EfficientDet Tutorial: inference, eval, and training

0. Install and view graph.

0.1 Install package and download source code/image.



In [0]:

    
%%capture
#@title
import os
import sys
import tensorflow.compat.v1 as tf

# Download source code.
if "efficientdet" not in os.getcwd():
  !git clone --depth 1 https://github.com/google/automl
  os.chdir('automl/efficientdet')
  sys.path.append('.')
  !pip install -r requirements.txt
  !pip install -U 'git+https://github.com/cocodataset/cocoapi.git#subdirectory=PythonAPI'
else:
  !git pull



In [0]:

    
MODEL = 'efficientdet-d0'  #@param

def download(m):
  if m not in os.listdir():
    !wget https://storage.googleapis.com/cloud-tpu-checkpoints/efficientdet/coco/{m}.tar.gz
    !tar zxf {m}.tar.gz
  ckpt_path = os.path.join(os.getcwd(), m)
  return ckpt_path

# Download checkpoint.
ckpt_path = download(MODEL)
print('Use model in {}'.format(ckpt_path))

# Prepare image and visualization settings.
image_url =  'https://user-images.githubusercontent.com/11736571/77320690-099af300-6d37-11ea-9d86-24f14dc2d540.png'#@param
image_name = 'img.png' #@param
!wget {image_url} -O img.png
import os
img_path = os.path.join(os.getcwd(), 'img.png')

min_score_thresh = 0.35  #@param
max_boxes_to_draw = 200  #@param
line_thickness =   2#@param

import PIL
# Get the largest of height/width and round to 128.
image_size = max(PIL.Image.open(img_path).size)









    



--2020-06-01 05:07:16--  https://storage.googleapis.com/cloud-tpu-checkpoints/efficientdet/coco/efficientdet-d0.tar.gz
Resolving storage.googleapis.com (storage.googleapis.com)... 173.194.217.128, 2607:f8b0:400c:c05::80
Connecting to storage.googleapis.com (storage.googleapis.com)|173.194.217.128|:443... connected.
HTTP request sent, awaiting response... 200 OK
Length: 28994253 (28M) [application/octet-stream]
Saving to: ‘efficientdet-d0.tar.gz’

efficientdet-d0.tar 100%[===================>]  27.65M   164MB/s    in 0.2s    

2020-06-01 05:07:16 (164 MB/s) - ‘efficientdet-d0.tar.gz’ saved [28994253/28994253]

Use model in /content/automl/efficientdet/efficientdet-d0
--2020-06-01 05:07:18--  https://user-images.githubusercontent.com/11736571/77320690-099af300-6d37-11ea-9d86-24f14dc2d540.png
Resolving user-images.githubusercontent.com (user-images.githubusercontent.com)... 151.101.0.133, 151.101.64.133, 151.101.128.133, ...
Connecting to user-images.githubusercontent.com (user-images.githubusercontent.com)|151.101.0.133|:443... connected.
HTTP request sent, awaiting response... 200 OK
Length: 4080549 (3.9M) [image/png]
Saving to: ‘img.png’

img.png             100%[===================>]   3.89M  12.8MB/s    in 0.3s    

2020-06-01 05:07:18 (12.8 MB/s) - ‘img.png’ saved [4080549/4080549]

0.2 View graph in TensorBoard



In [0]:

    
!python model_inspect.py --model_name={MODEL} --logdir=logs &> /dev/null
%load_ext tensorboard
%tensorboard --logdir logs

1. inference

1.1 Benchmark network latency

There are two types of latency: network latency and end-to-end latency.

network latency: from the first conv op to the network class and box prediction.
end-to-end latency: from image preprocessing, network, to the final postprocessing to generate a annotated new image.



In [0]:

    
# benchmaak network latency
!python model_inspect.py --runmode=bm --model_name=efficientdet-d4 --hparams="mixed_precision=true"

# With colab + Tesla T4 GPU, here are the batch size 1 latency summary:
# D0 (AP=33.5):  14.9ms,  FPS = 67.2   (batch size 8 FPS=)
# D1 (AP=39.6):  22.7ms,  FPS = 44.1   (batch size 8 FPS=)
# D2 (AP=43.0):  27.9ms,  FPS = 35.8   (batch size 8 FPS=)
# D3 (AP=45.8):  48.1ms,  FPS = 20.8   (batch size 8 FPS=)
# D4 (AP=49.4):  81.9ms,  FPS = 12.2   (batch size 8 FPS=)

1.2 Benchmark end-to-end latency



In [0]:

    
# Benchmark end-to-end latency (: preprocess + network + posprocess).
#
# With colab + Tesla T4 GPU, here are the batch size 1 latency summary:
# D0 (AP=33.5): 22.7ms,  FPS = 43.1   (batch size 4, FPS=)
# D1 (AP=39.6): 34.3ms,  FPS = 29.2   (batch size 4, FPS=)
# D2 (AP=43.0): 42.5ms,  FPS = 23.5   (batch size 4, FPS=)
# D3 (AP=45.8): 64.8ms,  FPS = 15.4   (batch size 4, FPS=)
# D4 (AP=49.4): 93.7ms,  FPS = 10.7   (batch size 4, FPS=)

m = 'efficientdet-d4'  # @param
batch_size = 1  # @param
m_path = download(m)

saved_model_dir = 'savedmodel'
!rm -rf {saved_model_dir}
!python model_inspect.py --runmode=saved_model --model_name={m} \
  --ckpt_path={m_path} --saved_model_dir={saved_model_dir} \
  --batch_size={batch_size}  --hparams="mixed_precision=true"
!python model_inspect.py --runmode=saved_model_benchmark --model_name={m} \
  --ckpt_path={m_path} --saved_model_dir={saved_model_dir}/{m}_frozen.pb \
  --batch_size={batch_size}  --hparams="mixed_precision=true" --input_image=testdata/img1.jpg









    



--2020-06-01 05:42:55--  https://storage.googleapis.com/cloud-tpu-checkpoints/efficientdet/coco/efficientdet-d4.tar.gz
Resolving storage.googleapis.com (storage.googleapis.com)... 173.194.216.128, 2607:f8b0:400c:c01::80
Connecting to storage.googleapis.com (storage.googleapis.com)|173.194.216.128|:443... connected.
HTTP request sent, awaiting response... 200 OK
Length: 152915674 (146M) [application/octet-stream]
Saving to: ‘efficientdet-d4.tar.gz’

efficientdet-d4.tar 100%[===================>] 145.83M   129MB/s    in 1.1s    

2020-06-01 05:42:57 (129 MB/s) - ‘efficientdet-d4.tar.gz’ saved [152915674/152915674]

2020-06-01 05:43:01.074313: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.1
2020-06-01 05:43:03.448788: I tensorflow/core/platform/profile_utils/cpu_utils.cc:102] CPU Frequency: 2300000000 Hz
2020-06-01 05:43:03.448968: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x1981100 initialized for platform Host (this does not guarantee that XLA will be used). Devices:
2020-06-01 05:43:03.449000: I tensorflow/compiler/xla/service/service.cc:176]   StreamExecutor device (0): Host, Default Version
2020-06-01 05:43:03.450763: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcuda.so.1
2020-06-01 05:43:03.592469: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:43:03.593133: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x19812c0 initialized for platform CUDA (this does not guarantee that XLA will be used). Devices:
2020-06-01 05:43:03.593171: I tensorflow/compiler/xla/service/service.cc:176]   StreamExecutor device (0): Tesla T4, Compute Capability 7.5
2020-06-01 05:43:03.593434: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:43:03.593956: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1561] Found device 0 with properties: 
pciBusID: 0000:00:04.0 name: Tesla T4 computeCapability: 7.5
coreClock: 1.59GHz coreCount: 40 deviceMemorySize: 14.73GiB deviceMemoryBandwidth: 298.08GiB/s
2020-06-01 05:43:03.594020: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.1
2020-06-01 05:43:03.595686: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcublas.so.10
2020-06-01 05:43:03.597259: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcufft.so.10
2020-06-01 05:43:03.597631: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcurand.so.10
2020-06-01 05:43:03.599125: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusolver.so.10
2020-06-01 05:43:03.599845: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusparse.so.10
2020-06-01 05:43:03.602709: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudnn.so.7
2020-06-01 05:43:03.602851: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:43:03.603445: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:43:03.603962: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1703] Adding visible gpu devices: 0
2020-06-01 05:43:03.604014: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.1
2020-06-01 05:43:04.140562: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1102] Device interconnect StreamExecutor with strength 1 edge matrix:
2020-06-01 05:43:04.140629: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1108]      0 
2020-06-01 05:43:04.140646: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1121] 0:   N 
2020-06-01 05:43:04.140884: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:43:04.141536: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:43:04.142083: W tensorflow/core/common_runtime/gpu/gpu_bfc_allocator.cc:39] Overriding allow_growth setting because the TF_FORCE_GPU_ALLOW_GROWTH environment variable is set. Original config value was 0.
2020-06-01 05:43:04.142134: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1247] Created TensorFlow device (/job:localhost/replica:0/task:0/device:GPU:0 with 13970 MB memory) -> physical GPU (device: 0, name: Tesla T4, pci bus id: 0000:00:04.0, compute capability: 7.5)
2020-06-01 05:43:04.227307: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:43:04.227940: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1561] Found device 0 with properties: 
pciBusID: 0000:00:04.0 name: Tesla T4 computeCapability: 7.5
coreClock: 1.59GHz coreCount: 40 deviceMemorySize: 14.73GiB deviceMemoryBandwidth: 298.08GiB/s
2020-06-01 05:43:04.228013: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.1
2020-06-01 05:43:04.228062: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcublas.so.10
2020-06-01 05:43:04.228087: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcufft.so.10
2020-06-01 05:43:04.228109: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcurand.so.10
2020-06-01 05:43:04.228130: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusolver.so.10
2020-06-01 05:43:04.228151: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusparse.so.10
2020-06-01 05:43:04.228172: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudnn.so.7
2020-06-01 05:43:04.228304: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:43:04.229034: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:43:04.229563: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1703] Adding visible gpu devices: 0
WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/ops/resource_variable_ops.py:1666: calling BaseResourceVariable.__init__ (from tensorflow.python.ops.resource_variable_ops) with constraint is deprecated and will be removed in a future version.
Instructions for updating:
If using Keras pass *_constraint arguments to layers.
W0601 05:43:06.053900 140469416372096 deprecation.py:506] From /usr/local/lib/python3.6/dist-packages/tensorflow/python/ops/resource_variable_ops.py:1666: calling BaseResourceVariable.__init__ (from tensorflow.python.ops.resource_variable_ops) with constraint is deprecated and will be removed in a future version.
Instructions for updating:
If using Keras pass *_constraint arguments to layers.
WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/profiler/internal/flops_registry.py:142: tensor_shape_from_node_def_name (from tensorflow.python.framework.graph_util_impl) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.compat.v1.graph_util.tensor_shape_from_node_def_name`
W0601 05:43:11.643087 140469416372096 deprecation.py:323] From /usr/local/lib/python3.6/dist-packages/tensorflow/python/profiler/internal/flops_registry.py:142: tensor_shape_from_node_def_name (from tensorflow.python.framework.graph_util_impl) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.compat.v1.graph_util.tensor_shape_from_node_def_name`
3 ops no flops stats due to incomplete shapes.
Parsing Inputs...
Incomplete shape.
Incomplete shape.
WARNING:tensorflow:From /content/automl/efficientdet/efficientdet_arch.py:138: conv2d (from tensorflow.python.layers.convolutional) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.keras.layers.Conv2D` instead.
W0601 05:43:12.194021 140469416372096 deprecation.py:323] From /content/automl/efficientdet/efficientdet_arch.py:138: conv2d (from tensorflow.python.layers.convolutional) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.keras.layers.Conv2D` instead.
WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/layers/convolutional.py:424: Layer.apply (from tensorflow.python.keras.engine.base_layer_v1) is deprecated and will be removed in a future version.
Instructions for updating:
Please use `layer.__call__` method instead.
W0601 05:43:12.197279 140469416372096 deprecation.py:323] From /usr/local/lib/python3.6/dist-packages/tensorflow/python/layers/convolutional.py:424: Layer.apply (from tensorflow.python.keras.engine.base_layer_v1) is deprecated and will be removed in a future version.
Instructions for updating:
Please use `layer.__call__` method instead.
WARNING:tensorflow:From /content/automl/efficientdet/efficientdet_arch.py:164: max_pooling2d (from tensorflow.python.layers.pooling) is deprecated and will be removed in a future version.
Instructions for updating:
Use keras.layers.MaxPooling2D instead.
W0601 05:43:12.273774 140469416372096 deprecation.py:323] From /content/automl/efficientdet/efficientdet_arch.py:164: max_pooling2d (from tensorflow.python.layers.pooling) is deprecated and will be removed in a future version.
Instructions for updating:
Use keras.layers.MaxPooling2D instead.
WARNING:tensorflow:From /content/automl/efficientdet/efficientdet_arch.py:685: separable_conv2d (from tensorflow.python.layers.convolutional) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.keras.layers.SeparableConv2D` instead.
W0601 05:43:12.303184 140469416372096 deprecation.py:323] From /content/automl/efficientdet/efficientdet_arch.py:685: separable_conv2d (from tensorflow.python.layers.convolutional) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.keras.layers.SeparableConv2D` instead.
3 ops no flops stats due to incomplete shapes.
Parsing Inputs...
Incomplete shape.
Incomplete shape.
3 ops no flops stats due to incomplete shapes.
Parsing Inputs...
Incomplete shape.
Incomplete shape.
WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/training/moving_averages.py:444: Variable.initialized_value (from tensorflow.python.ops.variables) is deprecated and will be removed in a future version.
Instructions for updating:
Use Variable.read_value. Variables in 2.X are initialized automatically both in eager and graph (inside tf.defun) contexts.
W0601 05:43:29.570928 140469416372096 deprecation.py:323] From /usr/local/lib/python3.6/dist-packages/tensorflow/python/training/moving_averages.py:444: Variable.initialized_value (from tensorflow.python.ops.variables) is deprecated and will be removed in a future version.
Instructions for updating:
Use Variable.read_value. Variables in 2.X are initialized automatically both in eager and graph (inside tf.defun) contexts.
WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/saved_model/signature_def_utils_impl.py:201: build_tensor_info (from tensorflow.python.saved_model.utils_impl) is deprecated and will be removed in a future version.
Instructions for updating:
This function will only be available through the v1 compatibility library as tf.compat.v1.saved_model.utils.build_tensor_info or tf.compat.v1.saved_model.build_tensor_info.
W0601 05:43:59.930180 140469416372096 deprecation.py:323] From /usr/local/lib/python3.6/dist-packages/tensorflow/python/saved_model/signature_def_utils_impl.py:201: build_tensor_info (from tensorflow.python.saved_model.utils_impl) is deprecated and will be removed in a future version.
Instructions for updating:
This function will only be available through the v1 compatibility library as tf.compat.v1.saved_model.utils.build_tensor_info or tf.compat.v1.saved_model.build_tensor_info.
WARNING:tensorflow:From /content/automl/efficientdet/inference.py:691: convert_variables_to_constants (from tensorflow.python.framework.graph_util_impl) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.compat.v1.graph_util.convert_variables_to_constants`
W0601 05:44:11.018419 140469416372096 deprecation.py:323] From /content/automl/efficientdet/inference.py:691: convert_variables_to_constants (from tensorflow.python.framework.graph_util_impl) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.compat.v1.graph_util.convert_variables_to_constants`
WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/graph_util_impl.py:359: extract_sub_graph (from tensorflow.python.framework.graph_util_impl) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.compat.v1.graph_util.extract_sub_graph`
W0601 05:44:11.018674 140469416372096 deprecation.py:323] From /usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/graph_util_impl.py:359: extract_sub_graph (from tensorflow.python.framework.graph_util_impl) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.compat.v1.graph_util.extract_sub_graph`
2020-06-01 05:44:16.731031: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.1
2020-06-01 05:44:19.085146: I tensorflow/core/platform/profile_utils/cpu_utils.cc:102] CPU Frequency: 2300000000 Hz
2020-06-01 05:44:19.085373: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x251f100 initialized for platform Host (this does not guarantee that XLA will be used). Devices:
2020-06-01 05:44:19.085408: I tensorflow/compiler/xla/service/service.cc:176]   StreamExecutor device (0): Host, Default Version
2020-06-01 05:44:19.087291: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcuda.so.1
2020-06-01 05:44:19.230882: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:44:19.231588: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x251f2c0 initialized for platform CUDA (this does not guarantee that XLA will be used). Devices:
2020-06-01 05:44:19.231624: I tensorflow/compiler/xla/service/service.cc:176]   StreamExecutor device (0): Tesla T4, Compute Capability 7.5
2020-06-01 05:44:19.231847: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:44:19.232455: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1561] Found device 0 with properties: 
pciBusID: 0000:00:04.0 name: Tesla T4 computeCapability: 7.5
coreClock: 1.59GHz coreCount: 40 deviceMemorySize: 14.73GiB deviceMemoryBandwidth: 298.08GiB/s
2020-06-01 05:44:19.232525: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.1
2020-06-01 05:44:19.234063: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcublas.so.10
2020-06-01 05:44:19.235648: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcufft.so.10
2020-06-01 05:44:19.235996: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcurand.so.10
2020-06-01 05:44:19.237520: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusolver.so.10
2020-06-01 05:44:19.238227: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusparse.so.10
2020-06-01 05:44:19.241182: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudnn.so.7
2020-06-01 05:44:19.241357: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:44:19.242002: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:44:19.242545: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1703] Adding visible gpu devices: 0
2020-06-01 05:44:19.242614: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.1
2020-06-01 05:44:19.784102: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1102] Device interconnect StreamExecutor with strength 1 edge matrix:
2020-06-01 05:44:19.784159: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1108]      0 
2020-06-01 05:44:19.784173: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1121] 0:   N 
2020-06-01 05:44:19.784431: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:44:19.785121: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:44:19.785723: W tensorflow/core/common_runtime/gpu/gpu_bfc_allocator.cc:39] Overriding allow_growth setting because the TF_FORCE_GPU_ALLOW_GROWTH environment variable is set. Original config value was 0.
2020-06-01 05:44:19.785773: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1247] Created TensorFlow device (/job:localhost/replica:0/task:0/device:GPU:0 with 13970 MB memory) -> physical GPU (device: 0, name: Tesla T4, pci bus id: 0000:00:04.0, compute capability: 7.5)
2020-06-01 05:44:22.260525: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudnn.so.7
2020-06-01 05:44:23.974135: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcublas.so.10
Per batch inference time:  0.09369662179997248
FPS:  10.672743379530186

1.3 Inference images.



In [0]:

    
# first export a saved model.
saved_model_dir = 'savedmodel'
!rm -rf {saved_model_dir}
!python model_inspect.py --runmode=saved_model --model_name={MODEL} \
  --ckpt_path={ckpt_path} --saved_model_dir={saved_model_dir}

# Then run saved_model_infer to do inference.
# Notably: batch_size, image_size must be the same as when it is exported.
serve_image_out = 'serve_image_out'
!mkdir {serve_image_out}

!python model_inspect.py --runmode=saved_model_infer \
  --saved_model_dir={saved_model_dir} \
  --model_name={MODEL}  --input_image=testdata/img1.jpg  \
  --output_image_dir={serve_image_out} \
  --min_score_thresh={min_score_thresh}  --max_boxes_to_draw={max_boxes_to_draw}









    



2020-05-31 22:18:29.834258: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.1
2020-05-31 22:18:31.984806: I tensorflow/core/platform/profile_utils/cpu_utils.cc:102] CPU Frequency: 2300000000 Hz
2020-05-31 22:18:31.984976: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x20ff100 initialized for platform Host (this does not guarantee that XLA will be used). Devices:
2020-05-31 22:18:31.985007: I tensorflow/compiler/xla/service/service.cc:176]   StreamExecutor device (0): Host, Default Version
2020-05-31 22:18:31.986636: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcuda.so.1
2020-05-31 22:18:32.117308: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-05-31 22:18:32.117967: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x20ff2c0 initialized for platform CUDA (this does not guarantee that XLA will be used). Devices:
2020-05-31 22:18:32.117998: I tensorflow/compiler/xla/service/service.cc:176]   StreamExecutor device (0): Tesla T4, Compute Capability 7.5
2020-05-31 22:18:32.118194: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-05-31 22:18:32.118719: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1561] Found device 0 with properties: 
pciBusID: 0000:00:04.0 name: Tesla T4 computeCapability: 7.5
coreClock: 1.59GHz coreCount: 40 deviceMemorySize: 14.73GiB deviceMemoryBandwidth: 298.08GiB/s
2020-05-31 22:18:32.118768: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.1
2020-05-31 22:18:32.120078: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcublas.so.10
2020-05-31 22:18:32.121500: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcufft.so.10
2020-05-31 22:18:32.121824: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcurand.so.10
2020-05-31 22:18:32.123138: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusolver.so.10
2020-05-31 22:18:32.123796: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusparse.so.10
2020-05-31 22:18:32.126344: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudnn.so.7
2020-05-31 22:18:32.126500: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-05-31 22:18:32.127069: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-05-31 22:18:32.127552: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1703] Adding visible gpu devices: 0
2020-05-31 22:18:32.127597: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.1
2020-05-31 22:18:32.624626: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1102] Device interconnect StreamExecutor with strength 1 edge matrix:
2020-05-31 22:18:32.624685: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1108]      0 
2020-05-31 22:18:32.624697: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1121] 0:   N 
2020-05-31 22:18:32.624908: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-05-31 22:18:32.625519: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-05-31 22:18:32.626014: W tensorflow/core/common_runtime/gpu/gpu_bfc_allocator.cc:39] Overriding allow_growth setting because the TF_FORCE_GPU_ALLOW_GROWTH environment variable is set. Original config value was 0.
2020-05-31 22:18:32.626056: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1247] Created TensorFlow device (/job:localhost/replica:0/task:0/device:GPU:0 with 13970 MB memory) -> physical GPU (device: 0, name: Tesla T4, pci bus id: 0000:00:04.0, compute capability: 7.5)
2020-05-31 22:18:32.674343: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-05-31 22:18:32.674888: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1561] Found device 0 with properties: 
pciBusID: 0000:00:04.0 name: Tesla T4 computeCapability: 7.5
coreClock: 1.59GHz coreCount: 40 deviceMemorySize: 14.73GiB deviceMemoryBandwidth: 298.08GiB/s
2020-05-31 22:18:32.674931: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.1
2020-05-31 22:18:32.674979: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcublas.so.10
2020-05-31 22:18:32.675000: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcufft.so.10
2020-05-31 22:18:32.675019: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcurand.so.10
2020-05-31 22:18:32.675037: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusolver.so.10
2020-05-31 22:18:32.675055: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusparse.so.10
2020-05-31 22:18:32.675073: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudnn.so.7
2020-05-31 22:18:32.675164: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-05-31 22:18:32.675714: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-05-31 22:18:32.676183: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1703] Adding visible gpu devices: 0
WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/ops/resource_variable_ops.py:1666: calling BaseResourceVariable.__init__ (from tensorflow.python.ops.resource_variable_ops) with constraint is deprecated and will be removed in a future version.
Instructions for updating:
If using Keras pass *_constraint arguments to layers.
W0531 22:18:34.165251 139695374006144 deprecation.py:506] From /usr/local/lib/python3.6/dist-packages/tensorflow/python/ops/resource_variable_ops.py:1666: calling BaseResourceVariable.__init__ (from tensorflow.python.ops.resource_variable_ops) with constraint is deprecated and will be removed in a future version.
Instructions for updating:
If using Keras pass *_constraint arguments to layers.
WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/profiler/internal/flops_registry.py:142: tensor_shape_from_node_def_name (from tensorflow.python.framework.graph_util_impl) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.compat.v1.graph_util.tensor_shape_from_node_def_name`
W0531 22:18:37.391232 139695374006144 deprecation.py:323] From /usr/local/lib/python3.6/dist-packages/tensorflow/python/profiler/internal/flops_registry.py:142: tensor_shape_from_node_def_name (from tensorflow.python.framework.graph_util_impl) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.compat.v1.graph_util.tensor_shape_from_node_def_name`
3 ops no flops stats due to incomplete shapes.
Parsing Inputs...
Incomplete shape.
Incomplete shape.
WARNING:tensorflow:From /content/automl/efficientdet/efficientdet_arch.py:138: conv2d (from tensorflow.python.layers.convolutional) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.keras.layers.Conv2D` instead.
W0531 22:18:37.628946 139695374006144 deprecation.py:323] From /content/automl/efficientdet/efficientdet_arch.py:138: conv2d (from tensorflow.python.layers.convolutional) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.keras.layers.Conv2D` instead.
WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/layers/convolutional.py:424: Layer.apply (from tensorflow.python.keras.engine.base_layer_v1) is deprecated and will be removed in a future version.
Instructions for updating:
Please use `layer.__call__` method instead.
W0531 22:18:37.632798 139695374006144 deprecation.py:323] From /usr/local/lib/python3.6/dist-packages/tensorflow/python/layers/convolutional.py:424: Layer.apply (from tensorflow.python.keras.engine.base_layer_v1) is deprecated and will be removed in a future version.
Instructions for updating:
Please use `layer.__call__` method instead.
WARNING:tensorflow:From /content/automl/efficientdet/efficientdet_arch.py:164: max_pooling2d (from tensorflow.python.layers.pooling) is deprecated and will be removed in a future version.
Instructions for updating:
Use keras.layers.MaxPooling2D instead.
W0531 22:18:37.685599 139695374006144 deprecation.py:323] From /content/automl/efficientdet/efficientdet_arch.py:164: max_pooling2d (from tensorflow.python.layers.pooling) is deprecated and will be removed in a future version.
Instructions for updating:
Use keras.layers.MaxPooling2D instead.
WARNING:tensorflow:From /content/automl/efficientdet/efficientdet_arch.py:685: separable_conv2d (from tensorflow.python.layers.convolutional) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.keras.layers.SeparableConv2D` instead.
W0531 22:18:37.708653 139695374006144 deprecation.py:323] From /content/automl/efficientdet/efficientdet_arch.py:685: separable_conv2d (from tensorflow.python.layers.convolutional) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.keras.layers.SeparableConv2D` instead.
3 ops no flops stats due to incomplete shapes.
Parsing Inputs...
Incomplete shape.
Incomplete shape.
3 ops no flops stats due to incomplete shapes.
Parsing Inputs...
Incomplete shape.
Incomplete shape.
WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/training/moving_averages.py:444: Variable.initialized_value (from tensorflow.python.ops.variables) is deprecated and will be removed in a future version.
Instructions for updating:
Use Variable.read_value. Variables in 2.X are initialized automatically both in eager and graph (inside tf.defun) contexts.
W0531 22:18:44.171763 139695374006144 deprecation.py:323] From /usr/local/lib/python3.6/dist-packages/tensorflow/python/training/moving_averages.py:444: Variable.initialized_value (from tensorflow.python.ops.variables) is deprecated and will be removed in a future version.
Instructions for updating:
Use Variable.read_value. Variables in 2.X are initialized automatically both in eager and graph (inside tf.defun) contexts.
WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/saved_model/signature_def_utils_impl.py:201: build_tensor_info (from tensorflow.python.saved_model.utils_impl) is deprecated and will be removed in a future version.
Instructions for updating:
This function will only be available through the v1 compatibility library as tf.compat.v1.saved_model.utils.build_tensor_info or tf.compat.v1.saved_model.build_tensor_info.
W0531 22:18:57.940969 139695374006144 deprecation.py:323] From /usr/local/lib/python3.6/dist-packages/tensorflow/python/saved_model/signature_def_utils_impl.py:201: build_tensor_info (from tensorflow.python.saved_model.utils_impl) is deprecated and will be removed in a future version.
Instructions for updating:
This function will only be available through the v1 compatibility library as tf.compat.v1.saved_model.utils.build_tensor_info or tf.compat.v1.saved_model.build_tensor_info.
WARNING:tensorflow:From /content/automl/efficientdet/inference.py:690: convert_variables_to_constants (from tensorflow.python.framework.graph_util_impl) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.compat.v1.graph_util.convert_variables_to_constants`
W0531 22:19:03.135875 139695374006144 deprecation.py:323] From /content/automl/efficientdet/inference.py:690: convert_variables_to_constants (from tensorflow.python.framework.graph_util_impl) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.compat.v1.graph_util.convert_variables_to_constants`
WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/graph_util_impl.py:359: extract_sub_graph (from tensorflow.python.framework.graph_util_impl) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.compat.v1.graph_util.extract_sub_graph`
W0531 22:19:03.136129 139695374006144 deprecation.py:323] From /usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/graph_util_impl.py:359: extract_sub_graph (from tensorflow.python.framework.graph_util_impl) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.compat.v1.graph_util.extract_sub_graph`
2020-05-31 22:19:06.848442: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.1
2020-05-31 22:19:09.071057: I tensorflow/core/platform/profile_utils/cpu_utils.cc:102] CPU Frequency: 2300000000 Hz
2020-05-31 22:19:09.071237: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x12bf100 initialized for platform Host (this does not guarantee that XLA will be used). Devices:
2020-05-31 22:19:09.071269: I tensorflow/compiler/xla/service/service.cc:176]   StreamExecutor device (0): Host, Default Version
2020-05-31 22:19:09.073086: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcuda.so.1
2020-05-31 22:19:09.205153: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-05-31 22:19:09.205811: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x12bf2c0 initialized for platform CUDA (this does not guarantee that XLA will be used). Devices:
2020-05-31 22:19:09.205841: I tensorflow/compiler/xla/service/service.cc:176]   StreamExecutor device (0): Tesla T4, Compute Capability 7.5
2020-05-31 22:19:09.206047: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-05-31 22:19:09.206568: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1561] Found device 0 with properties: 
pciBusID: 0000:00:04.0 name: Tesla T4 computeCapability: 7.5
coreClock: 1.59GHz coreCount: 40 deviceMemorySize: 14.73GiB deviceMemoryBandwidth: 298.08GiB/s
2020-05-31 22:19:09.206617: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.1
2020-05-31 22:19:09.207943: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcublas.so.10
2020-05-31 22:19:09.209330: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcufft.so.10
2020-05-31 22:19:09.209664: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcurand.so.10
2020-05-31 22:19:09.210969: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusolver.so.10
2020-05-31 22:19:09.211593: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusparse.so.10
2020-05-31 22:19:09.214126: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudnn.so.7
2020-05-31 22:19:09.214251: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-05-31 22:19:09.214834: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-05-31 22:19:09.215304: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1703] Adding visible gpu devices: 0
2020-05-31 22:19:09.215348: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.1
2020-05-31 22:19:09.724754: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1102] Device interconnect StreamExecutor with strength 1 edge matrix:
2020-05-31 22:19:09.724808: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1108]      0 
2020-05-31 22:19:09.724820: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1121] 0:   N 
2020-05-31 22:19:09.725035: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-05-31 22:19:09.725656: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-05-31 22:19:09.726145: W tensorflow/core/common_runtime/gpu/gpu_bfc_allocator.cc:39] Overriding allow_growth setting because the TF_FORCE_GPU_ALLOW_GROWTH environment variable is set. Original config value was 0.
2020-05-31 22:19:09.726185: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1247] Created TensorFlow device (/job:localhost/replica:0/task:0/device:GPU:0 with 13970 MB memory) -> physical GPU (device: 0, name: Tesla T4, pci bus id: 0000:00:04.0, compute capability: 7.5)
WARNING:tensorflow:From /content/automl/efficientdet/inference.py:678: load (from tensorflow.python.saved_model.loader_impl) is deprecated and will be removed in a future version.
Instructions for updating:
This function will only be available through the v1 compatibility library as tf.compat.v1.saved_model.loader.load or tf.compat.v1.saved_model.load. There will be a new function for importing SavedModels in Tensorflow 2.0.
W0531 22:19:09.726969 139773881436032 deprecation.py:323] From /content/automl/efficientdet/inference.py:678: load (from tensorflow.python.saved_model.loader_impl) is deprecated and will be removed in a future version.
Instructions for updating:
This function will only be available through the v1 compatibility library as tf.compat.v1.saved_model.loader.load or tf.compat.v1.saved_model.load. There will be a new function for importing SavedModels in Tensorflow 2.0.
all_files= ['testdata/img1.jpg']
2020-05-31 22:19:15.098209: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudnn.so.7
2020-05-31 22:19:16.755365: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcublas.so.10



In [0]:

    
from IPython import display
display.display(display.Image(os.path.join(serve_image_out, '0.jpg')))



In [0]:

    
# In case you need to specify different image size or batch size or #boxes, then
# you need to export a new saved model and run the inferernce.

serve_image_out = 'serve_image_out'
!mkdir {serve_image_out}
saved_model_dir = 'savedmodel'
!rm -rf {saved_model_dir}

# Step 1: export model
!python model_inspect.py --runmode=saved_model \
  --model_name=efficientdet-d0 --ckpt_path=efficientdet-d0 \
  --hparams="image_size=1920x1280" --saved_model_dir={saved_model_dir}

# Step 2: do inference with saved model.
!python model_inspect.py --runmode=saved_model_infer \
  --model_name=efficientdet-d0 --saved_model_dir={saved_model_dir} \
  --input_image=img.png --output_image_dir={serve_image_out} \
  --min_score_thresh={min_score_thresh}  --max_boxes_to_draw={max_boxes_to_draw}

from IPython import display
display.display(display.Image(os.path.join(serve_image_out, '0.jpg')))









    



2020-06-01 05:46:14.071332: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.1
2020-06-01 05:46:16.401257: I tensorflow/core/platform/profile_utils/cpu_utils.cc:102] CPU Frequency: 2300000000 Hz
2020-06-01 05:46:16.401496: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x270f100 initialized for platform Host (this does not guarantee that XLA will be used). Devices:
2020-06-01 05:46:16.401531: I tensorflow/compiler/xla/service/service.cc:176]   StreamExecutor device (0): Host, Default Version
2020-06-01 05:46:16.403436: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcuda.so.1
2020-06-01 05:46:16.547796: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:46:16.548486: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x270f2c0 initialized for platform CUDA (this does not guarantee that XLA will be used). Devices:
2020-06-01 05:46:16.548520: I tensorflow/compiler/xla/service/service.cc:176]   StreamExecutor device (0): Tesla T4, Compute Capability 7.5
2020-06-01 05:46:16.548728: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:46:16.549250: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1561] Found device 0 with properties: 
pciBusID: 0000:00:04.0 name: Tesla T4 computeCapability: 7.5
coreClock: 1.59GHz coreCount: 40 deviceMemorySize: 14.73GiB deviceMemoryBandwidth: 298.08GiB/s
2020-06-01 05:46:16.549316: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.1
2020-06-01 05:46:16.550811: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcublas.so.10
2020-06-01 05:46:16.552313: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcufft.so.10
2020-06-01 05:46:16.552651: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcurand.so.10
2020-06-01 05:46:16.554048: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusolver.so.10
2020-06-01 05:46:16.554769: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusparse.so.10
2020-06-01 05:46:16.557801: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudnn.so.7
2020-06-01 05:46:16.557946: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:46:16.558600: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:46:16.559170: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1703] Adding visible gpu devices: 0
2020-06-01 05:46:16.559220: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.1
2020-06-01 05:46:17.087089: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1102] Device interconnect StreamExecutor with strength 1 edge matrix:
2020-06-01 05:46:17.087152: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1108]      0 
2020-06-01 05:46:17.087166: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1121] 0:   N 
2020-06-01 05:46:17.087410: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:46:17.088037: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:46:17.088655: W tensorflow/core/common_runtime/gpu/gpu_bfc_allocator.cc:39] Overriding allow_growth setting because the TF_FORCE_GPU_ALLOW_GROWTH environment variable is set. Original config value was 0.
2020-06-01 05:46:17.088701: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1247] Created TensorFlow device (/job:localhost/replica:0/task:0/device:GPU:0 with 13970 MB memory) -> physical GPU (device: 0, name: Tesla T4, pci bus id: 0000:00:04.0, compute capability: 7.5)
2020-06-01 05:46:17.142073: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:46:17.142901: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1561] Found device 0 with properties: 
pciBusID: 0000:00:04.0 name: Tesla T4 computeCapability: 7.5
coreClock: 1.59GHz coreCount: 40 deviceMemorySize: 14.73GiB deviceMemoryBandwidth: 298.08GiB/s
2020-06-01 05:46:17.142958: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.1
2020-06-01 05:46:17.143004: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcublas.so.10
2020-06-01 05:46:17.143029: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcufft.so.10
2020-06-01 05:46:17.143049: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcurand.so.10
2020-06-01 05:46:17.143070: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusolver.so.10
2020-06-01 05:46:17.143090: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusparse.so.10
2020-06-01 05:46:17.143110: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudnn.so.7
2020-06-01 05:46:17.143227: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:46:17.144030: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:46:17.144772: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1703] Adding visible gpu devices: 0
WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/ops/resource_variable_ops.py:1666: calling BaseResourceVariable.__init__ (from tensorflow.python.ops.resource_variable_ops) with constraint is deprecated and will be removed in a future version.
Instructions for updating:
If using Keras pass *_constraint arguments to layers.
W0601 05:46:18.736636 139905342916480 deprecation.py:506] From /usr/local/lib/python3.6/dist-packages/tensorflow/python/ops/resource_variable_ops.py:1666: calling BaseResourceVariable.__init__ (from tensorflow.python.ops.resource_variable_ops) with constraint is deprecated and will be removed in a future version.
Instructions for updating:
If using Keras pass *_constraint arguments to layers.
WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/profiler/internal/flops_registry.py:142: tensor_shape_from_node_def_name (from tensorflow.python.framework.graph_util_impl) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.compat.v1.graph_util.tensor_shape_from_node_def_name`
W0601 05:46:22.133676 139905342916480 deprecation.py:323] From /usr/local/lib/python3.6/dist-packages/tensorflow/python/profiler/internal/flops_registry.py:142: tensor_shape_from_node_def_name (from tensorflow.python.framework.graph_util_impl) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.compat.v1.graph_util.tensor_shape_from_node_def_name`
3 ops no flops stats due to incomplete shapes.
Parsing Inputs...
Incomplete shape.
Incomplete shape.
WARNING:tensorflow:From /content/automl/efficientdet/efficientdet_arch.py:138: conv2d (from tensorflow.python.layers.convolutional) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.keras.layers.Conv2D` instead.
W0601 05:46:22.451671 139905342916480 deprecation.py:323] From /content/automl/efficientdet/efficientdet_arch.py:138: conv2d (from tensorflow.python.layers.convolutional) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.keras.layers.Conv2D` instead.
WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/layers/convolutional.py:424: Layer.apply (from tensorflow.python.keras.engine.base_layer_v1) is deprecated and will be removed in a future version.
Instructions for updating:
Please use `layer.__call__` method instead.
W0601 05:46:22.454935 139905342916480 deprecation.py:323] From /usr/local/lib/python3.6/dist-packages/tensorflow/python/layers/convolutional.py:424: Layer.apply (from tensorflow.python.keras.engine.base_layer_v1) is deprecated and will be removed in a future version.
Instructions for updating:
Please use `layer.__call__` method instead.
WARNING:tensorflow:From /content/automl/efficientdet/efficientdet_arch.py:164: max_pooling2d (from tensorflow.python.layers.pooling) is deprecated and will be removed in a future version.
Instructions for updating:
Use keras.layers.MaxPooling2D instead.
W0601 05:46:22.505055 139905342916480 deprecation.py:323] From /content/automl/efficientdet/efficientdet_arch.py:164: max_pooling2d (from tensorflow.python.layers.pooling) is deprecated and will be removed in a future version.
Instructions for updating:
Use keras.layers.MaxPooling2D instead.
WARNING:tensorflow:From /content/automl/efficientdet/efficientdet_arch.py:685: separable_conv2d (from tensorflow.python.layers.convolutional) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.keras.layers.SeparableConv2D` instead.
W0601 05:46:22.525332 139905342916480 deprecation.py:323] From /content/automl/efficientdet/efficientdet_arch.py:685: separable_conv2d (from tensorflow.python.layers.convolutional) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.keras.layers.SeparableConv2D` instead.
3 ops no flops stats due to incomplete shapes.
Parsing Inputs...
Incomplete shape.
Incomplete shape.
3 ops no flops stats due to incomplete shapes.
Parsing Inputs...
Incomplete shape.
Incomplete shape.
WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/training/moving_averages.py:444: Variable.initialized_value (from tensorflow.python.ops.variables) is deprecated and will be removed in a future version.
Instructions for updating:
Use Variable.read_value. Variables in 2.X are initialized automatically both in eager and graph (inside tf.defun) contexts.
W0601 05:46:29.380645 139905342916480 deprecation.py:323] From /usr/local/lib/python3.6/dist-packages/tensorflow/python/training/moving_averages.py:444: Variable.initialized_value (from tensorflow.python.ops.variables) is deprecated and will be removed in a future version.
Instructions for updating:
Use Variable.read_value. Variables in 2.X are initialized automatically both in eager and graph (inside tf.defun) contexts.
WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/saved_model/signature_def_utils_impl.py:201: build_tensor_info (from tensorflow.python.saved_model.utils_impl) is deprecated and will be removed in a future version.
Instructions for updating:
This function will only be available through the v1 compatibility library as tf.compat.v1.saved_model.utils.build_tensor_info or tf.compat.v1.saved_model.build_tensor_info.
W0601 05:46:44.401570 139905342916480 deprecation.py:323] From /usr/local/lib/python3.6/dist-packages/tensorflow/python/saved_model/signature_def_utils_impl.py:201: build_tensor_info (from tensorflow.python.saved_model.utils_impl) is deprecated and will be removed in a future version.
Instructions for updating:
This function will only be available through the v1 compatibility library as tf.compat.v1.saved_model.utils.build_tensor_info or tf.compat.v1.saved_model.build_tensor_info.
WARNING:tensorflow:From /content/automl/efficientdet/inference.py:691: convert_variables_to_constants (from tensorflow.python.framework.graph_util_impl) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.compat.v1.graph_util.convert_variables_to_constants`
W0601 05:46:50.026528 139905342916480 deprecation.py:323] From /content/automl/efficientdet/inference.py:691: convert_variables_to_constants (from tensorflow.python.framework.graph_util_impl) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.compat.v1.graph_util.convert_variables_to_constants`
WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/graph_util_impl.py:359: extract_sub_graph (from tensorflow.python.framework.graph_util_impl) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.compat.v1.graph_util.extract_sub_graph`
W0601 05:46:50.026841 139905342916480 deprecation.py:323] From /usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/graph_util_impl.py:359: extract_sub_graph (from tensorflow.python.framework.graph_util_impl) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.compat.v1.graph_util.extract_sub_graph`
2020-06-01 05:46:54.256374: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.1
2020-06-01 05:46:56.608253: I tensorflow/core/platform/profile_utils/cpu_utils.cc:102] CPU Frequency: 2300000000 Hz
2020-06-01 05:46:56.608499: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x2fef100 initialized for platform Host (this does not guarantee that XLA will be used). Devices:
2020-06-01 05:46:56.608535: I tensorflow/compiler/xla/service/service.cc:176]   StreamExecutor device (0): Host, Default Version
2020-06-01 05:46:56.610375: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcuda.so.1
2020-06-01 05:46:56.757873: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:46:56.758619: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x2fef2c0 initialized for platform CUDA (this does not guarantee that XLA will be used). Devices:
2020-06-01 05:46:56.758659: I tensorflow/compiler/xla/service/service.cc:176]   StreamExecutor device (0): Tesla T4, Compute Capability 7.5
2020-06-01 05:46:56.759121: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:46:56.759876: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1561] Found device 0 with properties: 
pciBusID: 0000:00:04.0 name: Tesla T4 computeCapability: 7.5
coreClock: 1.59GHz coreCount: 40 deviceMemorySize: 14.73GiB deviceMemoryBandwidth: 298.08GiB/s
2020-06-01 05:46:56.759940: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.1
2020-06-01 05:46:56.761691: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcublas.so.10
2020-06-01 05:46:56.763507: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcufft.so.10
2020-06-01 05:46:56.763818: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcurand.so.10
2020-06-01 05:46:56.765201: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusolver.so.10
2020-06-01 05:46:56.765873: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusparse.so.10
2020-06-01 05:46:56.768656: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudnn.so.7
2020-06-01 05:46:56.768788: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:46:56.769388: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:46:56.769953: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1703] Adding visible gpu devices: 0
2020-06-01 05:46:56.770002: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.1
2020-06-01 05:46:57.314572: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1102] Device interconnect StreamExecutor with strength 1 edge matrix:
2020-06-01 05:46:57.314626: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1108]      0 
2020-06-01 05:46:57.314640: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1121] 0:   N 
2020-06-01 05:46:57.314882: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:46:57.315517: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:46:57.316053: W tensorflow/core/common_runtime/gpu/gpu_bfc_allocator.cc:39] Overriding allow_growth setting because the TF_FORCE_GPU_ALLOW_GROWTH environment variable is set. Original config value was 0.
2020-06-01 05:46:57.316097: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1247] Created TensorFlow device (/job:localhost/replica:0/task:0/device:GPU:0 with 13970 MB memory) -> physical GPU (device: 0, name: Tesla T4, pci bus id: 0000:00:04.0, compute capability: 7.5)
WARNING:tensorflow:From /content/automl/efficientdet/inference.py:679: load (from tensorflow.python.saved_model.loader_impl) is deprecated and will be removed in a future version.
Instructions for updating:
This function will only be available through the v1 compatibility library as tf.compat.v1.saved_model.loader.load or tf.compat.v1.saved_model.load. There will be a new function for importing SavedModels in Tensorflow 2.0.
W0601 05:46:57.316836 140378130581376 deprecation.py:323] From /content/automl/efficientdet/inference.py:679: load (from tensorflow.python.saved_model.loader_impl) is deprecated and will be removed in a future version.
Instructions for updating:
This function will only be available through the v1 compatibility library as tf.compat.v1.saved_model.loader.load or tf.compat.v1.saved_model.load. There will be a new function for importing SavedModels in Tensorflow 2.0.
all_files= ['./img.png']
2020-06-01 05:47:03.094128: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudnn.so.7
2020-06-01 05:47:05.000097: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcublas.so.10

1.4 Inference video



In [0]:

    
# step 0: download video
video_url = 'https://storage.googleapis.com/cloud-tpu-checkpoints/efficientdet/data/video480p.mov'  # @param
!wget {video_url} -O input.mov

# Step 1: export model
saved_model_dir = 'savedmodel'
!rm -rf {saved_model_dir}

!python model_inspect.py --runmode=saved_model \
  --model_name=efficientdet-d0 --ckpt_path=efficientdet-d0 \
  --saved_model_dir={saved_model_dir} --hparams="mixed_precision=true"

# Step 2: do inference with saved model using saved_model_video
!python model_inspect.py --runmode=saved_model_video \
  --model_name=efficientdet-d0   --ckpt_path=efficientdet-d0 \
  --saved_model_dir={saved_model_dir} --hparams="mixed_precision=true" \
  --input_video=input.mov --output_video=output.mov
# Then you can view the output.mov









    



--2020-06-01 05:08:08--  https://storage.googleapis.com/cloud-tpu-checkpoints/efficientdet/data/video480p.mov
Resolving storage.googleapis.com (storage.googleapis.com)... 172.217.204.128, 2607:f8b0:400c:c05::80
Connecting to storage.googleapis.com (storage.googleapis.com)|172.217.204.128|:443... connected.
HTTP request sent, awaiting response... 200 OK
Length: 18511760 (18M) [application/octet-stream]
Saving to: ‘input.mov’

input.mov           100%[===================>]  17.65M  --.-KB/s    in 0.1s    

2020-06-01 05:08:08 (119 MB/s) - ‘input.mov’ saved [18511760/18511760]

2020-06-01 05:08:10.640913: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.1
2020-06-01 05:08:12.964736: I tensorflow/core/platform/profile_utils/cpu_utils.cc:102] CPU Frequency: 2300000000 Hz
2020-06-01 05:08:12.965102: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x2c41480 initialized for platform Host (this does not guarantee that XLA will be used). Devices:
2020-06-01 05:08:12.965147: I tensorflow/compiler/xla/service/service.cc:176]   StreamExecutor device (0): Host, Default Version
2020-06-01 05:08:12.969921: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcuda.so.1
2020-06-01 05:08:13.189523: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:08:13.190261: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x2c41640 initialized for platform CUDA (this does not guarantee that XLA will be used). Devices:
2020-06-01 05:08:13.190315: I tensorflow/compiler/xla/service/service.cc:176]   StreamExecutor device (0): Tesla T4, Compute Capability 7.5
2020-06-01 05:08:13.191600: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:08:13.192165: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1561] Found device 0 with properties: 
pciBusID: 0000:00:04.0 name: Tesla T4 computeCapability: 7.5
coreClock: 1.59GHz coreCount: 40 deviceMemorySize: 14.73GiB deviceMemoryBandwidth: 298.08GiB/s
2020-06-01 05:08:13.192213: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.1
2020-06-01 05:08:13.416369: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcublas.so.10
2020-06-01 05:08:13.542814: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcufft.so.10
2020-06-01 05:08:13.564000: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcurand.so.10
2020-06-01 05:08:13.832152: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusolver.so.10
2020-06-01 05:08:13.852017: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusparse.so.10
2020-06-01 05:08:14.369495: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudnn.so.7
2020-06-01 05:08:14.369747: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:08:14.370506: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:08:14.371069: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1703] Adding visible gpu devices: 0
2020-06-01 05:08:14.374734: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.1
2020-06-01 05:08:21.144501: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1102] Device interconnect StreamExecutor with strength 1 edge matrix:
2020-06-01 05:08:21.144562: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1108]      0 
2020-06-01 05:08:21.144576: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1121] 0:   N 
2020-06-01 05:08:21.150487: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:08:21.151227: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:08:21.151811: W tensorflow/core/common_runtime/gpu/gpu_bfc_allocator.cc:39] Overriding allow_growth setting because the TF_FORCE_GPU_ALLOW_GROWTH environment variable is set. Original config value was 0.
2020-06-01 05:08:21.151864: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1247] Created TensorFlow device (/job:localhost/replica:0/task:0/device:GPU:0 with 13970 MB memory) -> physical GPU (device: 0, name: Tesla T4, pci bus id: 0000:00:04.0, compute capability: 7.5)
2020-06-01 05:08:21.223963: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:08:21.224553: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1561] Found device 0 with properties: 
pciBusID: 0000:00:04.0 name: Tesla T4 computeCapability: 7.5
coreClock: 1.59GHz coreCount: 40 deviceMemorySize: 14.73GiB deviceMemoryBandwidth: 298.08GiB/s
2020-06-01 05:08:21.224604: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.1
2020-06-01 05:08:21.224650: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcublas.so.10
2020-06-01 05:08:21.224675: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcufft.so.10
2020-06-01 05:08:21.224698: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcurand.so.10
2020-06-01 05:08:21.224720: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusolver.so.10
2020-06-01 05:08:21.224742: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusparse.so.10
2020-06-01 05:08:21.224764: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudnn.so.7
2020-06-01 05:08:21.224874: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:08:21.225451: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:08:21.225958: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1703] Adding visible gpu devices: 0
WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/ops/resource_variable_ops.py:1666: calling BaseResourceVariable.__init__ (from tensorflow.python.ops.resource_variable_ops) with constraint is deprecated and will be removed in a future version.
Instructions for updating:
If using Keras pass *_constraint arguments to layers.
W0601 05:08:22.807971 140451268405120 deprecation.py:506] From /usr/local/lib/python3.6/dist-packages/tensorflow/python/ops/resource_variable_ops.py:1666: calling BaseResourceVariable.__init__ (from tensorflow.python.ops.resource_variable_ops) with constraint is deprecated and will be removed in a future version.
Instructions for updating:
If using Keras pass *_constraint arguments to layers.
WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/profiler/internal/flops_registry.py:142: tensor_shape_from_node_def_name (from tensorflow.python.framework.graph_util_impl) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.compat.v1.graph_util.tensor_shape_from_node_def_name`
W0601 05:08:26.245461 140451268405120 deprecation.py:323] From /usr/local/lib/python3.6/dist-packages/tensorflow/python/profiler/internal/flops_registry.py:142: tensor_shape_from_node_def_name (from tensorflow.python.framework.graph_util_impl) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.compat.v1.graph_util.tensor_shape_from_node_def_name`
3 ops no flops stats due to incomplete shapes.
Parsing Inputs...
Incomplete shape.
Incomplete shape.
WARNING:tensorflow:From /content/automl/efficientdet/efficientdet_arch.py:138: conv2d (from tensorflow.python.layers.convolutional) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.keras.layers.Conv2D` instead.
W0601 05:08:26.472386 140451268405120 deprecation.py:323] From /content/automl/efficientdet/efficientdet_arch.py:138: conv2d (from tensorflow.python.layers.convolutional) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.keras.layers.Conv2D` instead.
WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/layers/convolutional.py:424: Layer.apply (from tensorflow.python.keras.engine.base_layer_v1) is deprecated and will be removed in a future version.
Instructions for updating:
Please use `layer.__call__` method instead.
W0601 05:08:26.475308 140451268405120 deprecation.py:323] From /usr/local/lib/python3.6/dist-packages/tensorflow/python/layers/convolutional.py:424: Layer.apply (from tensorflow.python.keras.engine.base_layer_v1) is deprecated and will be removed in a future version.
Instructions for updating:
Please use `layer.__call__` method instead.
WARNING:tensorflow:From /content/automl/efficientdet/efficientdet_arch.py:164: max_pooling2d (from tensorflow.python.layers.pooling) is deprecated and will be removed in a future version.
Instructions for updating:
Use keras.layers.MaxPooling2D instead.
W0601 05:08:26.531509 140451268405120 deprecation.py:323] From /content/automl/efficientdet/efficientdet_arch.py:164: max_pooling2d (from tensorflow.python.layers.pooling) is deprecated and will be removed in a future version.
Instructions for updating:
Use keras.layers.MaxPooling2D instead.
WARNING:tensorflow:From /content/automl/efficientdet/efficientdet_arch.py:685: separable_conv2d (from tensorflow.python.layers.convolutional) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.keras.layers.SeparableConv2D` instead.
W0601 05:08:26.551836 140451268405120 deprecation.py:323] From /content/automl/efficientdet/efficientdet_arch.py:685: separable_conv2d (from tensorflow.python.layers.convolutional) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.keras.layers.SeparableConv2D` instead.
3 ops no flops stats due to incomplete shapes.
Parsing Inputs...
Incomplete shape.
Incomplete shape.
3 ops no flops stats due to incomplete shapes.
Parsing Inputs...
Incomplete shape.
Incomplete shape.
WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/training/moving_averages.py:444: Variable.initialized_value (from tensorflow.python.ops.variables) is deprecated and will be removed in a future version.
Instructions for updating:
Use Variable.read_value. Variables in 2.X are initialized automatically both in eager and graph (inside tf.defun) contexts.
W0601 05:08:33.364987 140451268405120 deprecation.py:323] From /usr/local/lib/python3.6/dist-packages/tensorflow/python/training/moving_averages.py:444: Variable.initialized_value (from tensorflow.python.ops.variables) is deprecated and will be removed in a future version.
Instructions for updating:
Use Variable.read_value. Variables in 2.X are initialized automatically both in eager and graph (inside tf.defun) contexts.
WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/saved_model/signature_def_utils_impl.py:201: build_tensor_info (from tensorflow.python.saved_model.utils_impl) is deprecated and will be removed in a future version.
Instructions for updating:
This function will only be available through the v1 compatibility library as tf.compat.v1.saved_model.utils.build_tensor_info or tf.compat.v1.saved_model.build_tensor_info.
W0601 05:08:47.828152 140451268405120 deprecation.py:323] From /usr/local/lib/python3.6/dist-packages/tensorflow/python/saved_model/signature_def_utils_impl.py:201: build_tensor_info (from tensorflow.python.saved_model.utils_impl) is deprecated and will be removed in a future version.
Instructions for updating:
This function will only be available through the v1 compatibility library as tf.compat.v1.saved_model.utils.build_tensor_info or tf.compat.v1.saved_model.build_tensor_info.
WARNING:tensorflow:From /content/automl/efficientdet/inference.py:691: convert_variables_to_constants (from tensorflow.python.framework.graph_util_impl) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.compat.v1.graph_util.convert_variables_to_constants`
W0601 05:08:53.265198 140451268405120 deprecation.py:323] From /content/automl/efficientdet/inference.py:691: convert_variables_to_constants (from tensorflow.python.framework.graph_util_impl) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.compat.v1.graph_util.convert_variables_to_constants`
WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/graph_util_impl.py:359: extract_sub_graph (from tensorflow.python.framework.graph_util_impl) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.compat.v1.graph_util.extract_sub_graph`
W0601 05:08:53.265506 140451268405120 deprecation.py:323] From /usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/graph_util_impl.py:359: extract_sub_graph (from tensorflow.python.framework.graph_util_impl) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.compat.v1.graph_util.extract_sub_graph`
2020-06-01 05:08:56.710234: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.1
2020-06-01 05:08:59.535479: I tensorflow/core/platform/profile_utils/cpu_utils.cc:102] CPU Frequency: 2300000000 Hz
2020-06-01 05:08:59.535664: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x30d3100 initialized for platform Host (this does not guarantee that XLA will be used). Devices:
2020-06-01 05:08:59.535696: I tensorflow/compiler/xla/service/service.cc:176]   StreamExecutor device (0): Host, Default Version
2020-06-01 05:08:59.537584: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcuda.so.1
2020-06-01 05:08:59.671441: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:08:59.672074: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x30d32c0 initialized for platform CUDA (this does not guarantee that XLA will be used). Devices:
2020-06-01 05:08:59.672104: I tensorflow/compiler/xla/service/service.cc:176]   StreamExecutor device (0): Tesla T4, Compute Capability 7.5
2020-06-01 05:08:59.672340: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:08:59.672877: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1561] Found device 0 with properties: 
pciBusID: 0000:00:04.0 name: Tesla T4 computeCapability: 7.5
coreClock: 1.59GHz coreCount: 40 deviceMemorySize: 14.73GiB deviceMemoryBandwidth: 298.08GiB/s
2020-06-01 05:08:59.672919: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.1
2020-06-01 05:08:59.674215: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcublas.so.10
2020-06-01 05:08:59.675592: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcufft.so.10
2020-06-01 05:08:59.675899: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcurand.so.10
2020-06-01 05:08:59.677196: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusolver.so.10
2020-06-01 05:08:59.677865: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusparse.so.10
2020-06-01 05:08:59.680574: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudnn.so.7
2020-06-01 05:08:59.680727: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:08:59.681340: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:08:59.681854: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1703] Adding visible gpu devices: 0
2020-06-01 05:08:59.681902: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.1
2020-06-01 05:09:00.205193: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1102] Device interconnect StreamExecutor with strength 1 edge matrix:
2020-06-01 05:09:00.205250: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1108]      0 
2020-06-01 05:09:00.205280: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1121] 0:   N 
2020-06-01 05:09:00.205510: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:09:00.206118: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:09:00.206641: W tensorflow/core/common_runtime/gpu/gpu_bfc_allocator.cc:39] Overriding allow_growth setting because the TF_FORCE_GPU_ALLOW_GROWTH environment variable is set. Original config value was 0.
2020-06-01 05:09:00.206699: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1247] Created TensorFlow device (/job:localhost/replica:0/task:0/device:GPU:0 with 13970 MB memory) -> physical GPU (device: 0, name: Tesla T4, pci bus id: 0000:00:04.0, compute capability: 7.5)
WARNING:tensorflow:From /content/automl/efficientdet/inference.py:679: load (from tensorflow.python.saved_model.loader_impl) is deprecated and will be removed in a future version.
Instructions for updating:
This function will only be available through the v1 compatibility library as tf.compat.v1.saved_model.loader.load or tf.compat.v1.saved_model.load. There will be a new function for importing SavedModels in Tensorflow 2.0.
W0601 05:09:00.207572 140294127970176 deprecation.py:323] From /content/automl/efficientdet/inference.py:679: load (from tensorflow.python.saved_model.loader_impl) is deprecated and will be removed in a future version.
Instructions for updating:
This function will only be available through the v1 compatibility library as tf.compat.v1.saved_model.loader.load or tf.compat.v1.saved_model.load. There will be a new function for importing SavedModels in Tensorflow 2.0.
2020-06-01 05:09:06.088973: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudnn.so.7
2020-06-01 05:09:11.133459: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcublas.so.10

3. COCO evaluation

3.1 COCO evaluation on validation set.



In [0]:

    
if 'val2017' not in os.listdir():
  !wget http://images.cocodataset.org/zips/val2017.zip
  !wget http://images.cocodataset.org/annotations/annotations_trainval2017.zip
  !unzip -q val2017.zip
  !unzip annotations_trainval2017.zip

  !mkdir tfrecord
  !PYTHONPATH=".:$PYTHONPATH"  python dataset/create_coco_tfrecord.py \
      --image_dir=val2017 \
      --caption_annotations_file=annotations/captions_val2017.json \
      --output_file_prefix=tfrecord/val \
      --num_shards=32









    



--2020-06-01 03:22:59--  http://images.cocodataset.org/zips/val2017.zip
Resolving images.cocodataset.org (images.cocodataset.org)... 52.216.239.115
Connecting to images.cocodataset.org (images.cocodataset.org)|52.216.239.115|:80... connected.
HTTP request sent, awaiting response... 200 OK
Length: 815585330 (778M) [application/zip]
Saving to: ‘val2017.zip’

val2017.zip         100%[===================>] 777.80M  95.2MB/s    in 8.1s    

2020-06-01 03:23:07 (96.1 MB/s) - ‘val2017.zip’ saved [815585330/815585330]

--2020-06-01 03:23:08--  http://images.cocodataset.org/annotations/annotations_trainval2017.zip
Resolving images.cocodataset.org (images.cocodataset.org)... 52.216.163.67
Connecting to images.cocodataset.org (images.cocodataset.org)|52.216.163.67|:80... connected.
HTTP request sent, awaiting response... 200 OK
Length: 252907541 (241M) [application/zip]
Saving to: ‘annotations_trainval2017.zip’

annotations_trainva 100%[===================>] 241.19M  96.2MB/s    in 2.5s    

2020-06-01 03:23:11 (96.2 MB/s) - ‘annotations_trainval2017.zip’ saved [252907541/252907541]

Archive:  annotations_trainval2017.zip
  inflating: annotations/instances_train2017.json  
  inflating: annotations/instances_val2017.json  
  inflating: annotations/captions_train2017.json  
  inflating: annotations/captions_val2017.json  
  inflating: annotations/person_keypoints_train2017.json  
  inflating: annotations/person_keypoints_val2017.json  
2020-06-01 03:23:39.581065: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.1
I0601 03:23:41.048378 140017000511360 create_coco_tfrecord.py:288] writing to output path: tfrecord/val
I0601 03:23:42.145988 140017000511360 create_coco_tfrecord.py:240] Building caption index.
I0601 03:23:42.152811 140017000511360 create_coco_tfrecord.py:252] 0 images are missing captions.
I0601 03:23:42.188177 140017000511360 create_coco_tfrecord.py:326] On image 0 of 5000
I0601 03:23:42.304900 140017000511360 create_coco_tfrecord.py:326] On image 100 of 5000
I0601 03:23:42.408726 140017000511360 create_coco_tfrecord.py:326] On image 200 of 5000
I0601 03:23:42.516799 140017000511360 create_coco_tfrecord.py:326] On image 300 of 5000
I0601 03:23:42.623024 140017000511360 create_coco_tfrecord.py:326] On image 400 of 5000
I0601 03:23:42.721782 140017000511360 create_coco_tfrecord.py:326] On image 500 of 5000
I0601 03:23:42.887866 140017000511360 create_coco_tfrecord.py:326] On image 600 of 5000
I0601 03:23:43.005725 140017000511360 create_coco_tfrecord.py:326] On image 700 of 5000
I0601 03:23:43.122723 140017000511360 create_coco_tfrecord.py:326] On image 800 of 5000
I0601 03:23:43.256930 140017000511360 create_coco_tfrecord.py:326] On image 900 of 5000
I0601 03:23:43.386459 140017000511360 create_coco_tfrecord.py:326] On image 1000 of 5000
I0601 03:23:43.513791 140017000511360 create_coco_tfrecord.py:326] On image 1100 of 5000
I0601 03:23:43.634377 140017000511360 create_coco_tfrecord.py:326] On image 1200 of 5000
I0601 03:23:43.748250 140017000511360 create_coco_tfrecord.py:326] On image 1300 of 5000
I0601 03:23:43.852913 140017000511360 create_coco_tfrecord.py:326] On image 1400 of 5000
I0601 03:23:43.977952 140017000511360 create_coco_tfrecord.py:326] On image 1500 of 5000
I0601 03:23:44.099335 140017000511360 create_coco_tfrecord.py:326] On image 1600 of 5000
I0601 03:23:44.223793 140017000511360 create_coco_tfrecord.py:326] On image 1700 of 5000
I0601 03:23:44.339614 140017000511360 create_coco_tfrecord.py:326] On image 1800 of 5000
I0601 03:23:44.453861 140017000511360 create_coco_tfrecord.py:326] On image 1900 of 5000
I0601 03:23:44.560936 140017000511360 create_coco_tfrecord.py:326] On image 2000 of 5000
I0601 03:23:44.825568 140017000511360 create_coco_tfrecord.py:326] On image 2100 of 5000
I0601 03:23:45.217720 140017000511360 create_coco_tfrecord.py:326] On image 2200 of 5000
I0601 03:23:45.526688 140017000511360 create_coco_tfrecord.py:326] On image 2300 of 5000
I0601 03:23:45.661227 140017000511360 create_coco_tfrecord.py:326] On image 2400 of 5000
I0601 03:23:45.872304 140017000511360 create_coco_tfrecord.py:326] On image 2500 of 5000
I0601 03:23:46.325947 140017000511360 create_coco_tfrecord.py:326] On image 2600 of 5000
I0601 03:23:46.455390 140017000511360 create_coco_tfrecord.py:326] On image 2700 of 5000
I0601 03:23:46.934686 140017000511360 create_coco_tfrecord.py:326] On image 2800 of 5000
I0601 03:23:47.429552 140017000511360 create_coco_tfrecord.py:326] On image 2900 of 5000
I0601 03:23:47.901286 140017000511360 create_coco_tfrecord.py:326] On image 3000 of 5000
I0601 03:23:48.331453 140017000511360 create_coco_tfrecord.py:326] On image 3100 of 5000
I0601 03:23:48.768493 140017000511360 create_coco_tfrecord.py:326] On image 3200 of 5000
I0601 03:23:49.232179 140017000511360 create_coco_tfrecord.py:326] On image 3300 of 5000
I0601 03:23:49.664278 140017000511360 create_coco_tfrecord.py:326] On image 3400 of 5000
I0601 03:23:50.062376 140017000511360 create_coco_tfrecord.py:326] On image 3500 of 5000
I0601 03:23:50.444368 140017000511360 create_coco_tfrecord.py:326] On image 3600 of 5000
I0601 03:23:50.883323 140017000511360 create_coco_tfrecord.py:326] On image 3700 of 5000
I0601 03:23:51.289299 140017000511360 create_coco_tfrecord.py:326] On image 3800 of 5000
I0601 03:23:51.697535 140017000511360 create_coco_tfrecord.py:326] On image 3900 of 5000
I0601 03:23:52.112248 140017000511360 create_coco_tfrecord.py:326] On image 4000 of 5000
I0601 03:23:52.515160 140017000511360 create_coco_tfrecord.py:326] On image 4100 of 5000
I0601 03:23:52.949388 140017000511360 create_coco_tfrecord.py:326] On image 4200 of 5000
I0601 03:23:53.368419 140017000511360 create_coco_tfrecord.py:326] On image 4300 of 5000
I0601 03:23:53.819533 140017000511360 create_coco_tfrecord.py:326] On image 4400 of 5000
I0601 03:23:54.243495 140017000511360 create_coco_tfrecord.py:326] On image 4500 of 5000
I0601 03:23:54.630336 140017000511360 create_coco_tfrecord.py:326] On image 4600 of 5000
I0601 03:23:55.044274 140017000511360 create_coco_tfrecord.py:326] On image 4700 of 5000
I0601 03:23:55.431686 140017000511360 create_coco_tfrecord.py:326] On image 4800 of 5000
I0601 03:23:55.863472 140017000511360 create_coco_tfrecord.py:326] On image 4900 of 5000
I0601 03:23:56.302472 140017000511360 create_coco_tfrecord.py:338] Finished writing, skipped 0 annotations.



In [0]:

    
# Evalute on validation set (takes about 10 mins for efficientdet-d0)
!python main.py --mode=eval  \
    --model_name={MODEL}  --model_dir={ckpt_path}  \
    --validation_file_pattern=tfrecord/val*  \
    --val_json_file=annotations/instances_val2017.json









    



2020-06-01 03:28:25.377482: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.1
WARNING:tensorflow:From main.py:234: The name tf.estimator.tpu.TPUConfig is deprecated. Please use tf.compat.v1.estimator.tpu.TPUConfig instead.

W0601 03:28:27.532853 140219029243776 module_wrapper.py:138] From main.py:234: The name tf.estimator.tpu.TPUConfig is deprecated. Please use tf.compat.v1.estimator.tpu.TPUConfig instead.

WARNING:tensorflow:From main.py:239: The name tf.estimator.tpu.InputPipelineConfig is deprecated. Please use tf.compat.v1.estimator.tpu.InputPipelineConfig instead.

W0601 03:28:27.533198 140219029243776 module_wrapper.py:138] From main.py:239: The name tf.estimator.tpu.InputPipelineConfig is deprecated. Please use tf.compat.v1.estimator.tpu.InputPipelineConfig instead.

WARNING:tensorflow:From main.py:247: The name tf.estimator.tpu.RunConfig is deprecated. Please use tf.compat.v1.estimator.tpu.RunConfig instead.

W0601 03:28:27.533394 140219029243776 module_wrapper.py:138] From main.py:247: The name tf.estimator.tpu.RunConfig is deprecated. Please use tf.compat.v1.estimator.tpu.RunConfig instead.

I0601 03:28:27.533618 140219029243776 main.py:262] {'name': 'efficientdet-d0', 'act_type': 'swish', 'image_size': (512, 512), 'target_size': None, 'input_rand_hflip': True, 'train_scale_min': 0.1, 'train_scale_max': 2.0, 'autoaugment_policy': None, 'use_augmix': False, 'augmix_params': (3, -1, 1), 'num_classes': 90, 'skip_crowd_during_training': True, 'label_id_mapping': None, 'max_instances_per_image': 100, 'min_level': 3, 'max_level': 7, 'num_scales': 3, 'aspect_ratios': [(1.0, 1.0), (1.4, 0.7), (0.7, 1.4)], 'anchor_scale': 4.0, 'is_training_bn': True, 'momentum': 0.9, 'optimizer': 'sgd', 'learning_rate': 0.08, 'lr_warmup_init': 0.008, 'lr_warmup_epoch': 1.0, 'first_lr_drop_epoch': 200.0, 'second_lr_drop_epoch': 250.0, 'poly_lr_power': 0.9, 'clip_gradients_norm': 10.0, 'num_epochs': 300, 'data_format': 'channels_last', 'alpha': 0.25, 'gamma': 1.5, 'delta': 0.1, 'box_loss_weight': 50.0, 'iou_loss_type': None, 'iou_loss_weight': 1.0, 'weight_decay': 4e-05, 'strategy': None, 'precision': None, 'box_class_repeats': 3, 'fpn_cell_repeats': 3, 'fpn_num_filters': 64, 'separable_conv': True, 'apply_bn_for_resampling': True, 'conv_after_downsample': False, 'conv_bn_act_pattern': False, 'use_native_resize_op': True, 'pooling_type': None, 'fpn_name': None, 'fpn_weight_method': None, 'fpn_config': None, 'survival_prob': None, 'img_summary_steps': None, 'lr_decay_method': 'cosine', 'moving_average_decay': 0.9998, 'ckpt_var_scope': None, 'var_exclude_expr': '.*/class-predict/.*', 'backbone_name': 'efficientnet-b0', 'backbone_config': None, 'var_freeze_expr': None, 'resnet_depth': 50, 'model_name': 'efficientdet-d0', 'iterations_per_loop': 100, 'model_dir': '/content/automl/efficientdet/efficientdet-d0', 'num_shards': 8, 'num_examples_per_epoch': 120000, 'backbone_ckpt': '', 'ckpt': None, 'val_json_file': 'annotations/instances_val2017.json', 'testdev_dir': None, 'mode': 'eval'}
WARNING:tensorflow:From main.py:317: The name tf.estimator.tpu.TPUEstimator is deprecated. Please use tf.compat.v1.estimator.tpu.TPUEstimator instead.

W0601 03:28:27.533797 140219029243776 module_wrapper.py:138] From main.py:317: The name tf.estimator.tpu.TPUEstimator is deprecated. Please use tf.compat.v1.estimator.tpu.TPUEstimator instead.

INFO:tensorflow:Using config: {'_model_dir': '/content/automl/efficientdet/efficientdet-d0', '_tf_random_seed': None, '_save_summary_steps': 100, '_save_checkpoints_steps': None, '_save_checkpoints_secs': 600, '_session_config': allow_soft_placement: true
, '_keep_checkpoint_max': 5, '_keep_checkpoint_every_n_hours': 10000, '_log_step_count_steps': None, '_train_distribute': None, '_device_fn': None, '_protocol': None, '_eval_distribute': None, '_experimental_distribute': None, '_experimental_max_worker_delay_secs': None, '_session_creation_timeout_secs': 7200, '_service': None, '_cluster_spec': ClusterSpec({}), '_task_type': 'worker', '_task_id': 0, '_global_id_in_cluster': 0, '_master': '', '_evaluation_master': '', '_is_chief': True, '_num_ps_replicas': 0, '_num_worker_replicas': 1, '_tpu_config': TPUConfig(iterations_per_loop=100, num_shards=8, num_cores_per_replica=None, per_host_input_for_training=3, tpu_job_name=None, initial_infeed_sleep_secs=None, input_partition_dims=None, eval_training_input_configuration=2, experimental_host_call_every_n_steps=1), '_cluster': None}
I0601 03:28:27.534311 140219029243776 estimator.py:191] Using config: {'_model_dir': '/content/automl/efficientdet/efficientdet-d0', '_tf_random_seed': None, '_save_summary_steps': 100, '_save_checkpoints_steps': None, '_save_checkpoints_secs': 600, '_session_config': allow_soft_placement: true
, '_keep_checkpoint_max': 5, '_keep_checkpoint_every_n_hours': 10000, '_log_step_count_steps': None, '_train_distribute': None, '_device_fn': None, '_protocol': None, '_eval_distribute': None, '_experimental_distribute': None, '_experimental_max_worker_delay_secs': None, '_session_creation_timeout_secs': 7200, '_service': None, '_cluster_spec': ClusterSpec({}), '_task_type': 'worker', '_task_id': 0, '_global_id_in_cluster': 0, '_master': '', '_evaluation_master': '', '_is_chief': True, '_num_ps_replicas': 0, '_num_worker_replicas': 1, '_tpu_config': TPUConfig(iterations_per_loop=100, num_shards=8, num_cores_per_replica=None, per_host_input_for_training=3, tpu_job_name=None, initial_infeed_sleep_secs=None, input_partition_dims=None, eval_training_input_configuration=2, experimental_host_call_every_n_steps=1), '_cluster': None}
INFO:tensorflow:_TPUContext: eval_on_tpu True
I0601 03:28:27.535187 140219029243776 tpu_context.py:216] _TPUContext: eval_on_tpu True
WARNING:tensorflow:eval_on_tpu ignored because use_tpu is False.
W0601 03:28:27.535668 140219029243776 tpu_context.py:218] eval_on_tpu ignored because use_tpu is False.
INFO:tensorflow:Waiting for new checkpoint at /content/automl/efficientdet/efficientdet-d0
I0601 03:28:27.535777 140219029243776 checkpoint_utils.py:125] Waiting for new checkpoint at /content/automl/efficientdet/efficientdet-d0
INFO:tensorflow:Found new checkpoint at /content/automl/efficientdet/efficientdet-d0/model
I0601 03:28:27.537644 140219029243776 checkpoint_utils.py:134] Found new checkpoint at /content/automl/efficientdet/efficientdet-d0/model
I0601 03:28:27.537798 140219029243776 main.py:337] Starting to evaluate.
WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/ops/resource_variable_ops.py:1666: calling BaseResourceVariable.__init__ (from tensorflow.python.ops.resource_variable_ops) with constraint is deprecated and will be removed in a future version.
Instructions for updating:
If using Keras pass *_constraint arguments to layers.
W0601 03:28:27.544394 140219029243776 deprecation.py:506] From /usr/local/lib/python3.6/dist-packages/tensorflow/python/ops/resource_variable_ops.py:1666: calling BaseResourceVariable.__init__ (from tensorflow.python.ops.resource_variable_ops) with constraint is deprecated and will be removed in a future version.
Instructions for updating:
If using Keras pass *_constraint arguments to layers.
WARNING:tensorflow:From /content/automl/efficientdet/dataloader.py:363: parallel_interleave (from tensorflow.python.data.experimental.ops.interleave_ops) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.data.Dataset.interleave(map_func, cycle_length, block_length, num_parallel_calls=tf.data.experimental.AUTOTUNE)` instead. If sloppy execution is desired, use `tf.data.Options.experimental_deterministic`.
W0601 03:28:27.563604 140219029243776 deprecation.py:323] From /content/automl/efficientdet/dataloader.py:363: parallel_interleave (from tensorflow.python.data.experimental.ops.interleave_ops) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.data.Dataset.interleave(map_func, cycle_length, block_length, num_parallel_calls=tf.data.experimental.AUTOTUNE)` instead. If sloppy execution is desired, use `tf.data.Options.experimental_deterministic`.
2020-06-01 03:28:27.667099: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcuda.so.1
2020-06-01 03:28:27.702213: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 03:28:27.702782: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1561] Found device 0 with properties: 
pciBusID: 0000:00:04.0 name: Tesla T4 computeCapability: 7.5
coreClock: 1.59GHz coreCount: 40 deviceMemorySize: 14.73GiB deviceMemoryBandwidth: 298.08GiB/s
2020-06-01 03:28:27.702825: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.1
2020-06-01 03:28:27.704274: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcublas.so.10
2020-06-01 03:28:27.705746: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcufft.so.10
2020-06-01 03:28:27.706108: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcurand.so.10
2020-06-01 03:28:27.707539: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusolver.so.10
2020-06-01 03:28:27.708273: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusparse.so.10
2020-06-01 03:28:27.711294: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudnn.so.7
2020-06-01 03:28:27.711441: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 03:28:27.712031: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 03:28:27.712554: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1703] Adding visible gpu devices: 0
INFO:tensorflow:Calling model_fn.
I0601 03:28:28.324099 140219029243776 estimator.py:1169] Calling model_fn.
INFO:tensorflow:Running eval on CPU/GPU
I0601 03:28:28.324348 140219029243776 tpu_estimator.py:3171] Running eval on CPU/GPU
I0601 03:28:28.324886 140219029243776 efficientdet_arch.py:720] act_type: swish
alpha: 0.25
anchor_scale: 4.0
apply_bn_for_resampling: true
aspect_ratios:
- !!python/tuple
    - 1.0
    - 1.0
- !!python/tuple
    - 1.4
    - 0.7
- !!python/tuple
    - 0.7
    - 1.4
augmix_params: !!python/tuple
- 3
- -1
- 1
autoaugment_policy: null
backbone_ckpt: ''
backbone_config: null
backbone_name: efficientnet-b0
batch_size: 1
box_class_repeats: 3
box_loss_weight: 50.0
ckpt: null
ckpt_var_scope: null
clip_gradients_norm: 10.0
conv_after_downsample: false
conv_bn_act_pattern: false
data_format: channels_last
delta: 0.1
first_lr_drop_epoch: 200.0
fpn_cell_repeats: 3
fpn_config: null
fpn_name: null
fpn_num_filters: 64
fpn_weight_method: null
gamma: 1.5
image_size: !!python/tuple
- 512
- 512
img_summary_steps: null
input_rand_hflip: false
iou_loss_type: null
iou_loss_weight: 1.0
is_training_bn: false
iterations_per_loop: 100
label_id_mapping: null
learning_rate: 0.08
lr_decay_method: cosine
lr_warmup_epoch: 1.0
lr_warmup_init: 0.008
max_instances_per_image: 100
max_level: 7
min_level: 3
mode: eval
model_dir: /content/automl/efficientdet/efficientdet-d0
model_name: efficientdet-d0
momentum: 0.9
moving_average_decay: 0.9998
name: efficientdet-d0
num_classes: 90
num_epochs: 300
num_examples_per_epoch: 120000
num_scales: 3
num_shards: 8
optimizer: sgd
poly_lr_power: 0.9
pooling_type: null
precision: null
resnet_depth: 50
second_lr_drop_epoch: 250.0
separable_conv: true
skip_crowd_during_training: true
strategy: null
survival_prob: null
target_size: null
testdev_dir: null
train_scale_max: 2.0
train_scale_min: 0.1
use_augmix: false
use_native_resize_op: true
use_tpu: false
val_json_file: annotations/instances_val2017.json
var_exclude_expr: .*/class-predict/.*
var_freeze_expr: null
weight_decay: 4.0e-05

I0601 03:28:28.329483 140219029243776 efficientnet_builder.py:224] global_params= GlobalParams(batch_norm_momentum=0.99, batch_norm_epsilon=0.001, dropout_rate=0.2, data_format='channels_last', num_classes=1000, width_coefficient=1.0, depth_coefficient=1.0, depth_divisor=8, min_depth=None, survival_prob=0.0, relu_fn=functools.partial(<function activation_fn at 0x7f86f29f3d90>, act_type='swish'), batch_norm=<class 'utils.BatchNormalization'>, use_se=True, local_pooling=None, condconv_num_experts=None, clip_projection_output=False, blocks_args=['r1_k3_s11_e1_i32_o16_se0.25', 'r2_k3_s22_e6_i16_o24_se0.25', 'r2_k5_s22_e6_i24_o40_se0.25', 'r3_k3_s22_e6_i40_o80_se0.25', 'r3_k5_s11_e6_i80_o112_se0.25', 'r4_k5_s22_e6_i112_o192_se0.25', 'r1_k3_s11_e6_i192_o320_se0.25'], fix_head_stem=None)
I0601 03:28:31.123013 140219029243776 api.py:587] Built stem layers with output shape: (1, 256, 256, 32)
I0601 03:28:33.629385 140219029243776 api.py:587] Block mb_conv_block  input shape: (1, 256, 256, 32)
I0601 03:28:33.661346 140219029243776 api.py:587] DWConv shape: (1, 256, 256, 32)
I0601 03:28:33.812110 140219029243776 api.py:587] Built Squeeze and Excitation with tensor shape: (1, 1, 1, 32)
I0601 03:28:33.841851 140219029243776 api.py:587] Project shape: (1, 256, 256, 16)
I0601 03:28:33.890565 140219029243776 api.py:587] Block mb_conv_block_1  input shape: (1, 256, 256, 16)
I0601 03:28:33.921314 140219029243776 api.py:587] Expand shape: (1, 256, 256, 96)
I0601 03:28:33.951348 140219029243776 api.py:587] DWConv shape: (1, 128, 128, 96)
I0601 03:28:33.984775 140219029243776 api.py:587] Built Squeeze and Excitation with tensor shape: (1, 1, 1, 96)
I0601 03:28:34.015750 140219029243776 api.py:587] Project shape: (1, 128, 128, 24)
I0601 03:28:34.024907 140219029243776 api.py:587] Block mb_conv_block_2  input shape: (1, 128, 128, 24)
I0601 03:28:34.058333 140219029243776 api.py:587] Expand shape: (1, 128, 128, 144)
I0601 03:28:34.089240 140219029243776 api.py:587] DWConv shape: (1, 128, 128, 144)
I0601 03:28:34.121758 140219029243776 api.py:587] Built Squeeze and Excitation with tensor shape: (1, 1, 1, 144)
I0601 03:28:34.151883 140219029243776 api.py:587] Project shape: (1, 128, 128, 24)
I0601 03:28:34.161335 140219029243776 api.py:587] Block mb_conv_block_3  input shape: (1, 128, 128, 24)
I0601 03:28:34.194899 140219029243776 api.py:587] Expand shape: (1, 128, 128, 144)
I0601 03:28:34.227180 140219029243776 api.py:587] DWConv shape: (1, 64, 64, 144)
I0601 03:28:34.267751 140219029243776 api.py:587] Built Squeeze and Excitation with tensor shape: (1, 1, 1, 144)
I0601 03:28:34.301784 140219029243776 api.py:587] Project shape: (1, 64, 64, 40)
I0601 03:28:34.313288 140219029243776 api.py:587] Block mb_conv_block_4  input shape: (1, 64, 64, 40)
I0601 03:28:34.347522 140219029243776 api.py:587] Expand shape: (1, 64, 64, 240)
I0601 03:28:34.378578 140219029243776 api.py:587] DWConv shape: (1, 64, 64, 240)
I0601 03:28:34.412726 140219029243776 api.py:587] Built Squeeze and Excitation with tensor shape: (1, 1, 1, 240)
I0601 03:28:34.442892 140219029243776 api.py:587] Project shape: (1, 64, 64, 40)
I0601 03:28:34.453632 140219029243776 api.py:587] Block mb_conv_block_5  input shape: (1, 64, 64, 40)
I0601 03:28:34.491663 140219029243776 api.py:587] Expand shape: (1, 64, 64, 240)
I0601 03:28:34.523128 140219029243776 api.py:587] DWConv shape: (1, 32, 32, 240)
I0601 03:28:34.555319 140219029243776 api.py:587] Built Squeeze and Excitation with tensor shape: (1, 1, 1, 240)
I0601 03:28:34.589114 140219029243776 api.py:587] Project shape: (1, 32, 32, 80)
I0601 03:28:34.601770 140219029243776 api.py:587] Block mb_conv_block_6  input shape: (1, 32, 32, 80)
I0601 03:28:34.633069 140219029243776 api.py:587] Expand shape: (1, 32, 32, 480)
I0601 03:28:34.669409 140219029243776 api.py:587] DWConv shape: (1, 32, 32, 480)
I0601 03:28:34.706420 140219029243776 api.py:587] Built Squeeze and Excitation with tensor shape: (1, 1, 1, 480)
I0601 03:28:34.737901 140219029243776 api.py:587] Project shape: (1, 32, 32, 80)
I0601 03:28:34.747094 140219029243776 api.py:587] Block mb_conv_block_7  input shape: (1, 32, 32, 80)
I0601 03:28:34.778591 140219029243776 api.py:587] Expand shape: (1, 32, 32, 480)
I0601 03:28:34.811817 140219029243776 api.py:587] DWConv shape: (1, 32, 32, 480)
I0601 03:28:34.843950 140219029243776 api.py:587] Built Squeeze and Excitation with tensor shape: (1, 1, 1, 480)
I0601 03:28:34.877232 140219029243776 api.py:587] Project shape: (1, 32, 32, 80)
I0601 03:28:34.886362 140219029243776 api.py:587] Block mb_conv_block_8  input shape: (1, 32, 32, 80)
I0601 03:28:34.917936 140219029243776 api.py:587] Expand shape: (1, 32, 32, 480)
I0601 03:28:34.949104 140219029243776 api.py:587] DWConv shape: (1, 32, 32, 480)
I0601 03:28:34.983436 140219029243776 api.py:587] Built Squeeze and Excitation with tensor shape: (1, 1, 1, 480)
I0601 03:28:35.017930 140219029243776 api.py:587] Project shape: (1, 32, 32, 112)
I0601 03:28:35.027091 140219029243776 api.py:587] Block mb_conv_block_9  input shape: (1, 32, 32, 112)
I0601 03:28:35.066807 140219029243776 api.py:587] Expand shape: (1, 32, 32, 672)
I0601 03:28:35.101598 140219029243776 api.py:587] DWConv shape: (1, 32, 32, 672)
I0601 03:28:35.134581 140219029243776 api.py:587] Built Squeeze and Excitation with tensor shape: (1, 1, 1, 672)
I0601 03:28:35.164892 140219029243776 api.py:587] Project shape: (1, 32, 32, 112)
I0601 03:28:35.174525 140219029243776 api.py:587] Block mb_conv_block_10  input shape: (1, 32, 32, 112)
I0601 03:28:35.207780 140219029243776 api.py:587] Expand shape: (1, 32, 32, 672)
I0601 03:28:35.238986 140219029243776 api.py:587] DWConv shape: (1, 32, 32, 672)
I0601 03:28:35.271428 140219029243776 api.py:587] Built Squeeze and Excitation with tensor shape: (1, 1, 1, 672)
I0601 03:28:35.307202 140219029243776 api.py:587] Project shape: (1, 32, 32, 112)
I0601 03:28:35.317617 140219029243776 api.py:587] Block mb_conv_block_11  input shape: (1, 32, 32, 112)
I0601 03:28:35.349665 140219029243776 api.py:587] Expand shape: (1, 32, 32, 672)
I0601 03:28:35.383484 140219029243776 api.py:587] DWConv shape: (1, 16, 16, 672)
I0601 03:28:35.418318 140219029243776 api.py:587] Built Squeeze and Excitation with tensor shape: (1, 1, 1, 672)
I0601 03:28:35.448390 140219029243776 api.py:587] Project shape: (1, 16, 16, 192)
I0601 03:28:35.458056 140219029243776 api.py:587] Block mb_conv_block_12  input shape: (1, 16, 16, 192)
I0601 03:28:35.499689 140219029243776 api.py:587] Expand shape: (1, 16, 16, 1152)
I0601 03:28:35.536298 140219029243776 api.py:587] DWConv shape: (1, 16, 16, 1152)
I0601 03:28:35.569783 140219029243776 api.py:587] Built Squeeze and Excitation with tensor shape: (1, 1, 1, 1152)
I0601 03:28:35.603983 140219029243776 api.py:587] Project shape: (1, 16, 16, 192)
I0601 03:28:35.613524 140219029243776 api.py:587] Block mb_conv_block_13  input shape: (1, 16, 16, 192)
I0601 03:28:35.648805 140219029243776 api.py:587] Expand shape: (1, 16, 16, 1152)
I0601 03:28:35.689206 140219029243776 api.py:587] DWConv shape: (1, 16, 16, 1152)
I0601 03:28:35.725426 140219029243776 api.py:587] Built Squeeze and Excitation with tensor shape: (1, 1, 1, 1152)
I0601 03:28:35.757848 140219029243776 api.py:587] Project shape: (1, 16, 16, 192)
I0601 03:28:35.767347 140219029243776 api.py:587] Block mb_conv_block_14  input shape: (1, 16, 16, 192)
I0601 03:28:35.807080 140219029243776 api.py:587] Expand shape: (1, 16, 16, 1152)
I0601 03:28:35.846334 140219029243776 api.py:587] DWConv shape: (1, 16, 16, 1152)
I0601 03:28:35.888501 140219029243776 api.py:587] Built Squeeze and Excitation with tensor shape: (1, 1, 1, 1152)
I0601 03:28:35.929219 140219029243776 api.py:587] Project shape: (1, 16, 16, 192)
I0601 03:28:35.940203 140219029243776 api.py:587] Block mb_conv_block_15  input shape: (1, 16, 16, 192)
I0601 03:28:35.979359 140219029243776 api.py:587] Expand shape: (1, 16, 16, 1152)
I0601 03:28:36.017061 140219029243776 api.py:587] DWConv shape: (1, 16, 16, 1152)
I0601 03:28:36.052354 140219029243776 api.py:587] Built Squeeze and Excitation with tensor shape: (1, 1, 1, 1152)
I0601 03:28:36.084471 140219029243776 api.py:587] Project shape: (1, 16, 16, 320)
WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/profiler/internal/flops_registry.py:142: tensor_shape_from_node_def_name (from tensorflow.python.framework.graph_util_impl) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.compat.v1.graph_util.tensor_shape_from_node_def_name`
W0601 03:28:36.094776 140219029243776 deprecation.py:323] From /usr/local/lib/python3.6/dist-packages/tensorflow/python/profiler/internal/flops_registry.py:142: tensor_shape_from_node_def_name (from tensorflow.python.framework.graph_util_impl) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.compat.v1.graph_util.tensor_shape_from_node_def_name`
Parsing Inputs...
I0601 03:28:36.338026 140219029243776 efficientdet_arch.py:725] backbone params/flops = 3.595388M, 1.933562589B
WARNING:tensorflow:From /content/automl/efficientdet/efficientdet_arch.py:138: conv2d (from tensorflow.python.layers.convolutional) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.keras.layers.Conv2D` instead.
W0601 03:28:36.338469 140219029243776 deprecation.py:323] From /content/automl/efficientdet/efficientdet_arch.py:138: conv2d (from tensorflow.python.layers.convolutional) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.keras.layers.Conv2D` instead.
WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/layers/convolutional.py:424: Layer.apply (from tensorflow.python.keras.engine.base_layer_v1) is deprecated and will be removed in a future version.
Instructions for updating:
Please use `layer.__call__` method instead.
W0601 03:28:36.341403 140219029243776 deprecation.py:323] From /usr/local/lib/python3.6/dist-packages/tensorflow/python/layers/convolutional.py:424: Layer.apply (from tensorflow.python.keras.engine.base_layer_v1) is deprecated and will be removed in a future version.
Instructions for updating:
Please use `layer.__call__` method instead.
WARNING:tensorflow:From /content/automl/efficientdet/efficientdet_arch.py:164: max_pooling2d (from tensorflow.python.layers.pooling) is deprecated and will be removed in a future version.
Instructions for updating:
Use keras.layers.MaxPooling2D instead.
W0601 03:28:36.389999 140219029243776 deprecation.py:323] From /content/automl/efficientdet/efficientdet_arch.py:164: max_pooling2d (from tensorflow.python.layers.pooling) is deprecated and will be removed in a future version.
Instructions for updating:
Use keras.layers.MaxPooling2D instead.
I0601 03:28:36.397655 140219029243776 efficientdet_arch.py:471] building cell 0
I0601 03:28:36.397964 140219029243776 efficientdet_arch.py:648] fnode 0 : {'feat_level': 6, 'inputs_offsets': [3, 4]}
WARNING:tensorflow:From /content/automl/efficientdet/efficientdet_arch.py:685: separable_conv2d (from tensorflow.python.layers.convolutional) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.keras.layers.SeparableConv2D` instead.
W0601 03:28:36.410160 140219029243776 deprecation.py:323] From /content/automl/efficientdet/efficientdet_arch.py:685: separable_conv2d (from tensorflow.python.layers.convolutional) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.keras.layers.SeparableConv2D` instead.
I0601 03:28:36.463976 140219029243776 efficientdet_arch.py:648] fnode 1 : {'feat_level': 5, 'inputs_offsets': [2, 5]}
I0601 03:28:36.579131 140219029243776 efficientdet_arch.py:648] fnode 2 : {'feat_level': 4, 'inputs_offsets': [1, 6]}
I0601 03:28:36.694639 140219029243776 efficientdet_arch.py:648] fnode 3 : {'feat_level': 3, 'inputs_offsets': [0, 7]}
I0601 03:28:36.810347 140219029243776 efficientdet_arch.py:648] fnode 4 : {'feat_level': 4, 'inputs_offsets': [1, 7, 8]}
I0601 03:28:36.929757 140219029243776 efficientdet_arch.py:648] fnode 5 : {'feat_level': 5, 'inputs_offsets': [2, 6, 9]}
I0601 03:28:37.053891 140219029243776 efficientdet_arch.py:648] fnode 6 : {'feat_level': 6, 'inputs_offsets': [3, 5, 10]}
I0601 03:28:37.132705 140219029243776 efficientdet_arch.py:648] fnode 7 : {'feat_level': 7, 'inputs_offsets': [4, 11]}
I0601 03:28:37.212006 140219029243776 efficientdet_arch.py:471] building cell 1
I0601 03:28:37.212349 140219029243776 efficientdet_arch.py:648] fnode 0 : {'feat_level': 6, 'inputs_offsets': [3, 4]}
I0601 03:28:37.284093 140219029243776 efficientdet_arch.py:648] fnode 1 : {'feat_level': 5, 'inputs_offsets': [2, 5]}
I0601 03:28:37.357714 140219029243776 efficientdet_arch.py:648] fnode 2 : {'feat_level': 4, 'inputs_offsets': [1, 6]}
I0601 03:28:37.438060 140219029243776 efficientdet_arch.py:648] fnode 3 : {'feat_level': 3, 'inputs_offsets': [0, 7]}
I0601 03:28:37.510998 140219029243776 efficientdet_arch.py:648] fnode 4 : {'feat_level': 4, 'inputs_offsets': [1, 7, 8]}
I0601 03:28:37.601469 140219029243776 efficientdet_arch.py:648] fnode 5 : {'feat_level': 5, 'inputs_offsets': [2, 6, 9]}
I0601 03:28:37.687102 140219029243776 efficientdet_arch.py:648] fnode 6 : {'feat_level': 6, 'inputs_offsets': [3, 5, 10]}
I0601 03:28:37.770025 140219029243776 efficientdet_arch.py:648] fnode 7 : {'feat_level': 7, 'inputs_offsets': [4, 11]}
I0601 03:28:37.852831 140219029243776 efficientdet_arch.py:471] building cell 2
I0601 03:28:37.853177 140219029243776 efficientdet_arch.py:648] fnode 0 : {'feat_level': 6, 'inputs_offsets': [3, 4]}
I0601 03:28:37.931208 140219029243776 efficientdet_arch.py:648] fnode 1 : {'feat_level': 5, 'inputs_offsets': [2, 5]}
I0601 03:28:38.012127 140219029243776 efficientdet_arch.py:648] fnode 2 : {'feat_level': 4, 'inputs_offsets': [1, 6]}
I0601 03:28:38.090590 140219029243776 efficientdet_arch.py:648] fnode 3 : {'feat_level': 3, 'inputs_offsets': [0, 7]}
I0601 03:28:38.170805 140219029243776 efficientdet_arch.py:648] fnode 4 : {'feat_level': 4, 'inputs_offsets': [1, 7, 8]}
I0601 03:28:38.261656 140219029243776 efficientdet_arch.py:648] fnode 5 : {'feat_level': 5, 'inputs_offsets': [2, 6, 9]}
I0601 03:28:38.449078 140219029243776 efficientdet_arch.py:648] fnode 6 : {'feat_level': 6, 'inputs_offsets': [3, 5, 10]}
I0601 03:28:38.541515 140219029243776 efficientdet_arch.py:648] fnode 7 : {'feat_level': 7, 'inputs_offsets': [4, 11]}
Parsing Inputs...
I0601 03:28:39.043738 140219029243776 efficientdet_arch.py:730] backbone+fpn params/flops = 3.791669M, 2.076332296B
Parsing Inputs...
I0601 03:28:42.122260 140219029243776 efficientdet_arch.py:735] backbone+fpn+box params/flops = 3.880067M, 2.535456184B
I0601 03:28:42.122621 140219029243776 det_model_fn.py:97] LR schedule method: cosine
WARNING:tensorflow:From /content/automl/efficientdet/det_model_fn.py:659: The name tf.estimator.tpu.TPUEstimatorSpec is deprecated. Please use tf.compat.v1.estimator.tpu.TPUEstimatorSpec instead.

W0601 03:28:42.439553 140219029243776 module_wrapper.py:138] From /content/automl/efficientdet/det_model_fn.py:659: The name tf.estimator.tpu.TPUEstimatorSpec is deprecated. Please use tf.compat.v1.estimator.tpu.TPUEstimatorSpec instead.

I0601 03:28:42.469991 140219029243776 det_model_fn.py:589] Eval val with groudtruths annotations/instances_val2017.json.
WARNING:tensorflow:From /content/automl/efficientdet/anchors.py:587: py_func (from tensorflow.python.ops.script_ops) is deprecated and will be removed in a future version.
Instructions for updating:
tf.py_func is deprecated in TF V2. Instead, there are two
    options available in V2.
    - tf.py_function takes a python function which manipulates tf eager
    tensors instead of numpy arrays. It's easy to convert a tf eager tensor to
    an ndarray (just call tensor.numpy()) but having access to eager tensors
    means `tf.py_function`s can use accelerators such as GPUs as well as
    being differentiable using a gradient tape.
    - tf.numpy_function maintains the semantics of the deprecated tf.py_func
    (it is not differentiable, and manipulates numpy arrays). It drops the
    stateful argument making all functions stateful.
    
W0601 03:28:42.478416 140219029243776 deprecation.py:323] From /content/automl/efficientdet/anchors.py:587: py_func (from tensorflow.python.ops.script_ops) is deprecated and will be removed in a future version.
Instructions for updating:
tf.py_func is deprecated in TF V2. Instead, there are two
    options available in V2.
    - tf.py_function takes a python function which manipulates tf eager
    tensors instead of numpy arrays. It's easy to convert a tf eager tensor to
    an ndarray (just call tensor.numpy()) but having access to eager tensors
    means `tf.py_function`s can use accelerators such as GPUs as well as
    being differentiable using a gradient tape.
    - tf.numpy_function maintains the semantics of the deprecated tf.py_func
    (it is not differentiable, and manipulates numpy arrays). It drops the
    stateful argument making all functions stateful.
    
loading annotations into memory...
Done (t=0.64s)
creating index...
index created!
I0601 03:28:43.161083 140219029243776 det_model_fn.py:652] Load EMA vars with ema_decay=0.999800
INFO:tensorflow:Done calling model_fn.
I0601 03:28:43.667938 140219029243776 estimator.py:1171] Done calling model_fn.
INFO:tensorflow:Starting evaluation at 2020-06-01T03:28:43Z
I0601 03:28:43.685333 140219029243776 evaluation.py:255] Starting evaluation at 2020-06-01T03:28:43Z
INFO:tensorflow:Graph was finalized.
I0601 03:28:44.014962 140219029243776 monitored_session.py:246] Graph was finalized.
2020-06-01 03:28:44.020247: I tensorflow/core/platform/profile_utils/cpu_utils.cc:102] CPU Frequency: 2300000000 Hz
2020-06-01 03:28:44.020470: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x1849100 initialized for platform Host (this does not guarantee that XLA will be used). Devices:
2020-06-01 03:28:44.020501: I tensorflow/compiler/xla/service/service.cc:176]   StreamExecutor device (0): Host, Default Version
2020-06-01 03:28:44.116278: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 03:28:44.116927: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x18492c0 initialized for platform CUDA (this does not guarantee that XLA will be used). Devices:
2020-06-01 03:28:44.116960: I tensorflow/compiler/xla/service/service.cc:176]   StreamExecutor device (0): Tesla T4, Compute Capability 7.5
2020-06-01 03:28:44.117208: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 03:28:44.117714: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1561] Found device 0 with properties: 
pciBusID: 0000:00:04.0 name: Tesla T4 computeCapability: 7.5
coreClock: 1.59GHz coreCount: 40 deviceMemorySize: 14.73GiB deviceMemoryBandwidth: 298.08GiB/s
2020-06-01 03:28:44.117767: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.1
2020-06-01 03:28:44.117812: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcublas.so.10
2020-06-01 03:28:44.117833: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcufft.so.10
2020-06-01 03:28:44.117852: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcurand.so.10
2020-06-01 03:28:44.117870: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusolver.so.10
2020-06-01 03:28:44.117893: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusparse.so.10
2020-06-01 03:28:44.117911: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudnn.so.7
2020-06-01 03:28:44.118006: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 03:28:44.118585: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 03:28:44.119077: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1703] Adding visible gpu devices: 0
2020-06-01 03:28:44.119129: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.1
2020-06-01 03:28:44.614816: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1102] Device interconnect StreamExecutor with strength 1 edge matrix:
2020-06-01 03:28:44.614873: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1108]      0 
2020-06-01 03:28:44.614886: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1121] 0:   N 
2020-06-01 03:28:44.615122: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 03:28:44.615721: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 03:28:44.616235: W tensorflow/core/common_runtime/gpu/gpu_bfc_allocator.cc:39] Overriding allow_growth setting because the TF_FORCE_GPU_ALLOW_GROWTH environment variable is set. Original config value was 0.
2020-06-01 03:28:44.616290: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1247] Created TensorFlow device (/job:localhost/replica:0/task:0/device:GPU:0 with 13970 MB memory) -> physical GPU (device: 0, name: Tesla T4, pci bus id: 0000:00:04.0, compute capability: 7.5)
INFO:tensorflow:Restoring parameters from /content/automl/efficientdet/efficientdet-d0/model
I0601 03:28:44.617812 140219029243776 saver.py:1293] Restoring parameters from /content/automl/efficientdet/efficientdet-d0/model
INFO:tensorflow:Running local_init_op.
I0601 03:28:45.706003 140219029243776 session_manager.py:505] Running local_init_op.
INFO:tensorflow:Done running local_init_op.
I0601 03:28:45.794684 140219029243776 session_manager.py:508] Done running local_init_op.
2020-06-01 03:28:48.918356: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudnn.so.7
2020-06-01 03:28:50.688065: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcublas.so.10
INFO:tensorflow:Evaluation [500/5000]
I0601 03:29:40.522579 140219029243776 evaluation.py:167] Evaluation [500/5000]
INFO:tensorflow:Evaluation [1000/5000]
I0601 03:30:29.711343 140219029243776 evaluation.py:167] Evaluation [1000/5000]
INFO:tensorflow:Evaluation [1500/5000]
I0601 03:31:20.120194 140219029243776 evaluation.py:167] Evaluation [1500/5000]
INFO:tensorflow:Evaluation [2000/5000]
I0601 03:32:09.853394 140219029243776 evaluation.py:167] Evaluation [2000/5000]
INFO:tensorflow:Evaluation [2500/5000]
I0601 03:32:58.698869 140219029243776 evaluation.py:167] Evaluation [2500/5000]
INFO:tensorflow:Evaluation [3000/5000]
I0601 03:33:49.175371 140219029243776 evaluation.py:167] Evaluation [3000/5000]
INFO:tensorflow:Evaluation [3500/5000]
I0601 03:34:39.156866 140219029243776 evaluation.py:167] Evaluation [3500/5000]
INFO:tensorflow:Evaluation [4000/5000]
I0601 03:35:28.423717 140219029243776 evaluation.py:167] Evaluation [4000/5000]
INFO:tensorflow:Evaluation [4500/5000]
I0601 03:36:17.885743 140219029243776 evaluation.py:167] Evaluation [4500/5000]
INFO:tensorflow:Evaluation [5000/5000]
I0601 03:37:07.498711 140219029243776 evaluation.py:167] Evaluation [5000/5000]
Loading and preparing results...
Converting ndarray to lists...
(500000, 7)
0/500000
DONE (t=3.43s)
creating index...
index created!
Running per image evaluation...
Evaluate annotation type *bbox*
DONE (t=68.06s).
Accumulating evaluation results...
DONE (t=9.95s).
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.335
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.515
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.353
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.125
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.388
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.526
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.287
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.439
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.465
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.193
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.549
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.685
INFO:tensorflow:Inference Time : 587.28611s
I0601 03:38:30.971660 140219029243776 evaluation.py:273] Inference Time : 587.28611s
INFO:tensorflow:Finished evaluation at 2020-06-01-03:38:30
I0601 03:38:30.971896 140219029243776 evaluation.py:276] Finished evaluation at 2020-06-01-03:38:30
INFO:tensorflow:Saving dict for global step 0: AP = 0.33491418, AP50 = 0.5154238, AP75 = 0.35295796, APl = 0.52646726, APm = 0.38791642, APs = 0.12475659, ARl = 0.6851952, ARm = 0.5492734, ARmax1 = 0.2873492, ARmax10 = 0.43856248, ARmax100 = 0.46534252, ARs = 0.19252816, box_loss = 0.0, cls_loss = 30.711409, global_step = 0, loss = 30.806612
I0601 03:38:30.972132 140219029243776 estimator.py:2066] Saving dict for global step 0: AP = 0.33491418, AP50 = 0.5154238, AP75 = 0.35295796, APl = 0.52646726, APm = 0.38791642, APs = 0.12475659, ARl = 0.6851952, ARm = 0.5492734, ARmax1 = 0.2873492, ARmax10 = 0.43856248, ARmax100 = 0.46534252, ARs = 0.19252816, box_loss = 0.0, cls_loss = 30.711409, global_step = 0, loss = 30.806612
INFO:tensorflow:Saving 'checkpoint_path' summary for global step 0: /content/automl/efficientdet/efficientdet-d0/model
I0601 03:38:31.859352 140219029243776 estimator.py:2127] Saving 'checkpoint_path' summary for global step 0: /content/automl/efficientdet/efficientdet-d0/model
INFO:tensorflow:evaluation_loop marked as finished
I0601 03:38:31.860275 140219029243776 error_handling.py:115] evaluation_loop marked as finished
I0601 03:38:31.860512 140219029243776 main.py:346] Eval results: {'AP': 0.33491418, 'AP50': 0.5154238, 'AP75': 0.35295796, 'APl': 0.52646726, 'APm': 0.38791642, 'APs': 0.12475659, 'ARl': 0.6851952, 'ARm': 0.5492734, 'ARmax1': 0.2873492, 'ARmax10': 0.43856248, 'ARmax100': 0.46534252, 'ARs': 0.19252816, 'box_loss': 0.0, 'cls_loss': 30.711409, 'loss': 30.806612, 'global_step': 0}
I0601 03:38:31.860629 140219029243776 main.py:352] /content/automl/efficientdet/efficientdet-d0/model has no global step info: stop!

3.2 COCO evaluation on test-dev.



In [0]:

    
# Eval on test-dev is slow (~40 mins), please be cautious. 
RUN_EXPENSIVE_TEST_DEV_EVAL = True  #@param

if RUN_EXPENSIVE_TEST_DEV_EVAL == True:
  !rm *.zip *.tar tfrecord/ val2017/   # Cleanup disk space
  # Download and convert test-dev data.
  if "test2017" not in os.listdir():
    !wget http://images.cocodataset.org/zips/test2017.zip
    !unzip -q test2017.zip
    !wget http://images.cocodataset.org/annotations/image_info_test2017.zip
    !unzip image_info_test2017.zip

    !mkdir tfrecord
    !PYTHONPATH=".:$PYTHONPATH"  python dataset/create_coco_tfrecord.py \
          --image_dir=test2017 \
          --image_info_file=annotations/image_info_test-dev2017.json \
          --output_file_prefix=tfrecord/testdev \
          --num_shards=32

  # Evalute on validation set: non-empty testdev_dir is the key pararmeter.
  # Also, test-dev has 20288 images rather than val 5000 images.
  !mkdir testdev_output
  !python main.py --mode=eval  \
      --model_name={MODEL}  --model_dir={ckpt_path}  \
      --validation_file_pattern=tfrecord/testdev*  \
      --eval_batch_size=8  --eval_samples=20288 \
      --testdev_dir='testdev_output'
  !rm -rf test2017  # delete images to release disk space.
  # Now you can submit testdev_output/detections_test-dev2017_test_results.json to
  # coco server: https://competitions.codalab.org/competitions/20794#participate

4. Training EfficientDets on PASCAL.

4.1 Prepare data



In [0]:

    
# Get pascal voc 2012 trainval data
import os
if 'VOCdevkit' not in os.listdir():
  !wget http://host.robots.ox.ac.uk/pascal/VOC/voc2012/VOCtrainval_11-May-2012.tar
  !tar xf VOCtrainval_11-May-2012.tar

  !mkdir tfrecord
  !PYTHONPATH=".:$PYTHONPATH"  python dataset/create_pascal_tfrecord.py  \
    --data_dir=VOCdevkit --year=VOC2012  --output_path=tfrecord/pascal

# Pascal has 5717 train images with 100 shards epoch, here we use a single shard
# for demo, but users should use all shards pascal-*-of-00100.tfrecord.
file_pattern = 'pascal-00000-of-00100.tfrecord'  # @param
images_per_epoch = 57 * len(tf.io.gfile.glob('tfrecord/' + file_pattern))
images_per_epoch = images_per_epoch // 8 * 8  # round to 64.
print('images_per_epoch = {}'.format(images_per_epoch))









    



--2020-06-01 05:10:36--  http://host.robots.ox.ac.uk/pascal/VOC/voc2012/VOCtrainval_11-May-2012.tar
Resolving host.robots.ox.ac.uk (host.robots.ox.ac.uk)... 129.67.94.152
Connecting to host.robots.ox.ac.uk (host.robots.ox.ac.uk)|129.67.94.152|:80... connected.
HTTP request sent, awaiting response... Read error (Connection reset by peer) in headers.
Retrying.

--2020-06-01 05:11:57--  (try: 2)  http://host.robots.ox.ac.uk/pascal/VOC/voc2012/VOCtrainval_11-May-2012.tar
Connecting to host.robots.ox.ac.uk (host.robots.ox.ac.uk)|129.67.94.152|:80... connected.
HTTP request sent, awaiting response... 200 OK
Length: 1999639040 (1.9G) [application/x-tar]
Saving to: ‘VOCtrainval_11-May-2012.tar’

VOCtrainval_11-May- 100%[===================>]   1.86G  13.7MB/s    in 2m 22s  

2020-06-01 05:20:26 (13.5 MB/s) - ‘VOCtrainval_11-May-2012.tar’ saved [1999639040/1999639040]

2020-06-01 05:21:21.622800: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.1
I0601 05:21:23.172357 140062335764352 create_pascal_tfrecord.py:254] writing to output path: tfrecord/pascal
I0601 05:21:23.176307 140062335764352 create_pascal_tfrecord.py:278] Reading from PASCAL VOC2012 dataset.
I0601 05:21:23.187967 140062335764352 create_pascal_tfrecord.py:287] On image 0 of 5717
/content/automl/efficientdet/dataset/tfrecord_util.py:78: FutureWarning: The behavior of this method will change in future versions. Use specific 'len(elem)' or 'elem is not None' test instead.
  if not xml:
I0601 05:21:23.309382 140062335764352 create_pascal_tfrecord.py:287] On image 100 of 5717
I0601 05:21:23.427004 140062335764352 create_pascal_tfrecord.py:287] On image 200 of 5717
I0601 05:21:23.540375 140062335764352 create_pascal_tfrecord.py:287] On image 300 of 5717
I0601 05:21:23.662686 140062335764352 create_pascal_tfrecord.py:287] On image 400 of 5717
I0601 05:21:23.777309 140062335764352 create_pascal_tfrecord.py:287] On image 500 of 5717
I0601 05:21:23.893491 140062335764352 create_pascal_tfrecord.py:287] On image 600 of 5717
I0601 05:21:24.012030 140062335764352 create_pascal_tfrecord.py:287] On image 700 of 5717
I0601 05:21:24.131017 140062335764352 create_pascal_tfrecord.py:287] On image 800 of 5717
I0601 05:21:24.249218 140062335764352 create_pascal_tfrecord.py:287] On image 900 of 5717
I0601 05:21:24.367812 140062335764352 create_pascal_tfrecord.py:287] On image 1000 of 5717
I0601 05:21:24.524589 140062335764352 create_pascal_tfrecord.py:287] On image 1100 of 5717
I0601 05:21:24.671167 140062335764352 create_pascal_tfrecord.py:287] On image 1200 of 5717
I0601 05:21:24.795677 140062335764352 create_pascal_tfrecord.py:287] On image 1300 of 5717
I0601 05:21:24.915077 140062335764352 create_pascal_tfrecord.py:287] On image 1400 of 5717
I0601 05:21:25.049030 140062335764352 create_pascal_tfrecord.py:287] On image 1500 of 5717
I0601 05:21:25.182934 140062335764352 create_pascal_tfrecord.py:287] On image 1600 of 5717
I0601 05:21:25.326709 140062335764352 create_pascal_tfrecord.py:287] On image 1700 of 5717
I0601 05:21:25.466770 140062335764352 create_pascal_tfrecord.py:287] On image 1800 of 5717
I0601 05:21:25.591415 140062335764352 create_pascal_tfrecord.py:287] On image 1900 of 5717
I0601 05:21:25.720433 140062335764352 create_pascal_tfrecord.py:287] On image 2000 of 5717
I0601 05:21:25.838430 140062335764352 create_pascal_tfrecord.py:287] On image 2100 of 5717
I0601 05:21:25.957128 140062335764352 create_pascal_tfrecord.py:287] On image 2200 of 5717
I0601 05:21:26.078567 140062335764352 create_pascal_tfrecord.py:287] On image 2300 of 5717
I0601 05:21:26.262683 140062335764352 create_pascal_tfrecord.py:287] On image 2400 of 5717
I0601 05:21:26.406095 140062335764352 create_pascal_tfrecord.py:287] On image 2500 of 5717
I0601 05:21:26.548389 140062335764352 create_pascal_tfrecord.py:287] On image 2600 of 5717
I0601 05:21:26.699499 140062335764352 create_pascal_tfrecord.py:287] On image 2700 of 5717
I0601 05:21:26.829630 140062335764352 create_pascal_tfrecord.py:287] On image 2800 of 5717
I0601 05:21:26.967325 140062335764352 create_pascal_tfrecord.py:287] On image 2900 of 5717
I0601 05:21:27.102216 140062335764352 create_pascal_tfrecord.py:287] On image 3000 of 5717
I0601 05:21:27.228751 140062335764352 create_pascal_tfrecord.py:287] On image 3100 of 5717
I0601 05:21:27.352252 140062335764352 create_pascal_tfrecord.py:287] On image 3200 of 5717
I0601 05:21:27.477318 140062335764352 create_pascal_tfrecord.py:287] On image 3300 of 5717
I0601 05:21:27.692891 140062335764352 create_pascal_tfrecord.py:287] On image 3400 of 5717
I0601 05:21:27.826326 140062335764352 create_pascal_tfrecord.py:287] On image 3500 of 5717
I0601 05:21:27.950285 140062335764352 create_pascal_tfrecord.py:287] On image 3600 of 5717
I0601 05:21:28.072462 140062335764352 create_pascal_tfrecord.py:287] On image 3700 of 5717
I0601 05:21:28.199001 140062335764352 create_pascal_tfrecord.py:287] On image 3800 of 5717
I0601 05:21:28.321695 140062335764352 create_pascal_tfrecord.py:287] On image 3900 of 5717
I0601 05:21:28.457443 140062335764352 create_pascal_tfrecord.py:287] On image 4000 of 5717
I0601 05:21:28.582388 140062335764352 create_pascal_tfrecord.py:287] On image 4100 of 5717
I0601 05:21:28.714639 140062335764352 create_pascal_tfrecord.py:287] On image 4200 of 5717
I0601 05:21:28.838026 140062335764352 create_pascal_tfrecord.py:287] On image 4300 of 5717
I0601 05:21:28.977360 140062335764352 create_pascal_tfrecord.py:287] On image 4400 of 5717
I0601 05:21:29.144394 140062335764352 create_pascal_tfrecord.py:287] On image 4500 of 5717
I0601 05:21:29.278243 140062335764352 create_pascal_tfrecord.py:287] On image 4600 of 5717
I0601 05:21:29.412878 140062335764352 create_pascal_tfrecord.py:287] On image 4700 of 5717
I0601 05:21:29.556532 140062335764352 create_pascal_tfrecord.py:287] On image 4800 of 5717
I0601 05:21:29.694362 140062335764352 create_pascal_tfrecord.py:287] On image 4900 of 5717
I0601 05:21:29.831758 140062335764352 create_pascal_tfrecord.py:287] On image 5000 of 5717
I0601 05:21:30.044910 140062335764352 create_pascal_tfrecord.py:287] On image 5100 of 5717
I0601 05:21:30.176852 140062335764352 create_pascal_tfrecord.py:287] On image 5200 of 5717
I0601 05:21:30.303508 140062335764352 create_pascal_tfrecord.py:287] On image 5300 of 5717
I0601 05:21:30.431598 140062335764352 create_pascal_tfrecord.py:287] On image 5400 of 5717
I0601 05:21:30.558156 140062335764352 create_pascal_tfrecord.py:287] On image 5500 of 5717
I0601 05:21:30.683752 140062335764352 create_pascal_tfrecord.py:287] On image 5600 of 5717
I0601 05:21:30.810799 140062335764352 create_pascal_tfrecord.py:287] On image 5700 of 5717
images_per_epoch = 56

4.2 Train Pascal VOC 2012 from ImageNet checkpoint for Backbone.



In [0]:

    
# Train efficientdet from scratch with backbone checkpoint.
backbone_name = {
    'efficientdet-d0': 'efficientnet-b0',
    'efficientdet-d1': 'efficientnet-b1',
    'efficientdet-d2': 'efficientnet-b2',
    'efficientdet-d3': 'efficientnet-b3',
    'efficientdet-d4': 'efficientnet-b4',
    'efficientdet-d5': 'efficientnet-b5',
    'efficientdet-d6': 'efficientnet-b6',
    'efficientdet-d7': 'efficientnet-b6',
}[MODEL]


# generating train tfrecord is large, so we skip the execution here.
import os
if backbone_name not in os.listdir():
  !wget https://storage.googleapis.com/cloud-tpu-checkpoints/efficientnet/ckptsaug/{backbone_name}.tar.gz
  !tar xf {backbone_name}.tar.gz

!mkdir /tmp/model_dir
# key option: use --backbone_ckpt rather than --ckpt.
# Don't use ema since we only train a few steps.
!python main.py --mode=train_and_eval \
    --training_file_pattern=tfrecord/{file_pattern} \
    --validation_file_pattern=tfrecord/{file_pattern} \
    --model_name={MODEL} \
    --model_dir=/tmp/model_dir/{MODEL}-scratch  \
    --backbone_ckpt={backbone_name} \
    --train_batch_size=4 \
    --eval_batch_size=4 --eval_samples={images_per_epoch}  \
    --num_examples_per_epoch={images_per_epoch}  --num_epochs=1  \
    --hparams="num_classes=20,moving_average_decay=0,mixed_precision=true"









    



--2020-06-01 05:36:57--  https://storage.googleapis.com/cloud-tpu-checkpoints/efficientnet/ckptsaug/efficientnet-b0.tar.gz
Resolving storage.googleapis.com (storage.googleapis.com)... 173.194.216.128, 2607:f8b0:400c:c13::80
Connecting to storage.googleapis.com (storage.googleapis.com)|173.194.216.128|:443... connected.
HTTP request sent, awaiting response... 200 OK
Length: 39302973 (37M) [application/gzip]
Saving to: ‘efficientnet-b0.tar.gz’

efficientnet-b0.tar 100%[===================>]  37.48M   162MB/s    in 0.2s    

2020-06-01 05:36:58 (162 MB/s) - ‘efficientnet-b0.tar.gz’ saved [39302973/39302973]

2020-06-01 05:37:00.448456: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.1
WARNING:tensorflow:From main.py:234: The name tf.estimator.tpu.TPUConfig is deprecated. Please use tf.compat.v1.estimator.tpu.TPUConfig instead.

W0601 05:37:02.781105 140460941842304 module_wrapper.py:138] From main.py:234: The name tf.estimator.tpu.TPUConfig is deprecated. Please use tf.compat.v1.estimator.tpu.TPUConfig instead.

WARNING:tensorflow:From main.py:239: The name tf.estimator.tpu.InputPipelineConfig is deprecated. Please use tf.compat.v1.estimator.tpu.InputPipelineConfig instead.

W0601 05:37:02.781442 140460941842304 module_wrapper.py:138] From main.py:239: The name tf.estimator.tpu.InputPipelineConfig is deprecated. Please use tf.compat.v1.estimator.tpu.InputPipelineConfig instead.

WARNING:tensorflow:From main.py:247: The name tf.estimator.tpu.RunConfig is deprecated. Please use tf.compat.v1.estimator.tpu.RunConfig instead.

W0601 05:37:02.781647 140460941842304 module_wrapper.py:138] From main.py:247: The name tf.estimator.tpu.RunConfig is deprecated. Please use tf.compat.v1.estimator.tpu.RunConfig instead.

I0601 05:37:02.781888 140460941842304 main.py:262] {'name': 'efficientdet-d0', 'act_type': 'swish', 'image_size': (512, 512), 'target_size': None, 'input_rand_hflip': True, 'train_scale_min': 0.1, 'train_scale_max': 2.0, 'autoaugment_policy': None, 'use_augmix': False, 'augmix_params': (3, -1, 1), 'num_classes': 20, 'skip_crowd_during_training': True, 'label_id_mapping': None, 'max_instances_per_image': 100, 'min_level': 3, 'max_level': 7, 'num_scales': 3, 'aspect_ratios': [(1.0, 1.0), (1.4, 0.7), (0.7, 1.4)], 'anchor_scale': 4.0, 'is_training_bn': True, 'momentum': 0.9, 'optimizer': 'sgd', 'learning_rate': 0.08, 'lr_warmup_init': 0.008, 'lr_warmup_epoch': 1.0, 'first_lr_drop_epoch': 200.0, 'second_lr_drop_epoch': 250.0, 'poly_lr_power': 0.9, 'clip_gradients_norm': 10.0, 'num_epochs': 1, 'data_format': 'channels_last', 'alpha': 0.25, 'gamma': 1.5, 'delta': 0.1, 'box_loss_weight': 50.0, 'iou_loss_type': None, 'iou_loss_weight': 1.0, 'weight_decay': 4e-05, 'strategy': None, 'precision': 'mixed_float16', 'box_class_repeats': 3, 'fpn_cell_repeats': 3, 'fpn_num_filters': 64, 'separable_conv': True, 'apply_bn_for_resampling': True, 'conv_after_downsample': False, 'conv_bn_act_pattern': False, 'use_native_resize_op': True, 'pooling_type': None, 'fpn_name': None, 'fpn_weight_method': None, 'fpn_config': None, 'survival_prob': None, 'img_summary_steps': None, 'lr_decay_method': 'cosine', 'moving_average_decay': 0, 'ckpt_var_scope': None, 'var_exclude_expr': '.*/class-predict/.*', 'backbone_name': 'efficientnet-b0', 'backbone_config': None, 'var_freeze_expr': None, 'resnet_depth': 50, 'model_name': 'efficientdet-d0', 'iterations_per_loop': 100, 'model_dir': '/tmp/model_dir/efficientdet-d0-scratch', 'num_shards': 8, 'num_examples_per_epoch': 56, 'backbone_ckpt': 'efficientnet-b0', 'ckpt': None, 'val_json_file': None, 'testdev_dir': None, 'mode': 'train_and_eval'}
I0601 05:37:02.782036 140460941842304 main.py:373] Starting training cycle, epoch: 0.
WARNING:tensorflow:From main.py:374: The name tf.estimator.tpu.TPUEstimator is deprecated. Please use tf.compat.v1.estimator.tpu.TPUEstimator instead.

W0601 05:37:02.782747 140460941842304 module_wrapper.py:138] From main.py:374: The name tf.estimator.tpu.TPUEstimator is deprecated. Please use tf.compat.v1.estimator.tpu.TPUEstimator instead.

INFO:tensorflow:Using config: {'_model_dir': '/tmp/model_dir/efficientdet-d0-scratch', '_tf_random_seed': None, '_save_summary_steps': 100, '_save_checkpoints_steps': None, '_save_checkpoints_secs': 600, '_session_config': allow_soft_placement: true
, '_keep_checkpoint_max': 5, '_keep_checkpoint_every_n_hours': 10000, '_log_step_count_steps': None, '_train_distribute': None, '_device_fn': None, '_protocol': None, '_eval_distribute': None, '_experimental_distribute': None, '_experimental_max_worker_delay_secs': None, '_session_creation_timeout_secs': 7200, '_service': None, '_cluster_spec': ClusterSpec({}), '_task_type': 'worker', '_task_id': 0, '_global_id_in_cluster': 0, '_master': '', '_evaluation_master': '', '_is_chief': True, '_num_ps_replicas': 0, '_num_worker_replicas': 1, '_tpu_config': TPUConfig(iterations_per_loop=100, num_shards=8, num_cores_per_replica=None, per_host_input_for_training=3, tpu_job_name=None, initial_infeed_sleep_secs=None, input_partition_dims=None, eval_training_input_configuration=2, experimental_host_call_every_n_steps=1), '_cluster': None}
I0601 05:37:02.783222 140460941842304 estimator.py:191] Using config: {'_model_dir': '/tmp/model_dir/efficientdet-d0-scratch', '_tf_random_seed': None, '_save_summary_steps': 100, '_save_checkpoints_steps': None, '_save_checkpoints_secs': 600, '_session_config': allow_soft_placement: true
, '_keep_checkpoint_max': 5, '_keep_checkpoint_every_n_hours': 10000, '_log_step_count_steps': None, '_train_distribute': None, '_device_fn': None, '_protocol': None, '_eval_distribute': None, '_experimental_distribute': None, '_experimental_max_worker_delay_secs': None, '_session_creation_timeout_secs': 7200, '_service': None, '_cluster_spec': ClusterSpec({}), '_task_type': 'worker', '_task_id': 0, '_global_id_in_cluster': 0, '_master': '', '_evaluation_master': '', '_is_chief': True, '_num_ps_replicas': 0, '_num_worker_replicas': 1, '_tpu_config': TPUConfig(iterations_per_loop=100, num_shards=8, num_cores_per_replica=None, per_host_input_for_training=3, tpu_job_name=None, initial_infeed_sleep_secs=None, input_partition_dims=None, eval_training_input_configuration=2, experimental_host_call_every_n_steps=1), '_cluster': None}
INFO:tensorflow:_TPUContext: eval_on_tpu True
I0601 05:37:02.783910 140460941842304 tpu_context.py:216] _TPUContext: eval_on_tpu True
WARNING:tensorflow:eval_on_tpu ignored because use_tpu is False.
W0601 05:37:02.784342 140460941842304 tpu_context.py:218] eval_on_tpu ignored because use_tpu is False.
WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/ops/resource_variable_ops.py:1666: calling BaseResourceVariable.__init__ (from tensorflow.python.ops.resource_variable_ops) with constraint is deprecated and will be removed in a future version.
Instructions for updating:
If using Keras pass *_constraint arguments to layers.
W0601 05:37:02.789717 140460941842304 deprecation.py:506] From /usr/local/lib/python3.6/dist-packages/tensorflow/python/ops/resource_variable_ops.py:1666: calling BaseResourceVariable.__init__ (from tensorflow.python.ops.resource_variable_ops) with constraint is deprecated and will be removed in a future version.
Instructions for updating:
If using Keras pass *_constraint arguments to layers.
WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/training/training_util.py:236: Variable.initialized_value (from tensorflow.python.ops.variables) is deprecated and will be removed in a future version.
Instructions for updating:
Use Variable.read_value. Variables in 2.X are initialized automatically both in eager and graph (inside tf.defun) contexts.
W0601 05:37:02.790105 140460941842304 deprecation.py:323] From /usr/local/lib/python3.6/dist-packages/tensorflow/python/training/training_util.py:236: Variable.initialized_value (from tensorflow.python.ops.variables) is deprecated and will be removed in a future version.
Instructions for updating:
Use Variable.read_value. Variables in 2.X are initialized automatically both in eager and graph (inside tf.defun) contexts.
2020-06-01 05:37:02.800769: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcuda.so.1
2020-06-01 05:37:02.836188: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:37:02.836775: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1561] Found device 0 with properties: 
pciBusID: 0000:00:04.0 name: Tesla T4 computeCapability: 7.5
coreClock: 1.59GHz coreCount: 40 deviceMemorySize: 14.73GiB deviceMemoryBandwidth: 298.08GiB/s
2020-06-01 05:37:02.836815: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.1
2020-06-01 05:37:02.839147: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcublas.so.10
2020-06-01 05:37:02.848785: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcufft.so.10
2020-06-01 05:37:02.849571: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcurand.so.10
2020-06-01 05:37:02.857847: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusolver.so.10
2020-06-01 05:37:02.858945: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusparse.so.10
2020-06-01 05:37:02.871841: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudnn.so.7
2020-06-01 05:37:02.872030: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:37:02.872713: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:37:02.873299: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1703] Adding visible gpu devices: 0
WARNING:tensorflow:From /content/automl/efficientdet/dataloader.py:363: parallel_interleave (from tensorflow.python.data.experimental.ops.interleave_ops) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.data.Dataset.interleave(map_func, cycle_length, block_length, num_parallel_calls=tf.data.experimental.AUTOTUNE)` instead. If sloppy execution is desired, use `tf.data.Options.experimental_deterministic`.
W0601 05:37:02.904621 140460941842304 deprecation.py:323] From /content/automl/efficientdet/dataloader.py:363: parallel_interleave (from tensorflow.python.data.experimental.ops.interleave_ops) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.data.Dataset.interleave(map_func, cycle_length, block_length, num_parallel_calls=tf.data.experimental.AUTOTUNE)` instead. If sloppy execution is desired, use `tf.data.Options.experimental_deterministic`.
I0601 05:37:03.094781 140460941842304 dataloader.py:83] target_size = (512, 512), output_size = (512, 512)
INFO:tensorflow:Calling model_fn.
I0601 05:37:03.714015 140460941842304 estimator.py:1169] Calling model_fn.
INFO:tensorflow:Running train on CPU/GPU
I0601 05:37:03.714310 140460941842304 tpu_estimator.py:3171] Running train on CPU/GPU
I0601 05:37:03.714726 140460941842304 utils.py:595] use mixed precision policy name mixed_float16
I0601 05:37:03.719992 140460941842304 efficientdet_arch.py:720] act_type: swish
alpha: 0.25
anchor_scale: 4.0
apply_bn_for_resampling: true
aspect_ratios:
- !!python/tuple
    - 1.0
    - 1.0
- !!python/tuple
    - 1.4
    - 0.7
- !!python/tuple
    - 0.7
    - 1.4
augmix_params: !!python/tuple
- 3
- -1
- 1
autoaugment_policy: null
backbone_ckpt: efficientnet-b0
backbone_config: null
backbone_name: efficientnet-b0
batch_size: 4
box_class_repeats: 3
box_loss_weight: 50.0
ckpt: null
ckpt_var_scope: null
clip_gradients_norm: 10.0
conv_after_downsample: false
conv_bn_act_pattern: false
data_format: channels_last
delta: 0.1
first_lr_drop_epoch: 200.0
fpn_cell_repeats: 3
fpn_config: null
fpn_name: null
fpn_num_filters: 64
fpn_weight_method: null
gamma: 1.5
image_size: !!python/tuple
- 512
- 512
img_summary_steps: null
input_rand_hflip: true
iou_loss_type: null
iou_loss_weight: 1.0
is_training_bn: true
iterations_per_loop: 100
label_id_mapping: null
learning_rate: 0.08
lr_decay_method: cosine
lr_warmup_epoch: 1.0
lr_warmup_init: 0.008
max_instances_per_image: 100
max_level: 7
min_level: 3
mode: train_and_eval
model_dir: /tmp/model_dir/efficientdet-d0-scratch
model_name: efficientdet-d0
momentum: 0.9
moving_average_decay: 0
name: efficientdet-d0
num_classes: 20
num_epochs: 1
num_examples_per_epoch: 56
num_scales: 3
num_shards: 8
optimizer: sgd
poly_lr_power: 0.9
pooling_type: null
precision: mixed_float16
resnet_depth: 50
second_lr_drop_epoch: 250.0
separable_conv: true
skip_crowd_during_training: true
strategy: null
survival_prob: null
target_size: null
testdev_dir: null
train_scale_max: 2.0
train_scale_min: 0.1
use_augmix: false
use_native_resize_op: true
use_tpu: false
val_json_file: null
var_exclude_expr: .*/class-predict/.*
var_freeze_expr: null
weight_decay: 4.0e-05

I0601 05:37:03.724804 140460941842304 efficientnet_builder.py:224] global_params= GlobalParams(batch_norm_momentum=0.99, batch_norm_epsilon=0.001, dropout_rate=0.2, data_format='channels_last', num_classes=1000, width_coefficient=1.0, depth_coefficient=1.0, depth_divisor=8, min_depth=None, survival_prob=0.0, relu_fn=functools.partial(<function activation_fn at 0x7fbf45bc8d90>, act_type='swish'), batch_norm=<class 'utils.BatchNormalization'>, use_se=True, local_pooling=None, condconv_num_experts=None, clip_projection_output=False, blocks_args=['r1_k3_s11_e1_i32_o16_se0.25', 'r2_k3_s22_e6_i16_o24_se0.25', 'r2_k5_s22_e6_i24_o40_se0.25', 'r3_k3_s22_e6_i40_o80_se0.25', 'r3_k5_s11_e6_i80_o112_se0.25', 'r4_k5_s22_e6_i112_o192_se0.25', 'r1_k3_s11_e6_i192_o320_se0.25'], fix_head_stem=None)
I0601 05:37:06.724712 140460941842304 api.py:587] Built stem layers with output shape: (4, 256, 256, 32)
I0601 05:37:09.435570 140460941842304 api.py:587] Block mb_conv_block  input shape: (4, 256, 256, 32)
I0601 05:37:09.477118 140460941842304 api.py:587] DWConv shape: (4, 256, 256, 32)
I0601 05:37:09.638726 140460941842304 api.py:587] Built Squeeze and Excitation with tensor shape: (4, 1, 1, 32)
I0601 05:37:09.676554 140460941842304 api.py:587] Project shape: (4, 256, 256, 16)
I0601 05:37:09.729519 140460941842304 api.py:587] Block mb_conv_block_1  input shape: (4, 256, 256, 16)
I0601 05:37:09.769014 140460941842304 api.py:587] Expand shape: (4, 256, 256, 96)
I0601 05:37:09.808395 140460941842304 api.py:587] DWConv shape: (4, 128, 128, 96)
I0601 05:37:09.844948 140460941842304 api.py:587] Built Squeeze and Excitation with tensor shape: (4, 1, 1, 96)
I0601 05:37:09.884656 140460941842304 api.py:587] Project shape: (4, 128, 128, 24)
I0601 05:37:09.894854 140460941842304 api.py:587] Block mb_conv_block_2  input shape: (4, 128, 128, 24)
I0601 05:37:09.938410 140460941842304 api.py:587] Expand shape: (4, 128, 128, 144)
I0601 05:37:09.976691 140460941842304 api.py:587] DWConv shape: (4, 128, 128, 144)
I0601 05:37:10.013263 140460941842304 api.py:587] Built Squeeze and Excitation with tensor shape: (4, 1, 1, 144)
I0601 05:37:10.061878 140460941842304 api.py:587] Project shape: (4, 128, 128, 24)
I0601 05:37:10.071763 140460941842304 api.py:587] Block mb_conv_block_3  input shape: (4, 128, 128, 24)
I0601 05:37:10.116038 140460941842304 api.py:587] Expand shape: (4, 128, 128, 144)
I0601 05:37:10.158890 140460941842304 api.py:587] DWConv shape: (4, 64, 64, 144)
I0601 05:37:10.195446 140460941842304 api.py:587] Built Squeeze and Excitation with tensor shape: (4, 1, 1, 144)
I0601 05:37:10.236959 140460941842304 api.py:587] Project shape: (4, 64, 64, 40)
I0601 05:37:10.246315 140460941842304 api.py:587] Block mb_conv_block_4  input shape: (4, 64, 64, 40)
I0601 05:37:10.285769 140460941842304 api.py:587] Expand shape: (4, 64, 64, 240)
I0601 05:37:10.331516 140460941842304 api.py:587] DWConv shape: (4, 64, 64, 240)
I0601 05:37:10.366286 140460941842304 api.py:587] Built Squeeze and Excitation with tensor shape: (4, 1, 1, 240)
I0601 05:37:10.403537 140460941842304 api.py:587] Project shape: (4, 64, 64, 40)
I0601 05:37:10.418687 140460941842304 api.py:587] Block mb_conv_block_5  input shape: (4, 64, 64, 40)
I0601 05:37:10.457679 140460941842304 api.py:587] Expand shape: (4, 64, 64, 240)
I0601 05:37:10.497891 140460941842304 api.py:587] DWConv shape: (4, 32, 32, 240)
I0601 05:37:10.532069 140460941842304 api.py:587] Built Squeeze and Excitation with tensor shape: (4, 1, 1, 240)
I0601 05:37:10.571993 140460941842304 api.py:587] Project shape: (4, 32, 32, 80)
I0601 05:37:10.582987 140460941842304 api.py:587] Block mb_conv_block_6  input shape: (4, 32, 32, 80)
I0601 05:37:10.630499 140460941842304 api.py:587] Expand shape: (4, 32, 32, 480)
I0601 05:37:10.675540 140460941842304 api.py:587] DWConv shape: (4, 32, 32, 480)
I0601 05:37:10.711717 140460941842304 api.py:587] Built Squeeze and Excitation with tensor shape: (4, 1, 1, 480)
I0601 05:37:10.753072 140460941842304 api.py:587] Project shape: (4, 32, 32, 80)
I0601 05:37:10.762947 140460941842304 api.py:587] Block mb_conv_block_7  input shape: (4, 32, 32, 80)
I0601 05:37:10.804822 140460941842304 api.py:587] Expand shape: (4, 32, 32, 480)
I0601 05:37:10.850904 140460941842304 api.py:587] DWConv shape: (4, 32, 32, 480)
I0601 05:37:10.886982 140460941842304 api.py:587] Built Squeeze and Excitation with tensor shape: (4, 1, 1, 480)
I0601 05:37:10.926297 140460941842304 api.py:587] Project shape: (4, 32, 32, 80)
I0601 05:37:10.936717 140460941842304 api.py:587] Block mb_conv_block_8  input shape: (4, 32, 32, 80)
I0601 05:37:10.979605 140460941842304 api.py:587] Expand shape: (4, 32, 32, 480)
I0601 05:37:11.026784 140460941842304 api.py:587] DWConv shape: (4, 32, 32, 480)
I0601 05:37:11.067464 140460941842304 api.py:587] Built Squeeze and Excitation with tensor shape: (4, 1, 1, 480)
I0601 05:37:11.106633 140460941842304 api.py:587] Project shape: (4, 32, 32, 112)
I0601 05:37:11.116317 140460941842304 api.py:587] Block mb_conv_block_9  input shape: (4, 32, 32, 112)
I0601 05:37:11.158664 140460941842304 api.py:587] Expand shape: (4, 32, 32, 672)
I0601 05:37:11.203032 140460941842304 api.py:587] DWConv shape: (4, 32, 32, 672)
I0601 05:37:11.242812 140460941842304 api.py:587] Built Squeeze and Excitation with tensor shape: (4, 1, 1, 672)
I0601 05:37:11.280435 140460941842304 api.py:587] Project shape: (4, 32, 32, 112)
I0601 05:37:11.289857 140460941842304 api.py:587] Block mb_conv_block_10  input shape: (4, 32, 32, 112)
I0601 05:37:11.334518 140460941842304 api.py:587] Expand shape: (4, 32, 32, 672)
I0601 05:37:11.375933 140460941842304 api.py:587] DWConv shape: (4, 32, 32, 672)
I0601 05:37:11.413398 140460941842304 api.py:587] Built Squeeze and Excitation with tensor shape: (4, 1, 1, 672)
I0601 05:37:11.456761 140460941842304 api.py:587] Project shape: (4, 32, 32, 112)
I0601 05:37:11.466616 140460941842304 api.py:587] Block mb_conv_block_11  input shape: (4, 32, 32, 112)
I0601 05:37:11.506339 140460941842304 api.py:587] Expand shape: (4, 32, 32, 672)
I0601 05:37:11.547044 140460941842304 api.py:587] DWConv shape: (4, 16, 16, 672)
I0601 05:37:11.586595 140460941842304 api.py:587] Built Squeeze and Excitation with tensor shape: (4, 1, 1, 672)
I0601 05:37:11.626645 140460941842304 api.py:587] Project shape: (4, 16, 16, 192)
I0601 05:37:11.640136 140460941842304 api.py:587] Block mb_conv_block_12  input shape: (4, 16, 16, 192)
I0601 05:37:11.687014 140460941842304 api.py:587] Expand shape: (4, 16, 16, 1152)
I0601 05:37:11.730680 140460941842304 api.py:587] DWConv shape: (4, 16, 16, 1152)
I0601 05:37:11.768292 140460941842304 api.py:587] Built Squeeze and Excitation with tensor shape: (4, 1, 1, 1152)
I0601 05:37:11.808110 140460941842304 api.py:587] Project shape: (4, 16, 16, 192)
I0601 05:37:11.817761 140460941842304 api.py:587] Block mb_conv_block_13  input shape: (4, 16, 16, 192)
I0601 05:37:11.865955 140460941842304 api.py:587] Expand shape: (4, 16, 16, 1152)
I0601 05:37:11.910137 140460941842304 api.py:587] DWConv shape: (4, 16, 16, 1152)
I0601 05:37:11.947402 140460941842304 api.py:587] Built Squeeze and Excitation with tensor shape: (4, 1, 1, 1152)
I0601 05:37:11.991628 140460941842304 api.py:587] Project shape: (4, 16, 16, 192)
I0601 05:37:12.002006 140460941842304 api.py:587] Block mb_conv_block_14  input shape: (4, 16, 16, 192)
I0601 05:37:12.057711 140460941842304 api.py:587] Expand shape: (4, 16, 16, 1152)
I0601 05:37:12.116494 140460941842304 api.py:587] DWConv shape: (4, 16, 16, 1152)
I0601 05:37:12.158459 140460941842304 api.py:587] Built Squeeze and Excitation with tensor shape: (4, 1, 1, 1152)
I0601 05:37:12.203221 140460941842304 api.py:587] Project shape: (4, 16, 16, 192)
I0601 05:37:12.212300 140460941842304 api.py:587] Block mb_conv_block_15  input shape: (4, 16, 16, 192)
I0601 05:37:12.260587 140460941842304 api.py:587] Expand shape: (4, 16, 16, 1152)
I0601 05:37:12.304630 140460941842304 api.py:587] DWConv shape: (4, 16, 16, 1152)
I0601 05:37:12.346603 140460941842304 api.py:587] Built Squeeze and Excitation with tensor shape: (4, 1, 1, 1152)
I0601 05:37:12.389416 140460941842304 api.py:587] Project shape: (4, 16, 16, 320)
WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/profiler/internal/flops_registry.py:142: tensor_shape_from_node_def_name (from tensorflow.python.framework.graph_util_impl) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.compat.v1.graph_util.tensor_shape_from_node_def_name`
W0601 05:37:12.399743 140460941842304 deprecation.py:323] From /usr/local/lib/python3.6/dist-packages/tensorflow/python/profiler/internal/flops_registry.py:142: tensor_shape_from_node_def_name (from tensorflow.python.framework.graph_util_impl) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.compat.v1.graph_util.tensor_shape_from_node_def_name`
Parsing Inputs...
I0601 05:37:12.734671 140460941842304 efficientdet_arch.py:725] backbone params/flops = 3.595388M, 7.723650081B
WARNING:tensorflow:From /content/automl/efficientdet/efficientdet_arch.py:138: conv2d (from tensorflow.python.layers.convolutional) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.keras.layers.Conv2D` instead.
W0601 05:37:12.735137 140460941842304 deprecation.py:323] From /content/automl/efficientdet/efficientdet_arch.py:138: conv2d (from tensorflow.python.layers.convolutional) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.keras.layers.Conv2D` instead.
WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/layers/convolutional.py:424: Layer.apply (from tensorflow.python.keras.engine.base_layer_v1) is deprecated and will be removed in a future version.
Instructions for updating:
Please use `layer.__call__` method instead.
W0601 05:37:12.738550 140460941842304 deprecation.py:323] From /usr/local/lib/python3.6/dist-packages/tensorflow/python/layers/convolutional.py:424: Layer.apply (from tensorflow.python.keras.engine.base_layer_v1) is deprecated and will be removed in a future version.
Instructions for updating:
Please use `layer.__call__` method instead.
WARNING:tensorflow:From /content/automl/efficientdet/efficientdet_arch.py:164: max_pooling2d (from tensorflow.python.layers.pooling) is deprecated and will be removed in a future version.
Instructions for updating:
Use keras.layers.MaxPooling2D instead.
W0601 05:37:12.796389 140460941842304 deprecation.py:323] From /content/automl/efficientdet/efficientdet_arch.py:164: max_pooling2d (from tensorflow.python.layers.pooling) is deprecated and will be removed in a future version.
Instructions for updating:
Use keras.layers.MaxPooling2D instead.
I0601 05:37:12.802982 140460941842304 efficientdet_arch.py:471] building cell 0
I0601 05:37:12.803340 140460941842304 efficientdet_arch.py:648] fnode 0 : {'feat_level': 6, 'inputs_offsets': [3, 4]}
WARNING:tensorflow:From /content/automl/efficientdet/efficientdet_arch.py:685: separable_conv2d (from tensorflow.python.layers.convolutional) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.keras.layers.SeparableConv2D` instead.
W0601 05:37:12.816923 140460941842304 deprecation.py:323] From /content/automl/efficientdet/efficientdet_arch.py:685: separable_conv2d (from tensorflow.python.layers.convolutional) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.keras.layers.SeparableConv2D` instead.
I0601 05:37:12.879780 140460941842304 efficientdet_arch.py:648] fnode 1 : {'feat_level': 5, 'inputs_offsets': [2, 5]}
I0601 05:37:13.014620 140460941842304 efficientdet_arch.py:648] fnode 2 : {'feat_level': 4, 'inputs_offsets': [1, 6]}
I0601 05:37:13.156741 140460941842304 efficientdet_arch.py:648] fnode 3 : {'feat_level': 3, 'inputs_offsets': [0, 7]}
I0601 05:37:13.296745 140460941842304 efficientdet_arch.py:648] fnode 4 : {'feat_level': 4, 'inputs_offsets': [1, 7, 8]}
I0601 05:37:13.445665 140460941842304 efficientdet_arch.py:648] fnode 5 : {'feat_level': 5, 'inputs_offsets': [2, 6, 9]}
I0601 05:37:13.594432 140460941842304 efficientdet_arch.py:648] fnode 6 : {'feat_level': 6, 'inputs_offsets': [3, 5, 10]}
I0601 05:37:13.688446 140460941842304 efficientdet_arch.py:648] fnode 7 : {'feat_level': 7, 'inputs_offsets': [4, 11]}
I0601 05:37:13.871573 140460941842304 efficientdet_arch.py:471] building cell 1
I0601 05:37:13.871955 140460941842304 efficientdet_arch.py:648] fnode 0 : {'feat_level': 6, 'inputs_offsets': [3, 4]}
I0601 05:37:13.965792 140460941842304 efficientdet_arch.py:648] fnode 1 : {'feat_level': 5, 'inputs_offsets': [2, 5]}
I0601 05:37:14.063542 140460941842304 efficientdet_arch.py:648] fnode 2 : {'feat_level': 4, 'inputs_offsets': [1, 6]}
I0601 05:37:14.167066 140460941842304 efficientdet_arch.py:648] fnode 3 : {'feat_level': 3, 'inputs_offsets': [0, 7]}
I0601 05:37:14.261668 140460941842304 efficientdet_arch.py:648] fnode 4 : {'feat_level': 4, 'inputs_offsets': [1, 7, 8]}
I0601 05:37:14.359102 140460941842304 efficientdet_arch.py:648] fnode 5 : {'feat_level': 5, 'inputs_offsets': [2, 6, 9]}
I0601 05:37:14.460067 140460941842304 efficientdet_arch.py:648] fnode 6 : {'feat_level': 6, 'inputs_offsets': [3, 5, 10]}
I0601 05:37:14.556290 140460941842304 efficientdet_arch.py:648] fnode 7 : {'feat_level': 7, 'inputs_offsets': [4, 11]}
I0601 05:37:14.650365 140460941842304 efficientdet_arch.py:471] building cell 2
I0601 05:37:14.650751 140460941842304 efficientdet_arch.py:648] fnode 0 : {'feat_level': 6, 'inputs_offsets': [3, 4]}
I0601 05:37:14.746739 140460941842304 efficientdet_arch.py:648] fnode 1 : {'feat_level': 5, 'inputs_offsets': [2, 5]}
I0601 05:37:14.839383 140460941842304 efficientdet_arch.py:648] fnode 2 : {'feat_level': 4, 'inputs_offsets': [1, 6]}
I0601 05:37:14.936065 140460941842304 efficientdet_arch.py:648] fnode 3 : {'feat_level': 3, 'inputs_offsets': [0, 7]}
I0601 05:37:15.035170 140460941842304 efficientdet_arch.py:648] fnode 4 : {'feat_level': 4, 'inputs_offsets': [1, 7, 8]}
I0601 05:37:15.149712 140460941842304 efficientdet_arch.py:648] fnode 5 : {'feat_level': 5, 'inputs_offsets': [2, 6, 9]}
I0601 05:37:15.257548 140460941842304 efficientdet_arch.py:648] fnode 6 : {'feat_level': 6, 'inputs_offsets': [3, 5, 10]}
I0601 05:37:15.366530 140460941842304 efficientdet_arch.py:648] fnode 7 : {'feat_level': 7, 'inputs_offsets': [4, 11]}
Parsing Inputs...
I0601 05:37:16.045335 140460941842304 efficientdet_arch.py:730] backbone+fpn params/flops = 3.791669M, 8.294161259B
Parsing Inputs...
I0601 05:37:19.741914 140460941842304 efficientdet_arch.py:735] backbone+fpn+box params/flops = 3.839117M, 9.243552137B
I0601 05:37:19.742230 140460941842304 utils.py:595] use mixed precision policy name float32
I0601 05:37:19.746633 140460941842304 det_model_fn.py:97] LR schedule method: cosine
I0601 05:37:19.968963 140460941842304 utils.py:394] Adding scale summary ('lrn_rate', <tf.Tensor 'Select:0' shape=() dtype=float32>)
I0601 05:37:19.969849 140460941842304 utils.py:394] Adding scale summary ('trainloss/cls_loss', <tf.Tensor 'AddN:0' shape=() dtype=float32>)
I0601 05:37:19.970671 140460941842304 utils.py:394] Adding scale summary ('trainloss/box_loss', <tf.Tensor 'AddN_1:0' shape=() dtype=float32>)
I0601 05:37:19.971507 140460941842304 utils.py:394] Adding scale summary ('trainloss/det_loss', <tf.Tensor 'add_4:0' shape=() dtype=float32>)
I0601 05:37:19.972293 140460941842304 utils.py:394] Adding scale summary ('trainloss/reg_l2_loss', <tf.Tensor 'mul_14:0' shape=() dtype=float32>)
I0601 05:37:19.973071 140460941842304 utils.py:394] Adding scale summary ('trainloss/loss', <tf.Tensor 'add_5:0' shape=() dtype=float32>)
I0601 05:37:19.973890 140460941842304 det_model_fn.py:537] clip gradients norm by 10.000000
I0601 05:37:24.866811 140460941842304 utils.py:394] Adding scale summary ('gnorm', <tf.Tensor 'clip/global_norm/global_norm:0' shape=() dtype=float32>)
WARNING:tensorflow:From /content/automl/efficientdet/det_model_fn.py:659: The name tf.estimator.tpu.TPUEstimatorSpec is deprecated. Please use tf.compat.v1.estimator.tpu.TPUEstimatorSpec instead.

W0601 05:37:27.070672 140460941842304 module_wrapper.py:138] From /content/automl/efficientdet/det_model_fn.py:659: The name tf.estimator.tpu.TPUEstimatorSpec is deprecated. Please use tf.compat.v1.estimator.tpu.TPUEstimatorSpec instead.

I0601 05:37:27.116415 140460941842304 det_model_fn.py:638] restore variables from efficientnet-b0
I0601 05:37:27.116562 140460941842304 utils.py:73] Init model from checkpoint efficientnet-b0
I0601 05:37:27.121120 140460941842304 utils.py:108] Init efficientnet-b0/stem/conv2d/kernel from ckpt var efficientnet-b0/stem/conv2d/kernel
I0601 05:37:27.121282 140460941842304 utils.py:108] Init efficientnet-b0/stem/tpu_batch_normalization/gamma from ckpt var efficientnet-b0/stem/tpu_batch_normalization/gamma
I0601 05:37:27.121384 140460941842304 utils.py:108] Init efficientnet-b0/stem/tpu_batch_normalization/beta from ckpt var efficientnet-b0/stem/tpu_batch_normalization/beta
I0601 05:37:27.121477 140460941842304 utils.py:108] Init efficientnet-b0/stem/tpu_batch_normalization/moving_mean from ckpt var efficientnet-b0/stem/tpu_batch_normalization/moving_mean
I0601 05:37:27.121554 140460941842304 utils.py:108] Init efficientnet-b0/stem/tpu_batch_normalization/moving_variance from ckpt var efficientnet-b0/stem/tpu_batch_normalization/moving_variance
I0601 05:37:27.121651 140460941842304 utils.py:108] Init efficientnet-b0/blocks_0/depthwise_conv2d/depthwise_kernel from ckpt var efficientnet-b0/blocks_0/depthwise_conv2d/depthwise_kernel
I0601 05:37:27.121726 140460941842304 utils.py:108] Init efficientnet-b0/blocks_0/tpu_batch_normalization/gamma from ckpt var efficientnet-b0/blocks_0/tpu_batch_normalization/gamma
I0601 05:37:27.121800 140460941842304 utils.py:108] Init efficientnet-b0/blocks_0/tpu_batch_normalization/beta from ckpt var efficientnet-b0/blocks_0/tpu_batch_normalization/beta
I0601 05:37:27.121912 140460941842304 utils.py:108] Init efficientnet-b0/blocks_0/tpu_batch_normalization/moving_mean from ckpt var efficientnet-b0/blocks_0/tpu_batch_normalization/moving_mean
I0601 05:37:27.122007 140460941842304 utils.py:108] Init efficientnet-b0/blocks_0/tpu_batch_normalization/moving_variance from ckpt var efficientnet-b0/blocks_0/tpu_batch_normalization/moving_variance
I0601 05:37:27.122094 140460941842304 utils.py:108] Init efficientnet-b0/blocks_0/se/conv2d/kernel from ckpt var efficientnet-b0/blocks_0/se/conv2d/kernel
I0601 05:37:27.122166 140460941842304 utils.py:108] Init efficientnet-b0/blocks_0/se/conv2d/bias from ckpt var efficientnet-b0/blocks_0/se/conv2d/bias
I0601 05:37:27.122240 140460941842304 utils.py:108] Init efficientnet-b0/blocks_0/se/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_0/se/conv2d_1/kernel
I0601 05:37:27.122339 140460941842304 utils.py:108] Init efficientnet-b0/blocks_0/se/conv2d_1/bias from ckpt var efficientnet-b0/blocks_0/se/conv2d_1/bias
I0601 05:37:27.122414 140460941842304 utils.py:108] Init efficientnet-b0/blocks_0/conv2d/kernel from ckpt var efficientnet-b0/blocks_0/conv2d/kernel
I0601 05:37:27.122487 140460941842304 utils.py:108] Init efficientnet-b0/blocks_0/tpu_batch_normalization_1/gamma from ckpt var efficientnet-b0/blocks_0/tpu_batch_normalization_1/gamma
I0601 05:37:27.122561 140460941842304 utils.py:108] Init efficientnet-b0/blocks_0/tpu_batch_normalization_1/beta from ckpt var efficientnet-b0/blocks_0/tpu_batch_normalization_1/beta
I0601 05:37:27.122643 140460941842304 utils.py:108] Init efficientnet-b0/blocks_0/tpu_batch_normalization_1/moving_mean from ckpt var efficientnet-b0/blocks_0/tpu_batch_normalization_1/moving_mean
I0601 05:37:27.122731 140460941842304 utils.py:108] Init efficientnet-b0/blocks_0/tpu_batch_normalization_1/moving_variance from ckpt var efficientnet-b0/blocks_0/tpu_batch_normalization_1/moving_variance
I0601 05:37:27.122803 140460941842304 utils.py:108] Init efficientnet-b0/blocks_1/conv2d/kernel from ckpt var efficientnet-b0/blocks_1/conv2d/kernel
I0601 05:37:27.122874 140460941842304 utils.py:108] Init efficientnet-b0/blocks_1/tpu_batch_normalization/gamma from ckpt var efficientnet-b0/blocks_1/tpu_batch_normalization/gamma
I0601 05:37:27.122954 140460941842304 utils.py:108] Init efficientnet-b0/blocks_1/tpu_batch_normalization/beta from ckpt var efficientnet-b0/blocks_1/tpu_batch_normalization/beta
I0601 05:37:27.123035 140460941842304 utils.py:108] Init efficientnet-b0/blocks_1/tpu_batch_normalization/moving_mean from ckpt var efficientnet-b0/blocks_1/tpu_batch_normalization/moving_mean
I0601 05:37:27.123108 140460941842304 utils.py:108] Init efficientnet-b0/blocks_1/tpu_batch_normalization/moving_variance from ckpt var efficientnet-b0/blocks_1/tpu_batch_normalization/moving_variance
I0601 05:37:27.123179 140460941842304 utils.py:108] Init efficientnet-b0/blocks_1/depthwise_conv2d/depthwise_kernel from ckpt var efficientnet-b0/blocks_1/depthwise_conv2d/depthwise_kernel
I0601 05:37:27.123252 140460941842304 utils.py:108] Init efficientnet-b0/blocks_1/tpu_batch_normalization_1/gamma from ckpt var efficientnet-b0/blocks_1/tpu_batch_normalization_1/gamma
I0601 05:37:27.123349 140460941842304 utils.py:108] Init efficientnet-b0/blocks_1/tpu_batch_normalization_1/beta from ckpt var efficientnet-b0/blocks_1/tpu_batch_normalization_1/beta
I0601 05:37:27.123430 140460941842304 utils.py:108] Init efficientnet-b0/blocks_1/tpu_batch_normalization_1/moving_mean from ckpt var efficientnet-b0/blocks_1/tpu_batch_normalization_1/moving_mean
I0601 05:37:27.123515 140460941842304 utils.py:108] Init efficientnet-b0/blocks_1/tpu_batch_normalization_1/moving_variance from ckpt var efficientnet-b0/blocks_1/tpu_batch_normalization_1/moving_variance
I0601 05:37:27.123599 140460941842304 utils.py:108] Init efficientnet-b0/blocks_1/se/conv2d/kernel from ckpt var efficientnet-b0/blocks_1/se/conv2d/kernel
I0601 05:37:27.123682 140460941842304 utils.py:108] Init efficientnet-b0/blocks_1/se/conv2d/bias from ckpt var efficientnet-b0/blocks_1/se/conv2d/bias
I0601 05:37:27.123755 140460941842304 utils.py:108] Init efficientnet-b0/blocks_1/se/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_1/se/conv2d_1/kernel
I0601 05:37:27.123831 140460941842304 utils.py:108] Init efficientnet-b0/blocks_1/se/conv2d_1/bias from ckpt var efficientnet-b0/blocks_1/se/conv2d_1/bias
I0601 05:37:27.123944 140460941842304 utils.py:108] Init efficientnet-b0/blocks_1/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_1/conv2d_1/kernel
I0601 05:37:27.124024 140460941842304 utils.py:108] Init efficientnet-b0/blocks_1/tpu_batch_normalization_2/gamma from ckpt var efficientnet-b0/blocks_1/tpu_batch_normalization_2/gamma
I0601 05:37:27.124098 140460941842304 utils.py:108] Init efficientnet-b0/blocks_1/tpu_batch_normalization_2/beta from ckpt var efficientnet-b0/blocks_1/tpu_batch_normalization_2/beta
I0601 05:37:27.124171 140460941842304 utils.py:108] Init efficientnet-b0/blocks_1/tpu_batch_normalization_2/moving_mean from ckpt var efficientnet-b0/blocks_1/tpu_batch_normalization_2/moving_mean
I0601 05:37:27.124250 140460941842304 utils.py:108] Init efficientnet-b0/blocks_1/tpu_batch_normalization_2/moving_variance from ckpt var efficientnet-b0/blocks_1/tpu_batch_normalization_2/moving_variance
I0601 05:37:27.124345 140460941842304 utils.py:108] Init efficientnet-b0/blocks_2/conv2d/kernel from ckpt var efficientnet-b0/blocks_2/conv2d/kernel
I0601 05:37:27.124422 140460941842304 utils.py:108] Init efficientnet-b0/blocks_2/tpu_batch_normalization/gamma from ckpt var efficientnet-b0/blocks_2/tpu_batch_normalization/gamma
I0601 05:37:27.124494 140460941842304 utils.py:108] Init efficientnet-b0/blocks_2/tpu_batch_normalization/beta from ckpt var efficientnet-b0/blocks_2/tpu_batch_normalization/beta
I0601 05:37:27.124572 140460941842304 utils.py:108] Init efficientnet-b0/blocks_2/tpu_batch_normalization/moving_mean from ckpt var efficientnet-b0/blocks_2/tpu_batch_normalization/moving_mean
I0601 05:37:27.124654 140460941842304 utils.py:108] Init efficientnet-b0/blocks_2/tpu_batch_normalization/moving_variance from ckpt var efficientnet-b0/blocks_2/tpu_batch_normalization/moving_variance
I0601 05:37:27.124723 140460941842304 utils.py:108] Init efficientnet-b0/blocks_2/depthwise_conv2d/depthwise_kernel from ckpt var efficientnet-b0/blocks_2/depthwise_conv2d/depthwise_kernel
I0601 05:37:27.124789 140460941842304 utils.py:108] Init efficientnet-b0/blocks_2/tpu_batch_normalization_1/gamma from ckpt var efficientnet-b0/blocks_2/tpu_batch_normalization_1/gamma
I0601 05:37:27.124865 140460941842304 utils.py:108] Init efficientnet-b0/blocks_2/tpu_batch_normalization_1/beta from ckpt var efficientnet-b0/blocks_2/tpu_batch_normalization_1/beta
I0601 05:37:27.124931 140460941842304 utils.py:108] Init efficientnet-b0/blocks_2/tpu_batch_normalization_1/moving_mean from ckpt var efficientnet-b0/blocks_2/tpu_batch_normalization_1/moving_mean
I0601 05:37:27.125029 140460941842304 utils.py:108] Init efficientnet-b0/blocks_2/tpu_batch_normalization_1/moving_variance from ckpt var efficientnet-b0/blocks_2/tpu_batch_normalization_1/moving_variance
I0601 05:37:27.125109 140460941842304 utils.py:108] Init efficientnet-b0/blocks_2/se/conv2d/kernel from ckpt var efficientnet-b0/blocks_2/se/conv2d/kernel
I0601 05:37:27.125190 140460941842304 utils.py:108] Init efficientnet-b0/blocks_2/se/conv2d/bias from ckpt var efficientnet-b0/blocks_2/se/conv2d/bias
I0601 05:37:27.125257 140460941842304 utils.py:108] Init efficientnet-b0/blocks_2/se/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_2/se/conv2d_1/kernel
I0601 05:37:27.125338 140460941842304 utils.py:108] Init efficientnet-b0/blocks_2/se/conv2d_1/bias from ckpt var efficientnet-b0/blocks_2/se/conv2d_1/bias
I0601 05:37:27.125397 140460941842304 utils.py:108] Init efficientnet-b0/blocks_2/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_2/conv2d_1/kernel
I0601 05:37:27.125455 140460941842304 utils.py:108] Init efficientnet-b0/blocks_2/tpu_batch_normalization_2/gamma from ckpt var efficientnet-b0/blocks_2/tpu_batch_normalization_2/gamma
I0601 05:37:27.125514 140460941842304 utils.py:108] Init efficientnet-b0/blocks_2/tpu_batch_normalization_2/beta from ckpt var efficientnet-b0/blocks_2/tpu_batch_normalization_2/beta
I0601 05:37:27.125575 140460941842304 utils.py:108] Init efficientnet-b0/blocks_2/tpu_batch_normalization_2/moving_mean from ckpt var efficientnet-b0/blocks_2/tpu_batch_normalization_2/moving_mean
I0601 05:37:27.125635 140460941842304 utils.py:108] Init efficientnet-b0/blocks_2/tpu_batch_normalization_2/moving_variance from ckpt var efficientnet-b0/blocks_2/tpu_batch_normalization_2/moving_variance
I0601 05:37:27.125697 140460941842304 utils.py:108] Init efficientnet-b0/blocks_3/conv2d/kernel from ckpt var efficientnet-b0/blocks_3/conv2d/kernel
I0601 05:37:27.125760 140460941842304 utils.py:108] Init efficientnet-b0/blocks_3/tpu_batch_normalization/gamma from ckpt var efficientnet-b0/blocks_3/tpu_batch_normalization/gamma
I0601 05:37:27.125820 140460941842304 utils.py:108] Init efficientnet-b0/blocks_3/tpu_batch_normalization/beta from ckpt var efficientnet-b0/blocks_3/tpu_batch_normalization/beta
I0601 05:37:27.125884 140460941842304 utils.py:108] Init efficientnet-b0/blocks_3/tpu_batch_normalization/moving_mean from ckpt var efficientnet-b0/blocks_3/tpu_batch_normalization/moving_mean
I0601 05:37:27.125946 140460941842304 utils.py:108] Init efficientnet-b0/blocks_3/tpu_batch_normalization/moving_variance from ckpt var efficientnet-b0/blocks_3/tpu_batch_normalization/moving_variance
I0601 05:37:27.126019 140460941842304 utils.py:108] Init efficientnet-b0/blocks_3/depthwise_conv2d/depthwise_kernel from ckpt var efficientnet-b0/blocks_3/depthwise_conv2d/depthwise_kernel
I0601 05:37:27.126083 140460941842304 utils.py:108] Init efficientnet-b0/blocks_3/tpu_batch_normalization_1/gamma from ckpt var efficientnet-b0/blocks_3/tpu_batch_normalization_1/gamma
I0601 05:37:27.126143 140460941842304 utils.py:108] Init efficientnet-b0/blocks_3/tpu_batch_normalization_1/beta from ckpt var efficientnet-b0/blocks_3/tpu_batch_normalization_1/beta
I0601 05:37:27.126204 140460941842304 utils.py:108] Init efficientnet-b0/blocks_3/tpu_batch_normalization_1/moving_mean from ckpt var efficientnet-b0/blocks_3/tpu_batch_normalization_1/moving_mean
I0601 05:37:27.126283 140460941842304 utils.py:108] Init efficientnet-b0/blocks_3/tpu_batch_normalization_1/moving_variance from ckpt var efficientnet-b0/blocks_3/tpu_batch_normalization_1/moving_variance
I0601 05:37:27.126359 140460941842304 utils.py:108] Init efficientnet-b0/blocks_3/se/conv2d/kernel from ckpt var efficientnet-b0/blocks_3/se/conv2d/kernel
I0601 05:37:27.126425 140460941842304 utils.py:108] Init efficientnet-b0/blocks_3/se/conv2d/bias from ckpt var efficientnet-b0/blocks_3/se/conv2d/bias
I0601 05:37:27.126497 140460941842304 utils.py:108] Init efficientnet-b0/blocks_3/se/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_3/se/conv2d_1/kernel
I0601 05:37:27.126557 140460941842304 utils.py:108] Init efficientnet-b0/blocks_3/se/conv2d_1/bias from ckpt var efficientnet-b0/blocks_3/se/conv2d_1/bias
I0601 05:37:27.126614 140460941842304 utils.py:108] Init efficientnet-b0/blocks_3/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_3/conv2d_1/kernel
I0601 05:37:27.126672 140460941842304 utils.py:108] Init efficientnet-b0/blocks_3/tpu_batch_normalization_2/gamma from ckpt var efficientnet-b0/blocks_3/tpu_batch_normalization_2/gamma
I0601 05:37:27.126732 140460941842304 utils.py:108] Init efficientnet-b0/blocks_3/tpu_batch_normalization_2/beta from ckpt var efficientnet-b0/blocks_3/tpu_batch_normalization_2/beta
I0601 05:37:27.126789 140460941842304 utils.py:108] Init efficientnet-b0/blocks_3/tpu_batch_normalization_2/moving_mean from ckpt var efficientnet-b0/blocks_3/tpu_batch_normalization_2/moving_mean
I0601 05:37:27.126847 140460941842304 utils.py:108] Init efficientnet-b0/blocks_3/tpu_batch_normalization_2/moving_variance from ckpt var efficientnet-b0/blocks_3/tpu_batch_normalization_2/moving_variance
I0601 05:37:27.126905 140460941842304 utils.py:108] Init efficientnet-b0/blocks_4/conv2d/kernel from ckpt var efficientnet-b0/blocks_4/conv2d/kernel
I0601 05:37:27.126962 140460941842304 utils.py:108] Init efficientnet-b0/blocks_4/tpu_batch_normalization/gamma from ckpt var efficientnet-b0/blocks_4/tpu_batch_normalization/gamma
I0601 05:37:27.127028 140460941842304 utils.py:108] Init efficientnet-b0/blocks_4/tpu_batch_normalization/beta from ckpt var efficientnet-b0/blocks_4/tpu_batch_normalization/beta
I0601 05:37:27.127087 140460941842304 utils.py:108] Init efficientnet-b0/blocks_4/tpu_batch_normalization/moving_mean from ckpt var efficientnet-b0/blocks_4/tpu_batch_normalization/moving_mean
I0601 05:37:27.127145 140460941842304 utils.py:108] Init efficientnet-b0/blocks_4/tpu_batch_normalization/moving_variance from ckpt var efficientnet-b0/blocks_4/tpu_batch_normalization/moving_variance
I0601 05:37:27.127201 140460941842304 utils.py:108] Init efficientnet-b0/blocks_4/depthwise_conv2d/depthwise_kernel from ckpt var efficientnet-b0/blocks_4/depthwise_conv2d/depthwise_kernel
I0601 05:37:27.127257 140460941842304 utils.py:108] Init efficientnet-b0/blocks_4/tpu_batch_normalization_1/gamma from ckpt var efficientnet-b0/blocks_4/tpu_batch_normalization_1/gamma
I0601 05:37:27.127334 140460941842304 utils.py:108] Init efficientnet-b0/blocks_4/tpu_batch_normalization_1/beta from ckpt var efficientnet-b0/blocks_4/tpu_batch_normalization_1/beta
I0601 05:37:27.127394 140460941842304 utils.py:108] Init efficientnet-b0/blocks_4/tpu_batch_normalization_1/moving_mean from ckpt var efficientnet-b0/blocks_4/tpu_batch_normalization_1/moving_mean
I0601 05:37:27.127453 140460941842304 utils.py:108] Init efficientnet-b0/blocks_4/tpu_batch_normalization_1/moving_variance from ckpt var efficientnet-b0/blocks_4/tpu_batch_normalization_1/moving_variance
I0601 05:37:27.127517 140460941842304 utils.py:108] Init efficientnet-b0/blocks_4/se/conv2d/kernel from ckpt var efficientnet-b0/blocks_4/se/conv2d/kernel
I0601 05:37:27.127576 140460941842304 utils.py:108] Init efficientnet-b0/blocks_4/se/conv2d/bias from ckpt var efficientnet-b0/blocks_4/se/conv2d/bias
I0601 05:37:27.127633 140460941842304 utils.py:108] Init efficientnet-b0/blocks_4/se/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_4/se/conv2d_1/kernel
I0601 05:37:27.127741 140460941842304 utils.py:108] Init efficientnet-b0/blocks_4/se/conv2d_1/bias from ckpt var efficientnet-b0/blocks_4/se/conv2d_1/bias
I0601 05:37:27.127798 140460941842304 utils.py:108] Init efficientnet-b0/blocks_4/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_4/conv2d_1/kernel
I0601 05:37:27.127856 140460941842304 utils.py:108] Init efficientnet-b0/blocks_4/tpu_batch_normalization_2/gamma from ckpt var efficientnet-b0/blocks_4/tpu_batch_normalization_2/gamma
I0601 05:37:27.127914 140460941842304 utils.py:108] Init efficientnet-b0/blocks_4/tpu_batch_normalization_2/beta from ckpt var efficientnet-b0/blocks_4/tpu_batch_normalization_2/beta
I0601 05:37:27.127989 140460941842304 utils.py:108] Init efficientnet-b0/blocks_4/tpu_batch_normalization_2/moving_mean from ckpt var efficientnet-b0/blocks_4/tpu_batch_normalization_2/moving_mean
I0601 05:37:27.128060 140460941842304 utils.py:108] Init efficientnet-b0/blocks_4/tpu_batch_normalization_2/moving_variance from ckpt var efficientnet-b0/blocks_4/tpu_batch_normalization_2/moving_variance
I0601 05:37:27.128121 140460941842304 utils.py:108] Init efficientnet-b0/blocks_5/conv2d/kernel from ckpt var efficientnet-b0/blocks_5/conv2d/kernel
I0601 05:37:27.128182 140460941842304 utils.py:108] Init efficientnet-b0/blocks_5/tpu_batch_normalization/gamma from ckpt var efficientnet-b0/blocks_5/tpu_batch_normalization/gamma
I0601 05:37:27.207009 140460941842304 utils.py:108] Init efficientnet-b0/blocks_5/tpu_batch_normalization/beta from ckpt var efficientnet-b0/blocks_5/tpu_batch_normalization/beta
I0601 05:37:27.207161 140460941842304 utils.py:108] Init efficientnet-b0/blocks_5/tpu_batch_normalization/moving_mean from ckpt var efficientnet-b0/blocks_5/tpu_batch_normalization/moving_mean
I0601 05:37:27.207300 140460941842304 utils.py:108] Init efficientnet-b0/blocks_5/tpu_batch_normalization/moving_variance from ckpt var efficientnet-b0/blocks_5/tpu_batch_normalization/moving_variance
I0601 05:37:27.207408 140460941842304 utils.py:108] Init efficientnet-b0/blocks_5/depthwise_conv2d/depthwise_kernel from ckpt var efficientnet-b0/blocks_5/depthwise_conv2d/depthwise_kernel
I0601 05:37:27.207518 140460941842304 utils.py:108] Init efficientnet-b0/blocks_5/tpu_batch_normalization_1/gamma from ckpt var efficientnet-b0/blocks_5/tpu_batch_normalization_1/gamma
I0601 05:37:27.207614 140460941842304 utils.py:108] Init efficientnet-b0/blocks_5/tpu_batch_normalization_1/beta from ckpt var efficientnet-b0/blocks_5/tpu_batch_normalization_1/beta
I0601 05:37:27.207706 140460941842304 utils.py:108] Init efficientnet-b0/blocks_5/tpu_batch_normalization_1/moving_mean from ckpt var efficientnet-b0/blocks_5/tpu_batch_normalization_1/moving_mean
I0601 05:37:27.207796 140460941842304 utils.py:108] Init efficientnet-b0/blocks_5/tpu_batch_normalization_1/moving_variance from ckpt var efficientnet-b0/blocks_5/tpu_batch_normalization_1/moving_variance
I0601 05:37:27.207886 140460941842304 utils.py:108] Init efficientnet-b0/blocks_5/se/conv2d/kernel from ckpt var efficientnet-b0/blocks_5/se/conv2d/kernel
I0601 05:37:27.207974 140460941842304 utils.py:108] Init efficientnet-b0/blocks_5/se/conv2d/bias from ckpt var efficientnet-b0/blocks_5/se/conv2d/bias
I0601 05:37:27.208061 140460941842304 utils.py:108] Init efficientnet-b0/blocks_5/se/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_5/se/conv2d_1/kernel
I0601 05:37:27.208147 140460941842304 utils.py:108] Init efficientnet-b0/blocks_5/se/conv2d_1/bias from ckpt var efficientnet-b0/blocks_5/se/conv2d_1/bias
I0601 05:37:27.208252 140460941842304 utils.py:108] Init efficientnet-b0/blocks_5/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_5/conv2d_1/kernel
I0601 05:37:27.208363 140460941842304 utils.py:108] Init efficientnet-b0/blocks_5/tpu_batch_normalization_2/gamma from ckpt var efficientnet-b0/blocks_5/tpu_batch_normalization_2/gamma
I0601 05:37:27.208455 140460941842304 utils.py:108] Init efficientnet-b0/blocks_5/tpu_batch_normalization_2/beta from ckpt var efficientnet-b0/blocks_5/tpu_batch_normalization_2/beta
I0601 05:37:27.208564 140460941842304 utils.py:108] Init efficientnet-b0/blocks_5/tpu_batch_normalization_2/moving_mean from ckpt var efficientnet-b0/blocks_5/tpu_batch_normalization_2/moving_mean
I0601 05:37:27.208651 140460941842304 utils.py:108] Init efficientnet-b0/blocks_5/tpu_batch_normalization_2/moving_variance from ckpt var efficientnet-b0/blocks_5/tpu_batch_normalization_2/moving_variance
I0601 05:37:27.208736 140460941842304 utils.py:108] Init efficientnet-b0/blocks_6/conv2d/kernel from ckpt var efficientnet-b0/blocks_6/conv2d/kernel
I0601 05:37:27.208820 140460941842304 utils.py:108] Init efficientnet-b0/blocks_6/tpu_batch_normalization/gamma from ckpt var efficientnet-b0/blocks_6/tpu_batch_normalization/gamma
I0601 05:37:27.208904 140460941842304 utils.py:108] Init efficientnet-b0/blocks_6/tpu_batch_normalization/beta from ckpt var efficientnet-b0/blocks_6/tpu_batch_normalization/beta
I0601 05:37:27.208990 140460941842304 utils.py:108] Init efficientnet-b0/blocks_6/tpu_batch_normalization/moving_mean from ckpt var efficientnet-b0/blocks_6/tpu_batch_normalization/moving_mean
I0601 05:37:27.209077 140460941842304 utils.py:108] Init efficientnet-b0/blocks_6/tpu_batch_normalization/moving_variance from ckpt var efficientnet-b0/blocks_6/tpu_batch_normalization/moving_variance
I0601 05:37:27.209163 140460941842304 utils.py:108] Init efficientnet-b0/blocks_6/depthwise_conv2d/depthwise_kernel from ckpt var efficientnet-b0/blocks_6/depthwise_conv2d/depthwise_kernel
I0601 05:37:27.209248 140460941842304 utils.py:108] Init efficientnet-b0/blocks_6/tpu_batch_normalization_1/gamma from ckpt var efficientnet-b0/blocks_6/tpu_batch_normalization_1/gamma
I0601 05:37:27.209354 140460941842304 utils.py:108] Init efficientnet-b0/blocks_6/tpu_batch_normalization_1/beta from ckpt var efficientnet-b0/blocks_6/tpu_batch_normalization_1/beta
I0601 05:37:27.209444 140460941842304 utils.py:108] Init efficientnet-b0/blocks_6/tpu_batch_normalization_1/moving_mean from ckpt var efficientnet-b0/blocks_6/tpu_batch_normalization_1/moving_mean
I0601 05:37:27.209539 140460941842304 utils.py:108] Init efficientnet-b0/blocks_6/tpu_batch_normalization_1/moving_variance from ckpt var efficientnet-b0/blocks_6/tpu_batch_normalization_1/moving_variance
I0601 05:37:27.209629 140460941842304 utils.py:108] Init efficientnet-b0/blocks_6/se/conv2d/kernel from ckpt var efficientnet-b0/blocks_6/se/conv2d/kernel
I0601 05:37:27.209716 140460941842304 utils.py:108] Init efficientnet-b0/blocks_6/se/conv2d/bias from ckpt var efficientnet-b0/blocks_6/se/conv2d/bias
I0601 05:37:27.209800 140460941842304 utils.py:108] Init efficientnet-b0/blocks_6/se/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_6/se/conv2d_1/kernel
I0601 05:37:27.209878 140460941842304 utils.py:108] Init efficientnet-b0/blocks_6/se/conv2d_1/bias from ckpt var efficientnet-b0/blocks_6/se/conv2d_1/bias
I0601 05:37:27.209957 140460941842304 utils.py:108] Init efficientnet-b0/blocks_6/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_6/conv2d_1/kernel
I0601 05:37:27.210039 140460941842304 utils.py:108] Init efficientnet-b0/blocks_6/tpu_batch_normalization_2/gamma from ckpt var efficientnet-b0/blocks_6/tpu_batch_normalization_2/gamma
I0601 05:37:27.210120 140460941842304 utils.py:108] Init efficientnet-b0/blocks_6/tpu_batch_normalization_2/beta from ckpt var efficientnet-b0/blocks_6/tpu_batch_normalization_2/beta
I0601 05:37:27.210199 140460941842304 utils.py:108] Init efficientnet-b0/blocks_6/tpu_batch_normalization_2/moving_mean from ckpt var efficientnet-b0/blocks_6/tpu_batch_normalization_2/moving_mean
I0601 05:37:27.210291 140460941842304 utils.py:108] Init efficientnet-b0/blocks_6/tpu_batch_normalization_2/moving_variance from ckpt var efficientnet-b0/blocks_6/tpu_batch_normalization_2/moving_variance
I0601 05:37:27.210397 140460941842304 utils.py:108] Init efficientnet-b0/blocks_7/conv2d/kernel from ckpt var efficientnet-b0/blocks_7/conv2d/kernel
I0601 05:37:27.210498 140460941842304 utils.py:108] Init efficientnet-b0/blocks_7/tpu_batch_normalization/gamma from ckpt var efficientnet-b0/blocks_7/tpu_batch_normalization/gamma
I0601 05:37:27.210580 140460941842304 utils.py:108] Init efficientnet-b0/blocks_7/tpu_batch_normalization/beta from ckpt var efficientnet-b0/blocks_7/tpu_batch_normalization/beta
I0601 05:37:27.210665 140460941842304 utils.py:108] Init efficientnet-b0/blocks_7/tpu_batch_normalization/moving_mean from ckpt var efficientnet-b0/blocks_7/tpu_batch_normalization/moving_mean
I0601 05:37:27.210749 140460941842304 utils.py:108] Init efficientnet-b0/blocks_7/tpu_batch_normalization/moving_variance from ckpt var efficientnet-b0/blocks_7/tpu_batch_normalization/moving_variance
I0601 05:37:27.210835 140460941842304 utils.py:108] Init efficientnet-b0/blocks_7/depthwise_conv2d/depthwise_kernel from ckpt var efficientnet-b0/blocks_7/depthwise_conv2d/depthwise_kernel
I0601 05:37:27.210939 140460941842304 utils.py:108] Init efficientnet-b0/blocks_7/tpu_batch_normalization_1/gamma from ckpt var efficientnet-b0/blocks_7/tpu_batch_normalization_1/gamma
I0601 05:37:27.211031 140460941842304 utils.py:108] Init efficientnet-b0/blocks_7/tpu_batch_normalization_1/beta from ckpt var efficientnet-b0/blocks_7/tpu_batch_normalization_1/beta
I0601 05:37:27.211124 140460941842304 utils.py:108] Init efficientnet-b0/blocks_7/tpu_batch_normalization_1/moving_mean from ckpt var efficientnet-b0/blocks_7/tpu_batch_normalization_1/moving_mean
I0601 05:37:27.211214 140460941842304 utils.py:108] Init efficientnet-b0/blocks_7/tpu_batch_normalization_1/moving_variance from ckpt var efficientnet-b0/blocks_7/tpu_batch_normalization_1/moving_variance
I0601 05:37:27.211322 140460941842304 utils.py:108] Init efficientnet-b0/blocks_7/se/conv2d/kernel from ckpt var efficientnet-b0/blocks_7/se/conv2d/kernel
I0601 05:37:27.211417 140460941842304 utils.py:108] Init efficientnet-b0/blocks_7/se/conv2d/bias from ckpt var efficientnet-b0/blocks_7/se/conv2d/bias
I0601 05:37:27.211518 140460941842304 utils.py:108] Init efficientnet-b0/blocks_7/se/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_7/se/conv2d_1/kernel
I0601 05:37:27.211609 140460941842304 utils.py:108] Init efficientnet-b0/blocks_7/se/conv2d_1/bias from ckpt var efficientnet-b0/blocks_7/se/conv2d_1/bias
I0601 05:37:27.211694 140460941842304 utils.py:108] Init efficientnet-b0/blocks_7/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_7/conv2d_1/kernel
I0601 05:37:27.211786 140460941842304 utils.py:108] Init efficientnet-b0/blocks_7/tpu_batch_normalization_2/gamma from ckpt var efficientnet-b0/blocks_7/tpu_batch_normalization_2/gamma
I0601 05:37:27.211865 140460941842304 utils.py:108] Init efficientnet-b0/blocks_7/tpu_batch_normalization_2/beta from ckpt var efficientnet-b0/blocks_7/tpu_batch_normalization_2/beta
I0601 05:37:27.211946 140460941842304 utils.py:108] Init efficientnet-b0/blocks_7/tpu_batch_normalization_2/moving_mean from ckpt var efficientnet-b0/blocks_7/tpu_batch_normalization_2/moving_mean
I0601 05:37:27.212031 140460941842304 utils.py:108] Init efficientnet-b0/blocks_7/tpu_batch_normalization_2/moving_variance from ckpt var efficientnet-b0/blocks_7/tpu_batch_normalization_2/moving_variance
I0601 05:37:27.212116 140460941842304 utils.py:108] Init efficientnet-b0/blocks_8/conv2d/kernel from ckpt var efficientnet-b0/blocks_8/conv2d/kernel
I0601 05:37:27.212202 140460941842304 utils.py:108] Init efficientnet-b0/blocks_8/tpu_batch_normalization/gamma from ckpt var efficientnet-b0/blocks_8/tpu_batch_normalization/gamma
I0601 05:37:27.212303 140460941842304 utils.py:108] Init efficientnet-b0/blocks_8/tpu_batch_normalization/beta from ckpt var efficientnet-b0/blocks_8/tpu_batch_normalization/beta
I0601 05:37:27.212392 140460941842304 utils.py:108] Init efficientnet-b0/blocks_8/tpu_batch_normalization/moving_mean from ckpt var efficientnet-b0/blocks_8/tpu_batch_normalization/moving_mean
I0601 05:37:27.212473 140460941842304 utils.py:108] Init efficientnet-b0/blocks_8/tpu_batch_normalization/moving_variance from ckpt var efficientnet-b0/blocks_8/tpu_batch_normalization/moving_variance
I0601 05:37:27.212568 140460941842304 utils.py:108] Init efficientnet-b0/blocks_8/depthwise_conv2d/depthwise_kernel from ckpt var efficientnet-b0/blocks_8/depthwise_conv2d/depthwise_kernel
I0601 05:37:27.212655 140460941842304 utils.py:108] Init efficientnet-b0/blocks_8/tpu_batch_normalization_1/gamma from ckpt var efficientnet-b0/blocks_8/tpu_batch_normalization_1/gamma
I0601 05:37:27.212738 140460941842304 utils.py:108] Init efficientnet-b0/blocks_8/tpu_batch_normalization_1/beta from ckpt var efficientnet-b0/blocks_8/tpu_batch_normalization_1/beta
I0601 05:37:27.212823 140460941842304 utils.py:108] Init efficientnet-b0/blocks_8/tpu_batch_normalization_1/moving_mean from ckpt var efficientnet-b0/blocks_8/tpu_batch_normalization_1/moving_mean
I0601 05:37:27.212906 140460941842304 utils.py:108] Init efficientnet-b0/blocks_8/tpu_batch_normalization_1/moving_variance from ckpt var efficientnet-b0/blocks_8/tpu_batch_normalization_1/moving_variance
I0601 05:37:27.212990 140460941842304 utils.py:108] Init efficientnet-b0/blocks_8/se/conv2d/kernel from ckpt var efficientnet-b0/blocks_8/se/conv2d/kernel
I0601 05:37:27.213074 140460941842304 utils.py:108] Init efficientnet-b0/blocks_8/se/conv2d/bias from ckpt var efficientnet-b0/blocks_8/se/conv2d/bias
I0601 05:37:27.213157 140460941842304 utils.py:108] Init efficientnet-b0/blocks_8/se/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_8/se/conv2d_1/kernel
I0601 05:37:27.213241 140460941842304 utils.py:108] Init efficientnet-b0/blocks_8/se/conv2d_1/bias from ckpt var efficientnet-b0/blocks_8/se/conv2d_1/bias
I0601 05:37:27.213342 140460941842304 utils.py:108] Init efficientnet-b0/blocks_8/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_8/conv2d_1/kernel
I0601 05:37:27.213423 140460941842304 utils.py:108] Init efficientnet-b0/blocks_8/tpu_batch_normalization_2/gamma from ckpt var efficientnet-b0/blocks_8/tpu_batch_normalization_2/gamma
I0601 05:37:27.213515 140460941842304 utils.py:108] Init efficientnet-b0/blocks_8/tpu_batch_normalization_2/beta from ckpt var efficientnet-b0/blocks_8/tpu_batch_normalization_2/beta
I0601 05:37:27.213601 140460941842304 utils.py:108] Init efficientnet-b0/blocks_8/tpu_batch_normalization_2/moving_mean from ckpt var efficientnet-b0/blocks_8/tpu_batch_normalization_2/moving_mean
I0601 05:37:27.213684 140460941842304 utils.py:108] Init efficientnet-b0/blocks_8/tpu_batch_normalization_2/moving_variance from ckpt var efficientnet-b0/blocks_8/tpu_batch_normalization_2/moving_variance
I0601 05:37:27.213773 140460941842304 utils.py:108] Init efficientnet-b0/blocks_9/conv2d/kernel from ckpt var efficientnet-b0/blocks_9/conv2d/kernel
I0601 05:37:27.213858 140460941842304 utils.py:108] Init efficientnet-b0/blocks_9/tpu_batch_normalization/gamma from ckpt var efficientnet-b0/blocks_9/tpu_batch_normalization/gamma
I0601 05:37:27.213962 140460941842304 utils.py:108] Init efficientnet-b0/blocks_9/tpu_batch_normalization/beta from ckpt var efficientnet-b0/blocks_9/tpu_batch_normalization/beta
I0601 05:37:27.214052 140460941842304 utils.py:108] Init efficientnet-b0/blocks_9/tpu_batch_normalization/moving_mean from ckpt var efficientnet-b0/blocks_9/tpu_batch_normalization/moving_mean
I0601 05:37:27.214140 140460941842304 utils.py:108] Init efficientnet-b0/blocks_9/tpu_batch_normalization/moving_variance from ckpt var efficientnet-b0/blocks_9/tpu_batch_normalization/moving_variance
I0601 05:37:27.214229 140460941842304 utils.py:108] Init efficientnet-b0/blocks_9/depthwise_conv2d/depthwise_kernel from ckpt var efficientnet-b0/blocks_9/depthwise_conv2d/depthwise_kernel
I0601 05:37:27.214336 140460941842304 utils.py:108] Init efficientnet-b0/blocks_9/tpu_batch_normalization_1/gamma from ckpt var efficientnet-b0/blocks_9/tpu_batch_normalization_1/gamma
I0601 05:37:27.214425 140460941842304 utils.py:108] Init efficientnet-b0/blocks_9/tpu_batch_normalization_1/beta from ckpt var efficientnet-b0/blocks_9/tpu_batch_normalization_1/beta
I0601 05:37:27.214522 140460941842304 utils.py:108] Init efficientnet-b0/blocks_9/tpu_batch_normalization_1/moving_mean from ckpt var efficientnet-b0/blocks_9/tpu_batch_normalization_1/moving_mean
I0601 05:37:27.214612 140460941842304 utils.py:108] Init efficientnet-b0/blocks_9/tpu_batch_normalization_1/moving_variance from ckpt var efficientnet-b0/blocks_9/tpu_batch_normalization_1/moving_variance
I0601 05:37:27.214701 140460941842304 utils.py:108] Init efficientnet-b0/blocks_9/se/conv2d/kernel from ckpt var efficientnet-b0/blocks_9/se/conv2d/kernel
I0601 05:37:27.214841 140460941842304 utils.py:108] Init efficientnet-b0/blocks_9/se/conv2d/bias from ckpt var efficientnet-b0/blocks_9/se/conv2d/bias
I0601 05:37:27.214942 140460941842304 utils.py:108] Init efficientnet-b0/blocks_9/se/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_9/se/conv2d_1/kernel
I0601 05:37:27.215043 140460941842304 utils.py:108] Init efficientnet-b0/blocks_9/se/conv2d_1/bias from ckpt var efficientnet-b0/blocks_9/se/conv2d_1/bias
I0601 05:37:27.215140 140460941842304 utils.py:108] Init efficientnet-b0/blocks_9/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_9/conv2d_1/kernel
I0601 05:37:27.215238 140460941842304 utils.py:108] Init efficientnet-b0/blocks_9/tpu_batch_normalization_2/gamma from ckpt var efficientnet-b0/blocks_9/tpu_batch_normalization_2/gamma
I0601 05:37:27.215354 140460941842304 utils.py:108] Init efficientnet-b0/blocks_9/tpu_batch_normalization_2/beta from ckpt var efficientnet-b0/blocks_9/tpu_batch_normalization_2/beta
I0601 05:37:27.215455 140460941842304 utils.py:108] Init efficientnet-b0/blocks_9/tpu_batch_normalization_2/moving_mean from ckpt var efficientnet-b0/blocks_9/tpu_batch_normalization_2/moving_mean
I0601 05:37:27.215564 140460941842304 utils.py:108] Init efficientnet-b0/blocks_9/tpu_batch_normalization_2/moving_variance from ckpt var efficientnet-b0/blocks_9/tpu_batch_normalization_2/moving_variance
I0601 05:37:27.215664 140460941842304 utils.py:108] Init efficientnet-b0/blocks_10/conv2d/kernel from ckpt var efficientnet-b0/blocks_10/conv2d/kernel
I0601 05:37:27.215763 140460941842304 utils.py:108] Init efficientnet-b0/blocks_10/tpu_batch_normalization/gamma from ckpt var efficientnet-b0/blocks_10/tpu_batch_normalization/gamma
I0601 05:37:27.215863 140460941842304 utils.py:108] Init efficientnet-b0/blocks_10/tpu_batch_normalization/beta from ckpt var efficientnet-b0/blocks_10/tpu_batch_normalization/beta
I0601 05:37:27.215981 140460941842304 utils.py:108] Init efficientnet-b0/blocks_10/tpu_batch_normalization/moving_mean from ckpt var efficientnet-b0/blocks_10/tpu_batch_normalization/moving_mean
I0601 05:37:27.216070 140460941842304 utils.py:108] Init efficientnet-b0/blocks_10/tpu_batch_normalization/moving_variance from ckpt var efficientnet-b0/blocks_10/tpu_batch_normalization/moving_variance
I0601 05:37:27.216159 140460941842304 utils.py:108] Init efficientnet-b0/blocks_10/depthwise_conv2d/depthwise_kernel from ckpt var efficientnet-b0/blocks_10/depthwise_conv2d/depthwise_kernel
I0601 05:37:27.216246 140460941842304 utils.py:108] Init efficientnet-b0/blocks_10/tpu_batch_normalization_1/gamma from ckpt var efficientnet-b0/blocks_10/tpu_batch_normalization_1/gamma
I0601 05:37:27.216351 140460941842304 utils.py:108] Init efficientnet-b0/blocks_10/tpu_batch_normalization_1/beta from ckpt var efficientnet-b0/blocks_10/tpu_batch_normalization_1/beta
I0601 05:37:27.216440 140460941842304 utils.py:108] Init efficientnet-b0/blocks_10/tpu_batch_normalization_1/moving_mean from ckpt var efficientnet-b0/blocks_10/tpu_batch_normalization_1/moving_mean
I0601 05:37:27.216543 140460941842304 utils.py:108] Init efficientnet-b0/blocks_10/tpu_batch_normalization_1/moving_variance from ckpt var efficientnet-b0/blocks_10/tpu_batch_normalization_1/moving_variance
I0601 05:37:27.216635 140460941842304 utils.py:108] Init efficientnet-b0/blocks_10/se/conv2d/kernel from ckpt var efficientnet-b0/blocks_10/se/conv2d/kernel
I0601 05:37:27.216723 140460941842304 utils.py:108] Init efficientnet-b0/blocks_10/se/conv2d/bias from ckpt var efficientnet-b0/blocks_10/se/conv2d/bias
I0601 05:37:27.216810 140460941842304 utils.py:108] Init efficientnet-b0/blocks_10/se/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_10/se/conv2d_1/kernel
I0601 05:37:27.216899 140460941842304 utils.py:108] Init efficientnet-b0/blocks_10/se/conv2d_1/bias from ckpt var efficientnet-b0/blocks_10/se/conv2d_1/bias
I0601 05:37:27.216988 140460941842304 utils.py:108] Init efficientnet-b0/blocks_10/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_10/conv2d_1/kernel
I0601 05:37:27.217077 140460941842304 utils.py:108] Init efficientnet-b0/blocks_10/tpu_batch_normalization_2/gamma from ckpt var efficientnet-b0/blocks_10/tpu_batch_normalization_2/gamma
I0601 05:37:27.217164 140460941842304 utils.py:108] Init efficientnet-b0/blocks_10/tpu_batch_normalization_2/beta from ckpt var efficientnet-b0/blocks_10/tpu_batch_normalization_2/beta
I0601 05:37:27.217254 140460941842304 utils.py:108] Init efficientnet-b0/blocks_10/tpu_batch_normalization_2/moving_mean from ckpt var efficientnet-b0/blocks_10/tpu_batch_normalization_2/moving_mean
I0601 05:37:27.217362 140460941842304 utils.py:108] Init efficientnet-b0/blocks_10/tpu_batch_normalization_2/moving_variance from ckpt var efficientnet-b0/blocks_10/tpu_batch_normalization_2/moving_variance
I0601 05:37:27.217463 140460941842304 utils.py:108] Init efficientnet-b0/blocks_11/conv2d/kernel from ckpt var efficientnet-b0/blocks_11/conv2d/kernel
I0601 05:37:27.217554 140460941842304 utils.py:108] Init efficientnet-b0/blocks_11/tpu_batch_normalization/gamma from ckpt var efficientnet-b0/blocks_11/tpu_batch_normalization/gamma
I0601 05:37:27.217638 140460941842304 utils.py:108] Init efficientnet-b0/blocks_11/tpu_batch_normalization/beta from ckpt var efficientnet-b0/blocks_11/tpu_batch_normalization/beta
I0601 05:37:27.217722 140460941842304 utils.py:108] Init efficientnet-b0/blocks_11/tpu_batch_normalization/moving_mean from ckpt var efficientnet-b0/blocks_11/tpu_batch_normalization/moving_mean
I0601 05:37:27.217807 140460941842304 utils.py:108] Init efficientnet-b0/blocks_11/tpu_batch_normalization/moving_variance from ckpt var efficientnet-b0/blocks_11/tpu_batch_normalization/moving_variance
I0601 05:37:27.217891 140460941842304 utils.py:108] Init efficientnet-b0/blocks_11/depthwise_conv2d/depthwise_kernel from ckpt var efficientnet-b0/blocks_11/depthwise_conv2d/depthwise_kernel
I0601 05:37:27.217975 140460941842304 utils.py:108] Init efficientnet-b0/blocks_11/tpu_batch_normalization_1/gamma from ckpt var efficientnet-b0/blocks_11/tpu_batch_normalization_1/gamma
I0601 05:37:27.218059 140460941842304 utils.py:108] Init efficientnet-b0/blocks_11/tpu_batch_normalization_1/beta from ckpt var efficientnet-b0/blocks_11/tpu_batch_normalization_1/beta
I0601 05:37:27.218143 140460941842304 utils.py:108] Init efficientnet-b0/blocks_11/tpu_batch_normalization_1/moving_mean from ckpt var efficientnet-b0/blocks_11/tpu_batch_normalization_1/moving_mean
I0601 05:37:27.218247 140460941842304 utils.py:108] Init efficientnet-b0/blocks_11/tpu_batch_normalization_1/moving_variance from ckpt var efficientnet-b0/blocks_11/tpu_batch_normalization_1/moving_variance
I0601 05:37:27.218357 140460941842304 utils.py:108] Init efficientnet-b0/blocks_11/se/conv2d/kernel from ckpt var efficientnet-b0/blocks_11/se/conv2d/kernel
I0601 05:37:27.218445 140460941842304 utils.py:108] Init efficientnet-b0/blocks_11/se/conv2d/bias from ckpt var efficientnet-b0/blocks_11/se/conv2d/bias
I0601 05:37:27.218544 140460941842304 utils.py:108] Init efficientnet-b0/blocks_11/se/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_11/se/conv2d_1/kernel
I0601 05:37:27.218634 140460941842304 utils.py:108] Init efficientnet-b0/blocks_11/se/conv2d_1/bias from ckpt var efficientnet-b0/blocks_11/se/conv2d_1/bias
I0601 05:37:27.218723 140460941842304 utils.py:108] Init efficientnet-b0/blocks_11/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_11/conv2d_1/kernel
I0601 05:37:27.218819 140460941842304 utils.py:108] Init efficientnet-b0/blocks_11/tpu_batch_normalization_2/gamma from ckpt var efficientnet-b0/blocks_11/tpu_batch_normalization_2/gamma
I0601 05:37:27.218907 140460941842304 utils.py:108] Init efficientnet-b0/blocks_11/tpu_batch_normalization_2/beta from ckpt var efficientnet-b0/blocks_11/tpu_batch_normalization_2/beta
I0601 05:37:27.218994 140460941842304 utils.py:108] Init efficientnet-b0/blocks_11/tpu_batch_normalization_2/moving_mean from ckpt var efficientnet-b0/blocks_11/tpu_batch_normalization_2/moving_mean
I0601 05:37:27.219083 140460941842304 utils.py:108] Init efficientnet-b0/blocks_11/tpu_batch_normalization_2/moving_variance from ckpt var efficientnet-b0/blocks_11/tpu_batch_normalization_2/moving_variance
I0601 05:37:27.219170 140460941842304 utils.py:108] Init efficientnet-b0/blocks_12/conv2d/kernel from ckpt var efficientnet-b0/blocks_12/conv2d/kernel
I0601 05:37:27.219257 140460941842304 utils.py:108] Init efficientnet-b0/blocks_12/tpu_batch_normalization/gamma from ckpt var efficientnet-b0/blocks_12/tpu_batch_normalization/gamma
I0601 05:37:27.219366 140460941842304 utils.py:108] Init efficientnet-b0/blocks_12/tpu_batch_normalization/beta from ckpt var efficientnet-b0/blocks_12/tpu_batch_normalization/beta
I0601 05:37:27.219455 140460941842304 utils.py:108] Init efficientnet-b0/blocks_12/tpu_batch_normalization/moving_mean from ckpt var efficientnet-b0/blocks_12/tpu_batch_normalization/moving_mean
I0601 05:37:27.219562 140460941842304 utils.py:108] Init efficientnet-b0/blocks_12/tpu_batch_normalization/moving_variance from ckpt var efficientnet-b0/blocks_12/tpu_batch_normalization/moving_variance
I0601 05:37:27.219646 140460941842304 utils.py:108] Init efficientnet-b0/blocks_12/depthwise_conv2d/depthwise_kernel from ckpt var efficientnet-b0/blocks_12/depthwise_conv2d/depthwise_kernel
I0601 05:37:27.219750 140460941842304 utils.py:108] Init efficientnet-b0/blocks_12/tpu_batch_normalization_1/gamma from ckpt var efficientnet-b0/blocks_12/tpu_batch_normalization_1/gamma
I0601 05:37:27.219845 140460941842304 utils.py:108] Init efficientnet-b0/blocks_12/tpu_batch_normalization_1/beta from ckpt var efficientnet-b0/blocks_12/tpu_batch_normalization_1/beta
I0601 05:37:27.219927 140460941842304 utils.py:108] Init efficientnet-b0/blocks_12/tpu_batch_normalization_1/moving_mean from ckpt var efficientnet-b0/blocks_12/tpu_batch_normalization_1/moving_mean
I0601 05:37:27.220011 140460941842304 utils.py:108] Init efficientnet-b0/blocks_12/tpu_batch_normalization_1/moving_variance from ckpt var efficientnet-b0/blocks_12/tpu_batch_normalization_1/moving_variance
I0601 05:37:27.220114 140460941842304 utils.py:108] Init efficientnet-b0/blocks_12/se/conv2d/kernel from ckpt var efficientnet-b0/blocks_12/se/conv2d/kernel
I0601 05:37:27.220211 140460941842304 utils.py:108] Init efficientnet-b0/blocks_12/se/conv2d/bias from ckpt var efficientnet-b0/blocks_12/se/conv2d/bias
I0601 05:37:27.220309 140460941842304 utils.py:108] Init efficientnet-b0/blocks_12/se/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_12/se/conv2d_1/kernel
I0601 05:37:27.220395 140460941842304 utils.py:108] Init efficientnet-b0/blocks_12/se/conv2d_1/bias from ckpt var efficientnet-b0/blocks_12/se/conv2d_1/bias
I0601 05:37:27.220478 140460941842304 utils.py:108] Init efficientnet-b0/blocks_12/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_12/conv2d_1/kernel
I0601 05:37:27.220593 140460941842304 utils.py:108] Init efficientnet-b0/blocks_12/tpu_batch_normalization_2/gamma from ckpt var efficientnet-b0/blocks_12/tpu_batch_normalization_2/gamma
I0601 05:37:27.220682 140460941842304 utils.py:108] Init efficientnet-b0/blocks_12/tpu_batch_normalization_2/beta from ckpt var efficientnet-b0/blocks_12/tpu_batch_normalization_2/beta
I0601 05:37:27.220764 140460941842304 utils.py:108] Init efficientnet-b0/blocks_12/tpu_batch_normalization_2/moving_mean from ckpt var efficientnet-b0/blocks_12/tpu_batch_normalization_2/moving_mean
I0601 05:37:27.220853 140460941842304 utils.py:108] Init efficientnet-b0/blocks_12/tpu_batch_normalization_2/moving_variance from ckpt var efficientnet-b0/blocks_12/tpu_batch_normalization_2/moving_variance
I0601 05:37:27.220942 140460941842304 utils.py:108] Init efficientnet-b0/blocks_13/conv2d/kernel from ckpt var efficientnet-b0/blocks_13/conv2d/kernel
I0601 05:37:27.221038 140460941842304 utils.py:108] Init efficientnet-b0/blocks_13/tpu_batch_normalization/gamma from ckpt var efficientnet-b0/blocks_13/tpu_batch_normalization/gamma
I0601 05:37:27.221127 140460941842304 utils.py:108] Init efficientnet-b0/blocks_13/tpu_batch_normalization/beta from ckpt var efficientnet-b0/blocks_13/tpu_batch_normalization/beta
I0601 05:37:27.221215 140460941842304 utils.py:108] Init efficientnet-b0/blocks_13/tpu_batch_normalization/moving_mean from ckpt var efficientnet-b0/blocks_13/tpu_batch_normalization/moving_mean
I0601 05:37:27.221319 140460941842304 utils.py:108] Init efficientnet-b0/blocks_13/tpu_batch_normalization/moving_variance from ckpt var efficientnet-b0/blocks_13/tpu_batch_normalization/moving_variance
I0601 05:37:27.221410 140460941842304 utils.py:108] Init efficientnet-b0/blocks_13/depthwise_conv2d/depthwise_kernel from ckpt var efficientnet-b0/blocks_13/depthwise_conv2d/depthwise_kernel
I0601 05:37:27.221506 140460941842304 utils.py:108] Init efficientnet-b0/blocks_13/tpu_batch_normalization_1/gamma from ckpt var efficientnet-b0/blocks_13/tpu_batch_normalization_1/gamma
I0601 05:37:27.221597 140460941842304 utils.py:108] Init efficientnet-b0/blocks_13/tpu_batch_normalization_1/beta from ckpt var efficientnet-b0/blocks_13/tpu_batch_normalization_1/beta
I0601 05:37:27.221686 140460941842304 utils.py:108] Init efficientnet-b0/blocks_13/tpu_batch_normalization_1/moving_mean from ckpt var efficientnet-b0/blocks_13/tpu_batch_normalization_1/moving_mean
I0601 05:37:27.221774 140460941842304 utils.py:108] Init efficientnet-b0/blocks_13/tpu_batch_normalization_1/moving_variance from ckpt var efficientnet-b0/blocks_13/tpu_batch_normalization_1/moving_variance
I0601 05:37:27.221896 140460941842304 utils.py:108] Init efficientnet-b0/blocks_13/se/conv2d/kernel from ckpt var efficientnet-b0/blocks_13/se/conv2d/kernel
I0601 05:37:27.221987 140460941842304 utils.py:108] Init efficientnet-b0/blocks_13/se/conv2d/bias from ckpt var efficientnet-b0/blocks_13/se/conv2d/bias
I0601 05:37:27.222090 140460941842304 utils.py:108] Init efficientnet-b0/blocks_13/se/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_13/se/conv2d_1/kernel
I0601 05:37:27.222178 140460941842304 utils.py:108] Init efficientnet-b0/blocks_13/se/conv2d_1/bias from ckpt var efficientnet-b0/blocks_13/se/conv2d_1/bias
I0601 05:37:27.222279 140460941842304 utils.py:108] Init efficientnet-b0/blocks_13/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_13/conv2d_1/kernel
I0601 05:37:27.222373 140460941842304 utils.py:108] Init efficientnet-b0/blocks_13/tpu_batch_normalization_2/gamma from ckpt var efficientnet-b0/blocks_13/tpu_batch_normalization_2/gamma
I0601 05:37:27.222463 140460941842304 utils.py:108] Init efficientnet-b0/blocks_13/tpu_batch_normalization_2/beta from ckpt var efficientnet-b0/blocks_13/tpu_batch_normalization_2/beta
I0601 05:37:27.222568 140460941842304 utils.py:108] Init efficientnet-b0/blocks_13/tpu_batch_normalization_2/moving_mean from ckpt var efficientnet-b0/blocks_13/tpu_batch_normalization_2/moving_mean
I0601 05:37:27.222660 140460941842304 utils.py:108] Init efficientnet-b0/blocks_13/tpu_batch_normalization_2/moving_variance from ckpt var efficientnet-b0/blocks_13/tpu_batch_normalization_2/moving_variance
I0601 05:37:27.222749 140460941842304 utils.py:108] Init efficientnet-b0/blocks_14/conv2d/kernel from ckpt var efficientnet-b0/blocks_14/conv2d/kernel
I0601 05:37:27.222837 140460941842304 utils.py:108] Init efficientnet-b0/blocks_14/tpu_batch_normalization/gamma from ckpt var efficientnet-b0/blocks_14/tpu_batch_normalization/gamma
I0601 05:37:27.222927 140460941842304 utils.py:108] Init efficientnet-b0/blocks_14/tpu_batch_normalization/beta from ckpt var efficientnet-b0/blocks_14/tpu_batch_normalization/beta
I0601 05:37:27.223024 140460941842304 utils.py:108] Init efficientnet-b0/blocks_14/tpu_batch_normalization/moving_mean from ckpt var efficientnet-b0/blocks_14/tpu_batch_normalization/moving_mean
I0601 05:37:27.223109 140460941842304 utils.py:108] Init efficientnet-b0/blocks_14/tpu_batch_normalization/moving_variance from ckpt var efficientnet-b0/blocks_14/tpu_batch_normalization/moving_variance
I0601 05:37:27.223192 140460941842304 utils.py:108] Init efficientnet-b0/blocks_14/depthwise_conv2d/depthwise_kernel from ckpt var efficientnet-b0/blocks_14/depthwise_conv2d/depthwise_kernel
I0601 05:37:27.223289 140460941842304 utils.py:108] Init efficientnet-b0/blocks_14/tpu_batch_normalization_1/gamma from ckpt var efficientnet-b0/blocks_14/tpu_batch_normalization_1/gamma
I0601 05:37:27.223378 140460941842304 utils.py:108] Init efficientnet-b0/blocks_14/tpu_batch_normalization_1/beta from ckpt var efficientnet-b0/blocks_14/tpu_batch_normalization_1/beta
I0601 05:37:27.223461 140460941842304 utils.py:108] Init efficientnet-b0/blocks_14/tpu_batch_normalization_1/moving_mean from ckpt var efficientnet-b0/blocks_14/tpu_batch_normalization_1/moving_mean
I0601 05:37:27.223575 140460941842304 utils.py:108] Init efficientnet-b0/blocks_14/tpu_batch_normalization_1/moving_variance from ckpt var efficientnet-b0/blocks_14/tpu_batch_normalization_1/moving_variance
I0601 05:37:27.223672 140460941842304 utils.py:108] Init efficientnet-b0/blocks_14/se/conv2d/kernel from ckpt var efficientnet-b0/blocks_14/se/conv2d/kernel
I0601 05:37:27.223776 140460941842304 utils.py:108] Init efficientnet-b0/blocks_14/se/conv2d/bias from ckpt var efficientnet-b0/blocks_14/se/conv2d/bias
I0601 05:37:27.223863 140460941842304 utils.py:108] Init efficientnet-b0/blocks_14/se/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_14/se/conv2d_1/kernel
I0601 05:37:27.223951 140460941842304 utils.py:108] Init efficientnet-b0/blocks_14/se/conv2d_1/bias from ckpt var efficientnet-b0/blocks_14/se/conv2d_1/bias
I0601 05:37:27.224049 140460941842304 utils.py:108] Init efficientnet-b0/blocks_14/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_14/conv2d_1/kernel
I0601 05:37:27.224131 140460941842304 utils.py:108] Init efficientnet-b0/blocks_14/tpu_batch_normalization_2/gamma from ckpt var efficientnet-b0/blocks_14/tpu_batch_normalization_2/gamma
I0601 05:37:27.224216 140460941842304 utils.py:108] Init efficientnet-b0/blocks_14/tpu_batch_normalization_2/beta from ckpt var efficientnet-b0/blocks_14/tpu_batch_normalization_2/beta
I0601 05:37:27.224336 140460941842304 utils.py:108] Init efficientnet-b0/blocks_14/tpu_batch_normalization_2/moving_mean from ckpt var efficientnet-b0/blocks_14/tpu_batch_normalization_2/moving_mean
I0601 05:37:27.224423 140460941842304 utils.py:108] Init efficientnet-b0/blocks_14/tpu_batch_normalization_2/moving_variance from ckpt var efficientnet-b0/blocks_14/tpu_batch_normalization_2/moving_variance
I0601 05:37:27.224520 140460941842304 utils.py:108] Init efficientnet-b0/blocks_15/conv2d/kernel from ckpt var efficientnet-b0/blocks_15/conv2d/kernel
I0601 05:37:27.224611 140460941842304 utils.py:108] Init efficientnet-b0/blocks_15/tpu_batch_normalization/gamma from ckpt var efficientnet-b0/blocks_15/tpu_batch_normalization/gamma
I0601 05:37:27.224711 140460941842304 utils.py:108] Init efficientnet-b0/blocks_15/tpu_batch_normalization/beta from ckpt var efficientnet-b0/blocks_15/tpu_batch_normalization/beta
I0601 05:37:27.224797 140460941842304 utils.py:108] Init efficientnet-b0/blocks_15/tpu_batch_normalization/moving_mean from ckpt var efficientnet-b0/blocks_15/tpu_batch_normalization/moving_mean
I0601 05:37:27.224882 140460941842304 utils.py:108] Init efficientnet-b0/blocks_15/tpu_batch_normalization/moving_variance from ckpt var efficientnet-b0/blocks_15/tpu_batch_normalization/moving_variance
I0601 05:37:27.224988 140460941842304 utils.py:108] Init efficientnet-b0/blocks_15/depthwise_conv2d/depthwise_kernel from ckpt var efficientnet-b0/blocks_15/depthwise_conv2d/depthwise_kernel
I0601 05:37:27.225078 140460941842304 utils.py:108] Init efficientnet-b0/blocks_15/tpu_batch_normalization_1/gamma from ckpt var efficientnet-b0/blocks_15/tpu_batch_normalization_1/gamma
I0601 05:37:27.225168 140460941842304 utils.py:108] Init efficientnet-b0/blocks_15/tpu_batch_normalization_1/beta from ckpt var efficientnet-b0/blocks_15/tpu_batch_normalization_1/beta
I0601 05:37:27.225256 140460941842304 utils.py:108] Init efficientnet-b0/blocks_15/tpu_batch_normalization_1/moving_mean from ckpt var efficientnet-b0/blocks_15/tpu_batch_normalization_1/moving_mean
I0601 05:37:27.225366 140460941842304 utils.py:108] Init efficientnet-b0/blocks_15/tpu_batch_normalization_1/moving_variance from ckpt var efficientnet-b0/blocks_15/tpu_batch_normalization_1/moving_variance
I0601 05:37:27.225460 140460941842304 utils.py:108] Init efficientnet-b0/blocks_15/se/conv2d/kernel from ckpt var efficientnet-b0/blocks_15/se/conv2d/kernel
I0601 05:37:27.225552 140460941842304 utils.py:108] Init efficientnet-b0/blocks_15/se/conv2d/bias from ckpt var efficientnet-b0/blocks_15/se/conv2d/bias
I0601 05:37:27.225637 140460941842304 utils.py:108] Init efficientnet-b0/blocks_15/se/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_15/se/conv2d_1/kernel
I0601 05:37:27.225721 140460941842304 utils.py:108] Init efficientnet-b0/blocks_15/se/conv2d_1/bias from ckpt var efficientnet-b0/blocks_15/se/conv2d_1/bias
I0601 05:37:27.225804 140460941842304 utils.py:108] Init efficientnet-b0/blocks_15/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_15/conv2d_1/kernel
I0601 05:37:27.225886 140460941842304 utils.py:108] Init efficientnet-b0/blocks_15/tpu_batch_normalization_2/gamma from ckpt var efficientnet-b0/blocks_15/tpu_batch_normalization_2/gamma
I0601 05:37:27.225969 140460941842304 utils.py:108] Init efficientnet-b0/blocks_15/tpu_batch_normalization_2/beta from ckpt var efficientnet-b0/blocks_15/tpu_batch_normalization_2/beta
I0601 05:37:27.226053 140460941842304 utils.py:108] Init efficientnet-b0/blocks_15/tpu_batch_normalization_2/moving_mean from ckpt var efficientnet-b0/blocks_15/tpu_batch_normalization_2/moving_mean
I0601 05:37:27.226139 140460941842304 utils.py:108] Init efficientnet-b0/blocks_15/tpu_batch_normalization_2/moving_variance from ckpt var efficientnet-b0/blocks_15/tpu_batch_normalization_2/moving_variance
INFO:tensorflow:Done calling model_fn.
I0601 05:37:27.906973 140460941842304 estimator.py:1171] Done calling model_fn.
INFO:tensorflow:Create CheckpointSaverHook.
I0601 05:37:27.908015 140460941842304 basic_session_run_hooks.py:546] Create CheckpointSaverHook.
INFO:tensorflow:Graph was finalized.
I0601 05:37:31.553620 140460941842304 monitored_session.py:246] Graph was finalized.
2020-06-01 05:37:31.559414: I tensorflow/core/platform/profile_utils/cpu_utils.cc:102] CPU Frequency: 2300000000 Hz
2020-06-01 05:37:31.559668: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x1b57100 initialized for platform Host (this does not guarantee that XLA will be used). Devices:
2020-06-01 05:37:31.559704: I tensorflow/compiler/xla/service/service.cc:176]   StreamExecutor device (0): Host, Default Version
2020-06-01 05:37:31.674881: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:37:31.675656: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x1b572c0 initialized for platform CUDA (this does not guarantee that XLA will be used). Devices:
2020-06-01 05:37:31.675693: I tensorflow/compiler/xla/service/service.cc:176]   StreamExecutor device (0): Tesla T4, Compute Capability 7.5
2020-06-01 05:37:31.675925: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:37:31.676551: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1561] Found device 0 with properties: 
pciBusID: 0000:00:04.0 name: Tesla T4 computeCapability: 7.5
coreClock: 1.59GHz coreCount: 40 deviceMemorySize: 14.73GiB deviceMemoryBandwidth: 298.08GiB/s
2020-06-01 05:37:31.676608: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.1
2020-06-01 05:37:31.676661: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcublas.so.10
2020-06-01 05:37:31.676689: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcufft.so.10
2020-06-01 05:37:31.676707: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcurand.so.10
2020-06-01 05:37:31.676728: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusolver.so.10
2020-06-01 05:37:31.676747: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusparse.so.10
2020-06-01 05:37:31.676767: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudnn.so.7
2020-06-01 05:37:31.676876: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:37:31.677540: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:37:31.678067: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1703] Adding visible gpu devices: 0
2020-06-01 05:37:31.678122: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.1
2020-06-01 05:37:32.222065: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1102] Device interconnect StreamExecutor with strength 1 edge matrix:
2020-06-01 05:37:32.222124: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1108]      0 
2020-06-01 05:37:32.222137: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1121] 0:   N 
2020-06-01 05:37:32.222377: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:37:32.223035: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:37:32.223671: W tensorflow/core/common_runtime/gpu/gpu_bfc_allocator.cc:39] Overriding allow_growth setting because the TF_FORCE_GPU_ALLOW_GROWTH environment variable is set. Original config value was 0.
2020-06-01 05:37:32.223744: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1247] Created TensorFlow device (/job:localhost/replica:0/task:0/device:GPU:0 with 13970 MB memory) -> physical GPU (device: 0, name: Tesla T4, pci bus id: 0000:00:04.0, compute capability: 7.5)
INFO:tensorflow:Running local_init_op.
I0601 05:37:35.726311 140460941842304 session_manager.py:505] Running local_init_op.
INFO:tensorflow:Done running local_init_op.
I0601 05:37:35.975380 140460941842304 session_manager.py:508] Done running local_init_op.
INFO:tensorflow:Calling checkpoint listeners before saving checkpoint 0...
I0601 05:37:45.047018 140460941842304 basic_session_run_hooks.py:614] Calling checkpoint listeners before saving checkpoint 0...
INFO:tensorflow:Saving checkpoints for 0 into /tmp/model_dir/efficientdet-d0-scratch/model.ckpt.
I0601 05:37:45.047802 140460941842304 basic_session_run_hooks.py:618] Saving checkpoints for 0 into /tmp/model_dir/efficientdet-d0-scratch/model.ckpt.
INFO:tensorflow:Calling checkpoint listeners after saving checkpoint 0...
I0601 05:37:47.314194 140460941842304 basic_session_run_hooks.py:626] Calling checkpoint listeners after saving checkpoint 0...
2020-06-01 05:37:56.700533: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudnn.so.7
2020-06-01 05:38:01.747040: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcublas.so.10
INFO:tensorflow:global_step/sec: 0.140179
I0601 05:38:10.825751 140460941842304 tpu_estimator.py:2350] global_step/sec: 0.140179
INFO:tensorflow:examples/sec: 0.560716
I0601 05:38:10.827028 140460941842304 tpu_estimator.py:2351] examples/sec: 0.560716
INFO:tensorflow:global_step/sec: 5.58317
I0601 05:38:11.004751 140460941842304 tpu_estimator.py:2350] global_step/sec: 5.58317
INFO:tensorflow:examples/sec: 22.3327
I0601 05:38:11.005145 140460941842304 tpu_estimator.py:2351] examples/sec: 22.3327
INFO:tensorflow:global_step/sec: 5.53731
I0601 05:38:11.185394 140460941842304 tpu_estimator.py:2350] global_step/sec: 5.53731
INFO:tensorflow:examples/sec: 22.1492
I0601 05:38:11.185792 140460941842304 tpu_estimator.py:2351] examples/sec: 22.1492
INFO:tensorflow:global_step/sec: 5.00635
I0601 05:38:11.385094 140460941842304 tpu_estimator.py:2350] global_step/sec: 5.00635
INFO:tensorflow:examples/sec: 20.0254
I0601 05:38:11.385536 140460941842304 tpu_estimator.py:2351] examples/sec: 20.0254
INFO:tensorflow:global_step/sec: 5.12939
I0601 05:38:11.580079 140460941842304 tpu_estimator.py:2350] global_step/sec: 5.12939
INFO:tensorflow:examples/sec: 20.5176
I0601 05:38:11.580518 140460941842304 tpu_estimator.py:2351] examples/sec: 20.5176
INFO:tensorflow:global_step/sec: 4.99735
I0601 05:38:11.780181 140460941842304 tpu_estimator.py:2350] global_step/sec: 4.99735
INFO:tensorflow:examples/sec: 19.9894
I0601 05:38:11.781100 140460941842304 tpu_estimator.py:2351] examples/sec: 19.9894
INFO:tensorflow:global_step/sec: 5.44052
I0601 05:38:11.963963 140460941842304 tpu_estimator.py:2350] global_step/sec: 5.44052
INFO:tensorflow:examples/sec: 21.7621
I0601 05:38:11.964417 140460941842304 tpu_estimator.py:2351] examples/sec: 21.7621
INFO:tensorflow:global_step/sec: 5.10901
I0601 05:38:12.159699 140460941842304 tpu_estimator.py:2350] global_step/sec: 5.10901
INFO:tensorflow:examples/sec: 20.436
I0601 05:38:12.160114 140460941842304 tpu_estimator.py:2351] examples/sec: 20.436
INFO:tensorflow:global_step/sec: 5.72686
I0601 05:38:12.334332 140460941842304 tpu_estimator.py:2350] global_step/sec: 5.72686
INFO:tensorflow:examples/sec: 22.9074
I0601 05:38:12.334732 140460941842304 tpu_estimator.py:2351] examples/sec: 22.9074
INFO:tensorflow:global_step/sec: 5.79841
I0601 05:38:12.506778 140460941842304 tpu_estimator.py:2350] global_step/sec: 5.79841
INFO:tensorflow:examples/sec: 23.1936
I0601 05:38:12.507008 140460941842304 tpu_estimator.py:2351] examples/sec: 23.1936
INFO:tensorflow:global_step/sec: 5.25588
I0601 05:38:12.697033 140460941842304 tpu_estimator.py:2350] global_step/sec: 5.25588
INFO:tensorflow:examples/sec: 21.0235
I0601 05:38:12.697489 140460941842304 tpu_estimator.py:2351] examples/sec: 21.0235
INFO:tensorflow:global_step/sec: 5.23154
I0601 05:38:12.888187 140460941842304 tpu_estimator.py:2350] global_step/sec: 5.23154
INFO:tensorflow:examples/sec: 20.9262
I0601 05:38:12.888642 140460941842304 tpu_estimator.py:2351] examples/sec: 20.9262
INFO:tensorflow:global_step/sec: 6.56178
I0601 05:38:13.040558 140460941842304 tpu_estimator.py:2350] global_step/sec: 6.56178
INFO:tensorflow:examples/sec: 26.2471
I0601 05:38:13.040964 140460941842304 tpu_estimator.py:2351] examples/sec: 26.2471
INFO:tensorflow:Calling checkpoint listeners before saving checkpoint 14...
I0601 05:38:13.041552 140460941842304 basic_session_run_hooks.py:614] Calling checkpoint listeners before saving checkpoint 14...
INFO:tensorflow:Saving checkpoints for 14 into /tmp/model_dir/efficientdet-d0-scratch/model.ckpt.
I0601 05:38:13.041719 140460941842304 basic_session_run_hooks.py:618] Saving checkpoints for 14 into /tmp/model_dir/efficientdet-d0-scratch/model.ckpt.
INFO:tensorflow:Calling checkpoint listeners after saving checkpoint 14...
I0601 05:38:14.448161 140460941842304 basic_session_run_hooks.py:626] Calling checkpoint listeners after saving checkpoint 14...
INFO:tensorflow:Loss for final step: 2.4904294.
I0601 05:38:15.084805 140460941842304 estimator.py:350] Loss for final step: 2.4904294.
INFO:tensorflow:training_loop marked as finished
I0601 05:38:15.085538 140460941842304 error_handling.py:115] training_loop marked as finished
I0601 05:38:15.085692 140460941842304 main.py:388] Starting evaluation cycle, epoch: 0.
INFO:tensorflow:Using config: {'_model_dir': '/tmp/model_dir/efficientdet-d0-scratch', '_tf_random_seed': None, '_save_summary_steps': 100, '_save_checkpoints_steps': None, '_save_checkpoints_secs': 600, '_session_config': allow_soft_placement: true
, '_keep_checkpoint_max': 5, '_keep_checkpoint_every_n_hours': 10000, '_log_step_count_steps': None, '_train_distribute': None, '_device_fn': None, '_protocol': None, '_eval_distribute': None, '_experimental_distribute': None, '_experimental_max_worker_delay_secs': None, '_session_creation_timeout_secs': 7200, '_service': None, '_cluster_spec': ClusterSpec({}), '_task_type': 'worker', '_task_id': 0, '_global_id_in_cluster': 0, '_master': '', '_evaluation_master': '', '_is_chief': True, '_num_ps_replicas': 0, '_num_worker_replicas': 1, '_tpu_config': TPUConfig(iterations_per_loop=100, num_shards=8, num_cores_per_replica=None, per_host_input_for_training=3, tpu_job_name=None, initial_infeed_sleep_secs=None, input_partition_dims=None, eval_training_input_configuration=2, experimental_host_call_every_n_steps=1), '_cluster': None}
I0601 05:38:15.086570 140460941842304 estimator.py:191] Using config: {'_model_dir': '/tmp/model_dir/efficientdet-d0-scratch', '_tf_random_seed': None, '_save_summary_steps': 100, '_save_checkpoints_steps': None, '_save_checkpoints_secs': 600, '_session_config': allow_soft_placement: true
, '_keep_checkpoint_max': 5, '_keep_checkpoint_every_n_hours': 10000, '_log_step_count_steps': None, '_train_distribute': None, '_device_fn': None, '_protocol': None, '_eval_distribute': None, '_experimental_distribute': None, '_experimental_max_worker_delay_secs': None, '_session_creation_timeout_secs': 7200, '_service': None, '_cluster_spec': ClusterSpec({}), '_task_type': 'worker', '_task_id': 0, '_global_id_in_cluster': 0, '_master': '', '_evaluation_master': '', '_is_chief': True, '_num_ps_replicas': 0, '_num_worker_replicas': 1, '_tpu_config': TPUConfig(iterations_per_loop=100, num_shards=8, num_cores_per_replica=None, per_host_input_for_training=3, tpu_job_name=None, initial_infeed_sleep_secs=None, input_partition_dims=None, eval_training_input_configuration=2, experimental_host_call_every_n_steps=1), '_cluster': None}
INFO:tensorflow:_TPUContext: eval_on_tpu True
I0601 05:38:15.087005 140460941842304 tpu_context.py:216] _TPUContext: eval_on_tpu True
WARNING:tensorflow:eval_on_tpu ignored because use_tpu is False.
W0601 05:38:15.087114 140460941842304 tpu_context.py:218] eval_on_tpu ignored because use_tpu is False.
INFO:tensorflow:Calling model_fn.
I0601 05:38:15.609617 140460941842304 estimator.py:1169] Calling model_fn.
INFO:tensorflow:Running eval on CPU/GPU
I0601 05:38:15.609894 140460941842304 tpu_estimator.py:3171] Running eval on CPU/GPU
I0601 05:38:15.610420 140460941842304 utils.py:595] use mixed precision policy name mixed_float16
I0601 05:38:15.611307 140460941842304 efficientdet_arch.py:720] act_type: swish
alpha: 0.25
anchor_scale: 4.0
apply_bn_for_resampling: true
aspect_ratios:
- !!python/tuple
    - 1.0
    - 1.0
- !!python/tuple
    - 1.4
    - 0.7
- !!python/tuple
    - 0.7
    - 1.4
augmix_params: !!python/tuple
- 3
- -1
- 1
autoaugment_policy: null
backbone_ckpt: efficientnet-b0
backbone_config: null
backbone_name: efficientnet-b0
batch_size: 4
box_class_repeats: 3
box_loss_weight: 50.0
ckpt: null
ckpt_var_scope: null
clip_gradients_norm: 10.0
conv_after_downsample: false
conv_bn_act_pattern: false
data_format: channels_last
delta: 0.1
first_lr_drop_epoch: 200.0
fpn_cell_repeats: 3
fpn_config: null
fpn_name: null
fpn_num_filters: 64
fpn_weight_method: null
gamma: 1.5
image_size: !!python/tuple
- 512
- 512
img_summary_steps: null
input_rand_hflip: false
iou_loss_type: null
iou_loss_weight: 1.0
is_training_bn: false
iterations_per_loop: 100
label_id_mapping: null
learning_rate: 0.08
lr_decay_method: cosine
lr_warmup_epoch: 1.0
lr_warmup_init: 0.008
max_instances_per_image: 100
max_level: 7
min_level: 3
mode: train_and_eval
model_dir: /tmp/model_dir/efficientdet-d0-scratch
model_name: efficientdet-d0
momentum: 0.9
moving_average_decay: 0
name: efficientdet-d0
num_classes: 20
num_epochs: 1
num_examples_per_epoch: 56
num_scales: 3
num_shards: 8
optimizer: sgd
poly_lr_power: 0.9
pooling_type: null
precision: mixed_float16
resnet_depth: 50
second_lr_drop_epoch: 250.0
separable_conv: true
skip_crowd_during_training: true
strategy: null
survival_prob: null
target_size: null
testdev_dir: null
train_scale_max: 2.0
train_scale_min: 0.1
use_augmix: false
use_native_resize_op: true
use_tpu: false
val_json_file: null
var_exclude_expr: .*/class-predict/.*
var_freeze_expr: null
weight_decay: 4.0e-05

I0601 05:38:15.615985 140460941842304 efficientnet_builder.py:224] global_params= GlobalParams(batch_norm_momentum=0.99, batch_norm_epsilon=0.001, dropout_rate=0.2, data_format='channels_last', num_classes=1000, width_coefficient=1.0, depth_coefficient=1.0, depth_divisor=8, min_depth=None, survival_prob=0.0, relu_fn=functools.partial(<function activation_fn at 0x7fbf45bc8d90>, act_type='swish'), batch_norm=<class 'utils.BatchNormalization'>, use_se=True, local_pooling=None, condconv_num_experts=None, clip_projection_output=False, blocks_args=['r1_k3_s11_e1_i32_o16_se0.25', 'r2_k3_s22_e6_i16_o24_se0.25', 'r2_k5_s22_e6_i24_o40_se0.25', 'r3_k3_s22_e6_i40_o80_se0.25', 'r3_k5_s11_e6_i80_o112_se0.25', 'r4_k5_s22_e6_i112_o192_se0.25', 'r1_k3_s11_e6_i192_o320_se0.25'], fix_head_stem=None)
I0601 05:38:15.879783 140460941842304 api.py:587] Built stem layers with output shape: (4, 256, 256, 32)
I0601 05:38:15.890318 140460941842304 api.py:587] Block mb_conv_block  input shape: (4, 256, 256, 32)
I0601 05:38:15.922745 140460941842304 api.py:587] DWConv shape: (4, 256, 256, 32)
I0601 05:38:15.959311 140460941842304 api.py:587] Built Squeeze and Excitation with tensor shape: (4, 1, 1, 32)
I0601 05:38:15.990694 140460941842304 api.py:587] Project shape: (4, 256, 256, 16)
I0601 05:38:16.003817 140460941842304 api.py:587] Block mb_conv_block_1  input shape: (4, 256, 256, 16)
I0601 05:38:16.038242 140460941842304 api.py:587] Expand shape: (4, 256, 256, 96)
I0601 05:38:16.073603 140460941842304 api.py:587] DWConv shape: (4, 128, 128, 96)
I0601 05:38:16.113133 140460941842304 api.py:587] Built Squeeze and Excitation with tensor shape: (4, 1, 1, 96)
I0601 05:38:16.147000 140460941842304 api.py:587] Project shape: (4, 128, 128, 24)
I0601 05:38:16.157407 140460941842304 api.py:587] Block mb_conv_block_2  input shape: (4, 128, 128, 24)
I0601 05:38:16.200918 140460941842304 api.py:587] Expand shape: (4, 128, 128, 144)
I0601 05:38:16.233961 140460941842304 api.py:587] DWConv shape: (4, 128, 128, 144)
I0601 05:38:16.269518 140460941842304 api.py:587] Built Squeeze and Excitation with tensor shape: (4, 1, 1, 144)
I0601 05:38:16.301350 140460941842304 api.py:587] Project shape: (4, 128, 128, 24)
I0601 05:38:16.312747 140460941842304 api.py:587] Block mb_conv_block_3  input shape: (4, 128, 128, 24)
I0601 05:38:16.346515 140460941842304 api.py:587] Expand shape: (4, 128, 128, 144)
I0601 05:38:16.381837 140460941842304 api.py:587] DWConv shape: (4, 64, 64, 144)
I0601 05:38:16.426858 140460941842304 api.py:587] Built Squeeze and Excitation with tensor shape: (4, 1, 1, 144)
I0601 05:38:16.459029 140460941842304 api.py:587] Project shape: (4, 64, 64, 40)
I0601 05:38:16.468211 140460941842304 api.py:587] Block mb_conv_block_4  input shape: (4, 64, 64, 40)
I0601 05:38:16.503680 140460941842304 api.py:587] Expand shape: (4, 64, 64, 240)
I0601 05:38:16.538903 140460941842304 api.py:587] DWConv shape: (4, 64, 64, 240)
I0601 05:38:16.573376 140460941842304 api.py:587] Built Squeeze and Excitation with tensor shape: (4, 1, 1, 240)
I0601 05:38:16.609440 140460941842304 api.py:587] Project shape: (4, 64, 64, 40)
I0601 05:38:16.619002 140460941842304 api.py:587] Block mb_conv_block_5  input shape: (4, 64, 64, 40)
I0601 05:38:16.652125 140460941842304 api.py:587] Expand shape: (4, 64, 64, 240)
I0601 05:38:16.685711 140460941842304 api.py:587] DWConv shape: (4, 32, 32, 240)
I0601 05:38:16.722809 140460941842304 api.py:587] Built Squeeze and Excitation with tensor shape: (4, 1, 1, 240)
I0601 05:38:16.753987 140460941842304 api.py:587] Project shape: (4, 32, 32, 80)
I0601 05:38:16.763768 140460941842304 api.py:587] Block mb_conv_block_6  input shape: (4, 32, 32, 80)
I0601 05:38:16.798378 140460941842304 api.py:587] Expand shape: (4, 32, 32, 480)
I0601 05:38:16.831592 140460941842304 api.py:587] DWConv shape: (4, 32, 32, 480)
I0601 05:38:16.866909 140460941842304 api.py:587] Built Squeeze and Excitation with tensor shape: (4, 1, 1, 480)
I0601 05:38:16.902912 140460941842304 api.py:587] Project shape: (4, 32, 32, 80)
I0601 05:38:16.915614 140460941842304 api.py:587] Block mb_conv_block_7  input shape: (4, 32, 32, 80)
I0601 05:38:16.951402 140460941842304 api.py:587] Expand shape: (4, 32, 32, 480)
I0601 05:38:16.985633 140460941842304 api.py:587] DWConv shape: (4, 32, 32, 480)
I0601 05:38:17.025779 140460941842304 api.py:587] Built Squeeze and Excitation with tensor shape: (4, 1, 1, 480)
I0601 05:38:17.058473 140460941842304 api.py:587] Project shape: (4, 32, 32, 80)
I0601 05:38:17.068195 140460941842304 api.py:587] Block mb_conv_block_8  input shape: (4, 32, 32, 80)
I0601 05:38:17.104321 140460941842304 api.py:587] Expand shape: (4, 32, 32, 480)
I0601 05:38:17.139899 140460941842304 api.py:587] DWConv shape: (4, 32, 32, 480)
I0601 05:38:17.178178 140460941842304 api.py:587] Built Squeeze and Excitation with tensor shape: (4, 1, 1, 480)
I0601 05:38:17.223506 140460941842304 api.py:587] Project shape: (4, 32, 32, 112)
I0601 05:38:17.233667 140460941842304 api.py:587] Block mb_conv_block_9  input shape: (4, 32, 32, 112)
I0601 05:38:17.267562 140460941842304 api.py:587] Expand shape: (4, 32, 32, 672)
I0601 05:38:17.300494 140460941842304 api.py:587] DWConv shape: (4, 32, 32, 672)
I0601 05:38:17.337823 140460941842304 api.py:587] Built Squeeze and Excitation with tensor shape: (4, 1, 1, 672)
I0601 05:38:17.370885 140460941842304 api.py:587] Project shape: (4, 32, 32, 112)
I0601 05:38:17.381354 140460941842304 api.py:587] Block mb_conv_block_10  input shape: (4, 32, 32, 112)
I0601 05:38:17.425960 140460941842304 api.py:587] Expand shape: (4, 32, 32, 672)
I0601 05:38:17.465528 140460941842304 api.py:587] DWConv shape: (4, 32, 32, 672)
I0601 05:38:17.501981 140460941842304 api.py:587] Built Squeeze and Excitation with tensor shape: (4, 1, 1, 672)
I0601 05:38:17.535618 140460941842304 api.py:587] Project shape: (4, 32, 32, 112)
I0601 05:38:17.545045 140460941842304 api.py:587] Block mb_conv_block_11  input shape: (4, 32, 32, 112)
I0601 05:38:17.577750 140460941842304 api.py:587] Expand shape: (4, 32, 32, 672)
I0601 05:38:17.613845 140460941842304 api.py:587] DWConv shape: (4, 16, 16, 672)
I0601 05:38:17.648429 140460941842304 api.py:587] Built Squeeze and Excitation with tensor shape: (4, 1, 1, 672)
I0601 05:38:17.679331 140460941842304 api.py:587] Project shape: (4, 16, 16, 192)
I0601 05:38:17.691922 140460941842304 api.py:587] Block mb_conv_block_12  input shape: (4, 16, 16, 192)
I0601 05:38:17.730681 140460941842304 api.py:587] Expand shape: (4, 16, 16, 1152)
I0601 05:38:17.768310 140460941842304 api.py:587] DWConv shape: (4, 16, 16, 1152)
I0601 05:38:17.805337 140460941842304 api.py:587] Built Squeeze and Excitation with tensor shape: (4, 1, 1, 1152)
I0601 05:38:17.839221 140460941842304 api.py:587] Project shape: (4, 16, 16, 192)
I0601 05:38:17.848821 140460941842304 api.py:587] Block mb_conv_block_13  input shape: (4, 16, 16, 192)
I0601 05:38:17.888498 140460941842304 api.py:587] Expand shape: (4, 16, 16, 1152)
I0601 05:38:17.929714 140460941842304 api.py:587] DWConv shape: (4, 16, 16, 1152)
I0601 05:38:17.967441 140460941842304 api.py:587] Built Squeeze and Excitation with tensor shape: (4, 1, 1, 1152)
I0601 05:38:18.000252 140460941842304 api.py:587] Project shape: (4, 16, 16, 192)
I0601 05:38:18.010449 140460941842304 api.py:587] Block mb_conv_block_14  input shape: (4, 16, 16, 192)
I0601 05:38:18.052438 140460941842304 api.py:587] Expand shape: (4, 16, 16, 1152)
I0601 05:38:18.093971 140460941842304 api.py:587] DWConv shape: (4, 16, 16, 1152)
I0601 05:38:18.132887 140460941842304 api.py:587] Built Squeeze and Excitation with tensor shape: (4, 1, 1, 1152)
I0601 05:38:18.166574 140460941842304 api.py:587] Project shape: (4, 16, 16, 192)
I0601 05:38:18.176187 140460941842304 api.py:587] Block mb_conv_block_15  input shape: (4, 16, 16, 192)
I0601 05:38:18.221516 140460941842304 api.py:587] Expand shape: (4, 16, 16, 1152)
I0601 05:38:18.258391 140460941842304 api.py:587] DWConv shape: (4, 16, 16, 1152)
I0601 05:38:18.294653 140460941842304 api.py:587] Built Squeeze and Excitation with tensor shape: (4, 1, 1, 1152)
I0601 05:38:18.327821 140460941842304 api.py:587] Project shape: (4, 16, 16, 320)
Parsing Inputs...
I0601 05:38:18.653207 140460941842304 efficientdet_arch.py:725] backbone params/flops = 3.595388M, 7.723610577B
I0601 05:38:18.711972 140460941842304 efficientdet_arch.py:471] building cell 0
I0601 05:38:18.712393 140460941842304 efficientdet_arch.py:648] fnode 0 : {'feat_level': 6, 'inputs_offsets': [3, 4]}
I0601 05:38:18.790723 140460941842304 efficientdet_arch.py:648] fnode 1 : {'feat_level': 5, 'inputs_offsets': [2, 5]}
I0601 05:38:18.910952 140460941842304 efficientdet_arch.py:648] fnode 2 : {'feat_level': 4, 'inputs_offsets': [1, 6]}
I0601 05:38:19.033912 140460941842304 efficientdet_arch.py:648] fnode 3 : {'feat_level': 3, 'inputs_offsets': [0, 7]}
I0601 05:38:19.155185 140460941842304 efficientdet_arch.py:648] fnode 4 : {'feat_level': 4, 'inputs_offsets': [1, 7, 8]}
I0601 05:38:19.300895 140460941842304 efficientdet_arch.py:648] fnode 5 : {'feat_level': 5, 'inputs_offsets': [2, 6, 9]}
I0601 05:38:19.448367 140460941842304 efficientdet_arch.py:648] fnode 6 : {'feat_level': 6, 'inputs_offsets': [3, 5, 10]}
I0601 05:38:19.545654 140460941842304 efficientdet_arch.py:648] fnode 7 : {'feat_level': 7, 'inputs_offsets': [4, 11]}
I0601 05:38:19.632055 140460941842304 efficientdet_arch.py:471] building cell 1
I0601 05:38:19.632450 140460941842304 efficientdet_arch.py:648] fnode 0 : {'feat_level': 6, 'inputs_offsets': [3, 4]}
I0601 05:38:19.719202 140460941842304 efficientdet_arch.py:648] fnode 1 : {'feat_level': 5, 'inputs_offsets': [2, 5]}
I0601 05:38:19.801726 140460941842304 efficientdet_arch.py:648] fnode 2 : {'feat_level': 4, 'inputs_offsets': [1, 6]}
I0601 05:38:19.890650 140460941842304 efficientdet_arch.py:648] fnode 3 : {'feat_level': 3, 'inputs_offsets': [0, 7]}
I0601 05:38:19.980336 140460941842304 efficientdet_arch.py:648] fnode 4 : {'feat_level': 4, 'inputs_offsets': [1, 7, 8]}
I0601 05:38:20.076389 140460941842304 efficientdet_arch.py:648] fnode 5 : {'feat_level': 5, 'inputs_offsets': [2, 6, 9]}
I0601 05:38:20.187259 140460941842304 efficientdet_arch.py:648] fnode 6 : {'feat_level': 6, 'inputs_offsets': [3, 5, 10]}
I0601 05:38:20.289635 140460941842304 efficientdet_arch.py:648] fnode 7 : {'feat_level': 7, 'inputs_offsets': [4, 11]}
I0601 05:38:20.381916 140460941842304 efficientdet_arch.py:471] building cell 2
I0601 05:38:20.382333 140460941842304 efficientdet_arch.py:648] fnode 0 : {'feat_level': 6, 'inputs_offsets': [3, 4]}
I0601 05:38:20.470864 140460941842304 efficientdet_arch.py:648] fnode 1 : {'feat_level': 5, 'inputs_offsets': [2, 5]}
I0601 05:38:20.571690 140460941842304 efficientdet_arch.py:648] fnode 2 : {'feat_level': 4, 'inputs_offsets': [1, 6]}
I0601 05:38:20.665080 140460941842304 efficientdet_arch.py:648] fnode 3 : {'feat_level': 3, 'inputs_offsets': [0, 7]}
I0601 05:38:20.769322 140460941842304 efficientdet_arch.py:648] fnode 4 : {'feat_level': 4, 'inputs_offsets': [1, 7, 8]}
I0601 05:38:20.869930 140460941842304 efficientdet_arch.py:648] fnode 5 : {'feat_level': 5, 'inputs_offsets': [2, 6, 9]}
I0601 05:38:20.977256 140460941842304 efficientdet_arch.py:648] fnode 6 : {'feat_level': 6, 'inputs_offsets': [3, 5, 10]}
I0601 05:38:21.077913 140460941842304 efficientdet_arch.py:648] fnode 7 : {'feat_level': 7, 'inputs_offsets': [4, 11]}
Parsing Inputs...
I0601 05:38:21.728706 140460941842304 efficientdet_arch.py:730] backbone+fpn params/flops = 3.791669M, 8.294117884B
Parsing Inputs...
I0601 05:38:25.041391 140460941842304 efficientdet_arch.py:735] backbone+fpn+box params/flops = 3.839117M, 9.243504892B
I0601 05:38:25.041746 140460941842304 utils.py:595] use mixed precision policy name float32
I0601 05:38:25.045943 140460941842304 det_model_fn.py:97] LR schedule method: cosine
I0601 05:38:25.389793 140460941842304 det_model_fn.py:589] Eval val with groudtruths None.
WARNING:tensorflow:From /content/automl/efficientdet/anchors.py:587: py_func (from tensorflow.python.ops.script_ops) is deprecated and will be removed in a future version.
Instructions for updating:
tf.py_func is deprecated in TF V2. Instead, there are two
    options available in V2.
    - tf.py_function takes a python function which manipulates tf eager
    tensors instead of numpy arrays. It's easy to convert a tf eager tensor to
    an ndarray (just call tensor.numpy()) but having access to eager tensors
    means `tf.py_function`s can use accelerators such as GPUs as well as
    being differentiable using a gradient tape.
    - tf.numpy_function maintains the semantics of the deprecated tf.py_func
    (it is not differentiable, and manipulates numpy arrays). It drops the
    stateful argument making all functions stateful.
    
W0601 05:38:25.398602 140460941842304 deprecation.py:323] From /content/automl/efficientdet/anchors.py:587: py_func (from tensorflow.python.ops.script_ops) is deprecated and will be removed in a future version.
Instructions for updating:
tf.py_func is deprecated in TF V2. Instead, there are two
    options available in V2.
    - tf.py_function takes a python function which manipulates tf eager
    tensors instead of numpy arrays. It's easy to convert a tf eager tensor to
    an ndarray (just call tensor.numpy()) but having access to eager tensors
    means `tf.py_function`s can use accelerators such as GPUs as well as
    being differentiable using a gradient tape.
    - tf.numpy_function maintains the semantics of the deprecated tf.py_func
    (it is not differentiable, and manipulates numpy arrays). It drops the
    stateful argument making all functions stateful.
    
INFO:tensorflow:Done calling model_fn.
I0601 05:38:25.449812 140460941842304 estimator.py:1171] Done calling model_fn.
INFO:tensorflow:Starting evaluation at 2020-06-01T05:38:25Z
I0601 05:38:25.467760 140460941842304 evaluation.py:255] Starting evaluation at 2020-06-01T05:38:25Z
INFO:tensorflow:Graph was finalized.
I0601 05:38:26.550288 140460941842304 monitored_session.py:246] Graph was finalized.
2020-06-01 05:38:26.550968: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:38:26.551510: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1561] Found device 0 with properties: 
pciBusID: 0000:00:04.0 name: Tesla T4 computeCapability: 7.5
coreClock: 1.59GHz coreCount: 40 deviceMemorySize: 14.73GiB deviceMemoryBandwidth: 298.08GiB/s
2020-06-01 05:38:26.551566: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.1
2020-06-01 05:38:26.551653: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcublas.so.10
2020-06-01 05:38:26.551682: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcufft.so.10
2020-06-01 05:38:26.551705: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcurand.so.10
2020-06-01 05:38:26.551735: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusolver.so.10
2020-06-01 05:38:26.551758: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusparse.so.10
2020-06-01 05:38:26.551779: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudnn.so.7
2020-06-01 05:38:26.551890: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:38:26.552491: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:38:26.552956: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1703] Adding visible gpu devices: 0
2020-06-01 05:38:26.553001: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1102] Device interconnect StreamExecutor with strength 1 edge matrix:
2020-06-01 05:38:26.553014: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1108]      0 
2020-06-01 05:38:26.553024: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1121] 0:   N 
2020-06-01 05:38:26.553140: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:38:26.553665: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:38:26.554150: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1247] Created TensorFlow device (/job:localhost/replica:0/task:0/device:GPU:0 with 13970 MB memory) -> physical GPU (device: 0, name: Tesla T4, pci bus id: 0000:00:04.0, compute capability: 7.5)
INFO:tensorflow:Restoring parameters from /tmp/model_dir/efficientdet-d0-scratch/model.ckpt-14
I0601 05:38:26.555339 140460941842304 saver.py:1293] Restoring parameters from /tmp/model_dir/efficientdet-d0-scratch/model.ckpt-14
INFO:tensorflow:Running local_init_op.
I0601 05:38:27.867699 140460941842304 session_manager.py:505] Running local_init_op.
INFO:tensorflow:Done running local_init_op.
I0601 05:38:27.978962 140460941842304 session_manager.py:508] Done running local_init_op.
INFO:tensorflow:Evaluation [1/14]
I0601 05:38:31.654607 140460941842304 evaluation.py:167] Evaluation [1/14]
INFO:tensorflow:Evaluation [2/14]
I0601 05:38:32.161419 140460941842304 evaluation.py:167] Evaluation [2/14]
INFO:tensorflow:Evaluation [3/14]
I0601 05:38:32.635456 140460941842304 evaluation.py:167] Evaluation [3/14]
INFO:tensorflow:Evaluation [4/14]
I0601 05:38:33.120565 140460941842304 evaluation.py:167] Evaluation [4/14]
INFO:tensorflow:Evaluation [5/14]
I0601 05:38:33.597185 140460941842304 evaluation.py:167] Evaluation [5/14]
INFO:tensorflow:Evaluation [6/14]
I0601 05:38:34.080617 140460941842304 evaluation.py:167] Evaluation [6/14]
INFO:tensorflow:Evaluation [7/14]
I0601 05:38:34.650703 140460941842304 evaluation.py:167] Evaluation [7/14]
INFO:tensorflow:Evaluation [8/14]
I0601 05:38:35.131882 140460941842304 evaluation.py:167] Evaluation [8/14]
INFO:tensorflow:Evaluation [9/14]
I0601 05:38:35.616208 140460941842304 evaluation.py:167] Evaluation [9/14]
INFO:tensorflow:Evaluation [10/14]
I0601 05:38:36.107428 140460941842304 evaluation.py:167] Evaluation [10/14]
INFO:tensorflow:Evaluation [11/14]
I0601 05:38:36.596470 140460941842304 evaluation.py:167] Evaluation [11/14]
INFO:tensorflow:Evaluation [12/14]
I0601 05:38:37.073881 140460941842304 evaluation.py:167] Evaluation [12/14]
INFO:tensorflow:Evaluation [13/14]
I0601 05:38:37.664412 140460941842304 evaluation.py:167] Evaluation [13/14]
INFO:tensorflow:Evaluation [14/14]
I0601 05:38:38.225920 140460941842304 evaluation.py:167] Evaluation [14/14]
creating index...
index created!
Loading and preparing results...
Converting ndarray to lists...
(5600, 7)
0/5600
DONE (t=0.02s)
creating index...
index created!
Running per image evaluation...
Evaluate annotation type *bbox*
DONE (t=0.29s).
Accumulating evaluation results...
DONE (t=0.07s).
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.003
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = -1.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = -1.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.002
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.002
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = -1.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = -1.000
INFO:tensorflow:Inference Time : 13.39342s
I0601 05:38:38.861362 140460941842304 evaluation.py:273] Inference Time : 13.39342s
INFO:tensorflow:Finished evaluation at 2020-06-01-05:38:38
I0601 05:38:38.861632 140460941842304 evaluation.py:276] Finished evaluation at 2020-06-01-05:38:38
INFO:tensorflow:Saving dict for global step 14: AP = 1.4968081e-06, AP50 = 4.20262e-06, AP75 = 2.8654225e-07, APl = -1.0, APm = -1.0, APs = 0.0025247524, ARl = -1.0, ARm = -1.0, ARmax1 = 0.0, ARmax10 = 0.0, ARmax100 = 0.00225, ARs = 0.00225, box_loss = 0.012224401, cls_loss = 1.2661626, global_step = 14, loss = 2.4825947
I0601 05:38:38.861842 140460941842304 estimator.py:2066] Saving dict for global step 14: AP = 1.4968081e-06, AP50 = 4.20262e-06, AP75 = 2.8654225e-07, APl = -1.0, APm = -1.0, APs = 0.0025247524, ARl = -1.0, ARm = -1.0, ARmax1 = 0.0, ARmax10 = 0.0, ARmax100 = 0.00225, ARs = 0.00225, box_loss = 0.012224401, cls_loss = 1.2661626, global_step = 14, loss = 2.4825947
INFO:tensorflow:Saving 'checkpoint_path' summary for global step 14: /tmp/model_dir/efficientdet-d0-scratch/model.ckpt-14
I0601 05:38:39.915983 140460941842304 estimator.py:2127] Saving 'checkpoint_path' summary for global step 14: /tmp/model_dir/efficientdet-d0-scratch/model.ckpt-14
INFO:tensorflow:evaluation_loop marked as finished
I0601 05:38:39.916708 140460941842304 error_handling.py:115] evaluation_loop marked as finished
I0601 05:38:39.916910 140460941842304 main.py:411] Evaluation results: {'AP': 1.4968081e-06, 'AP50': 4.20262e-06, 'AP75': 2.8654225e-07, 'APl': -1.0, 'APm': -1.0, 'APs': 0.0025247524, 'ARl': -1.0, 'ARm': -1.0, 'ARmax1': 0.0, 'ARmax10': 0.0, 'ARmax100': 0.00225, 'ARs': 0.00225, 'box_loss': 0.012224401, 'cls_loss': 1.2661626, 'loss': 2.4825947, 'global_step': 14}
INFO:tensorflow:/tmp/model_dir/efficientdet-d0-scratch/archive/model.ckpt-14 is not in all_model_checkpoint_paths. Manually adding it.
I0601 05:38:39.959321 140460941842304 checkpoint_management.py:102] /tmp/model_dir/efficientdet-d0-scratch/archive/model.ckpt-14 is not in all_model_checkpoint_paths. Manually adding it.
I0601 05:38:39.959901 140460941842304 utils.py:491] Copying checkpoint /tmp/model_dir/efficientdet-d0-scratch/model.ckpt-14 to /tmp/model_dir/efficientdet-d0-scratch/archive

4.3 Train Pascal VOC 2012 from COCO checkpoint for the whole net.



In [0]:

    
# generating train tfrecord is large, so we skip the execution here.
import os
if MODEL not in os.listdir():
  !wget https://storage.googleapis.com/cloud-tpu-checkpoints/efficientdet/coco/{MODEL}.tar.gz
  !tar xf {MODEL}.tar.gz

!mkdir /tmp/model_dir/
# key option: use --ckpt rather than --backbone_ckpt.
!python main.py --mode=train_and_eval \
    --training_file_pattern=tfrecord/{file_pattern} \
    --validation_file_pattern=tfrecord/{file_pattern} \
    --model_name={MODEL} \
    --model_dir=/tmp/model_dir/{MODEL}-finetune \
    --ckpt={MODEL} \
    --train_batch_size=8 \
    --eval_batch_size=8 --eval_samples={images_per_epoch}  \
    --num_examples_per_epoch={images_per_epoch}  --num_epochs=1  \
    --hparams="num_classes=20,moving_average_decay=0,mixed_precision=true"









    



mkdir: cannot create directory ‘/tmp/model_dir/’: File exists
2020-06-01 05:38:43.365099: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.1
WARNING:tensorflow:From main.py:234: The name tf.estimator.tpu.TPUConfig is deprecated. Please use tf.compat.v1.estimator.tpu.TPUConfig instead.

W0601 05:38:45.587865 140233166403456 module_wrapper.py:138] From main.py:234: The name tf.estimator.tpu.TPUConfig is deprecated. Please use tf.compat.v1.estimator.tpu.TPUConfig instead.

WARNING:tensorflow:From main.py:239: The name tf.estimator.tpu.InputPipelineConfig is deprecated. Please use tf.compat.v1.estimator.tpu.InputPipelineConfig instead.

W0601 05:38:45.588135 140233166403456 module_wrapper.py:138] From main.py:239: The name tf.estimator.tpu.InputPipelineConfig is deprecated. Please use tf.compat.v1.estimator.tpu.InputPipelineConfig instead.

WARNING:tensorflow:From main.py:247: The name tf.estimator.tpu.RunConfig is deprecated. Please use tf.compat.v1.estimator.tpu.RunConfig instead.

W0601 05:38:45.588366 140233166403456 module_wrapper.py:138] From main.py:247: The name tf.estimator.tpu.RunConfig is deprecated. Please use tf.compat.v1.estimator.tpu.RunConfig instead.

I0601 05:38:45.588604 140233166403456 main.py:262] {'name': 'efficientdet-d0', 'act_type': 'swish', 'image_size': (512, 512), 'target_size': None, 'input_rand_hflip': True, 'train_scale_min': 0.1, 'train_scale_max': 2.0, 'autoaugment_policy': None, 'use_augmix': False, 'augmix_params': (3, -1, 1), 'num_classes': 20, 'skip_crowd_during_training': True, 'label_id_mapping': None, 'max_instances_per_image': 100, 'min_level': 3, 'max_level': 7, 'num_scales': 3, 'aspect_ratios': [(1.0, 1.0), (1.4, 0.7), (0.7, 1.4)], 'anchor_scale': 4.0, 'is_training_bn': True, 'momentum': 0.9, 'optimizer': 'sgd', 'learning_rate': 0.08, 'lr_warmup_init': 0.008, 'lr_warmup_epoch': 1.0, 'first_lr_drop_epoch': 200.0, 'second_lr_drop_epoch': 250.0, 'poly_lr_power': 0.9, 'clip_gradients_norm': 10.0, 'num_epochs': 1, 'data_format': 'channels_last', 'alpha': 0.25, 'gamma': 1.5, 'delta': 0.1, 'box_loss_weight': 50.0, 'iou_loss_type': None, 'iou_loss_weight': 1.0, 'weight_decay': 4e-05, 'strategy': None, 'precision': 'mixed_float16', 'box_class_repeats': 3, 'fpn_cell_repeats': 3, 'fpn_num_filters': 64, 'separable_conv': True, 'apply_bn_for_resampling': True, 'conv_after_downsample': False, 'conv_bn_act_pattern': False, 'use_native_resize_op': True, 'pooling_type': None, 'fpn_name': None, 'fpn_weight_method': None, 'fpn_config': None, 'survival_prob': None, 'img_summary_steps': None, 'lr_decay_method': 'cosine', 'moving_average_decay': 0, 'ckpt_var_scope': None, 'var_exclude_expr': '.*/class-predict/.*', 'backbone_name': 'efficientnet-b0', 'backbone_config': None, 'var_freeze_expr': None, 'resnet_depth': 50, 'model_name': 'efficientdet-d0', 'iterations_per_loop': 100, 'model_dir': '/tmp/model_dir/efficientdet-d0-finetune', 'num_shards': 8, 'num_examples_per_epoch': 56, 'backbone_ckpt': '', 'ckpt': 'efficientdet-d0', 'val_json_file': None, 'testdev_dir': None, 'mode': 'train_and_eval'}
I0601 05:38:45.588753 140233166403456 main.py:373] Starting training cycle, epoch: 0.
WARNING:tensorflow:From main.py:374: The name tf.estimator.tpu.TPUEstimator is deprecated. Please use tf.compat.v1.estimator.tpu.TPUEstimator instead.

W0601 05:38:45.589545 140233166403456 module_wrapper.py:138] From main.py:374: The name tf.estimator.tpu.TPUEstimator is deprecated. Please use tf.compat.v1.estimator.tpu.TPUEstimator instead.

INFO:tensorflow:Using config: {'_model_dir': '/tmp/model_dir/efficientdet-d0-finetune', '_tf_random_seed': None, '_save_summary_steps': 100, '_save_checkpoints_steps': None, '_save_checkpoints_secs': 600, '_session_config': allow_soft_placement: true
, '_keep_checkpoint_max': 5, '_keep_checkpoint_every_n_hours': 10000, '_log_step_count_steps': None, '_train_distribute': None, '_device_fn': None, '_protocol': None, '_eval_distribute': None, '_experimental_distribute': None, '_experimental_max_worker_delay_secs': None, '_session_creation_timeout_secs': 7200, '_service': None, '_cluster_spec': ClusterSpec({}), '_task_type': 'worker', '_task_id': 0, '_global_id_in_cluster': 0, '_master': '', '_evaluation_master': '', '_is_chief': True, '_num_ps_replicas': 0, '_num_worker_replicas': 1, '_tpu_config': TPUConfig(iterations_per_loop=100, num_shards=8, num_cores_per_replica=None, per_host_input_for_training=3, tpu_job_name=None, initial_infeed_sleep_secs=None, input_partition_dims=None, eval_training_input_configuration=2, experimental_host_call_every_n_steps=1), '_cluster': None}
I0601 05:38:45.590010 140233166403456 estimator.py:191] Using config: {'_model_dir': '/tmp/model_dir/efficientdet-d0-finetune', '_tf_random_seed': None, '_save_summary_steps': 100, '_save_checkpoints_steps': None, '_save_checkpoints_secs': 600, '_session_config': allow_soft_placement: true
, '_keep_checkpoint_max': 5, '_keep_checkpoint_every_n_hours': 10000, '_log_step_count_steps': None, '_train_distribute': None, '_device_fn': None, '_protocol': None, '_eval_distribute': None, '_experimental_distribute': None, '_experimental_max_worker_delay_secs': None, '_session_creation_timeout_secs': 7200, '_service': None, '_cluster_spec': ClusterSpec({}), '_task_type': 'worker', '_task_id': 0, '_global_id_in_cluster': 0, '_master': '', '_evaluation_master': '', '_is_chief': True, '_num_ps_replicas': 0, '_num_worker_replicas': 1, '_tpu_config': TPUConfig(iterations_per_loop=100, num_shards=8, num_cores_per_replica=None, per_host_input_for_training=3, tpu_job_name=None, initial_infeed_sleep_secs=None, input_partition_dims=None, eval_training_input_configuration=2, experimental_host_call_every_n_steps=1), '_cluster': None}
INFO:tensorflow:_TPUContext: eval_on_tpu True
I0601 05:38:45.590748 140233166403456 tpu_context.py:216] _TPUContext: eval_on_tpu True
WARNING:tensorflow:eval_on_tpu ignored because use_tpu is False.
W0601 05:38:45.591170 140233166403456 tpu_context.py:218] eval_on_tpu ignored because use_tpu is False.
WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/ops/resource_variable_ops.py:1666: calling BaseResourceVariable.__init__ (from tensorflow.python.ops.resource_variable_ops) with constraint is deprecated and will be removed in a future version.
Instructions for updating:
If using Keras pass *_constraint arguments to layers.
W0601 05:38:45.596488 140233166403456 deprecation.py:506] From /usr/local/lib/python3.6/dist-packages/tensorflow/python/ops/resource_variable_ops.py:1666: calling BaseResourceVariable.__init__ (from tensorflow.python.ops.resource_variable_ops) with constraint is deprecated and will be removed in a future version.
Instructions for updating:
If using Keras pass *_constraint arguments to layers.
WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/training/training_util.py:236: Variable.initialized_value (from tensorflow.python.ops.variables) is deprecated and will be removed in a future version.
Instructions for updating:
Use Variable.read_value. Variables in 2.X are initialized automatically both in eager and graph (inside tf.defun) contexts.
W0601 05:38:45.596863 140233166403456 deprecation.py:323] From /usr/local/lib/python3.6/dist-packages/tensorflow/python/training/training_util.py:236: Variable.initialized_value (from tensorflow.python.ops.variables) is deprecated and will be removed in a future version.
Instructions for updating:
Use Variable.read_value. Variables in 2.X are initialized automatically both in eager and graph (inside tf.defun) contexts.
2020-06-01 05:38:45.605005: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcuda.so.1
2020-06-01 05:38:45.639140: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:38:45.639714: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1561] Found device 0 with properties: 
pciBusID: 0000:00:04.0 name: Tesla T4 computeCapability: 7.5
coreClock: 1.59GHz coreCount: 40 deviceMemorySize: 14.73GiB deviceMemoryBandwidth: 298.08GiB/s
2020-06-01 05:38:45.639756: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.1
2020-06-01 05:38:45.641296: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcublas.so.10
2020-06-01 05:38:45.642996: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcufft.so.10
2020-06-01 05:38:45.643349: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcurand.so.10
2020-06-01 05:38:45.645110: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusolver.so.10
2020-06-01 05:38:45.646026: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusparse.so.10
2020-06-01 05:38:45.649821: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudnn.so.7
2020-06-01 05:38:45.649959: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:38:45.650550: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:38:45.651030: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1703] Adding visible gpu devices: 0
WARNING:tensorflow:From /content/automl/efficientdet/dataloader.py:363: parallel_interleave (from tensorflow.python.data.experimental.ops.interleave_ops) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.data.Dataset.interleave(map_func, cycle_length, block_length, num_parallel_calls=tf.data.experimental.AUTOTUNE)` instead. If sloppy execution is desired, use `tf.data.Options.experimental_deterministic`.
W0601 05:38:45.680050 140233166403456 deprecation.py:323] From /content/automl/efficientdet/dataloader.py:363: parallel_interleave (from tensorflow.python.data.experimental.ops.interleave_ops) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.data.Dataset.interleave(map_func, cycle_length, block_length, num_parallel_calls=tf.data.experimental.AUTOTUNE)` instead. If sloppy execution is desired, use `tf.data.Options.experimental_deterministic`.
I0601 05:38:45.853356 140233166403456 dataloader.py:83] target_size = (512, 512), output_size = (512, 512)
INFO:tensorflow:Calling model_fn.
I0601 05:38:46.429213 140233166403456 estimator.py:1169] Calling model_fn.
INFO:tensorflow:Running train on CPU/GPU
I0601 05:38:46.429509 140233166403456 tpu_estimator.py:3171] Running train on CPU/GPU
I0601 05:38:46.429973 140233166403456 utils.py:595] use mixed precision policy name mixed_float16
I0601 05:38:46.435803 140233166403456 efficientdet_arch.py:720] act_type: swish
alpha: 0.25
anchor_scale: 4.0
apply_bn_for_resampling: true
aspect_ratios:
- !!python/tuple
    - 1.0
    - 1.0
- !!python/tuple
    - 1.4
    - 0.7
- !!python/tuple
    - 0.7
    - 1.4
augmix_params: !!python/tuple
- 3
- -1
- 1
autoaugment_policy: null
backbone_ckpt: ''
backbone_config: null
backbone_name: efficientnet-b0
batch_size: 8
box_class_repeats: 3
box_loss_weight: 50.0
ckpt: efficientdet-d0
ckpt_var_scope: null
clip_gradients_norm: 10.0
conv_after_downsample: false
conv_bn_act_pattern: false
data_format: channels_last
delta: 0.1
first_lr_drop_epoch: 200.0
fpn_cell_repeats: 3
fpn_config: null
fpn_name: null
fpn_num_filters: 64
fpn_weight_method: null
gamma: 1.5
image_size: !!python/tuple
- 512
- 512
img_summary_steps: null
input_rand_hflip: true
iou_loss_type: null
iou_loss_weight: 1.0
is_training_bn: true
iterations_per_loop: 100
label_id_mapping: null
learning_rate: 0.08
lr_decay_method: cosine
lr_warmup_epoch: 1.0
lr_warmup_init: 0.008
max_instances_per_image: 100
max_level: 7
min_level: 3
mode: train_and_eval
model_dir: /tmp/model_dir/efficientdet-d0-finetune
model_name: efficientdet-d0
momentum: 0.9
moving_average_decay: 0
name: efficientdet-d0
num_classes: 20
num_epochs: 1
num_examples_per_epoch: 56
num_scales: 3
num_shards: 8
optimizer: sgd
poly_lr_power: 0.9
pooling_type: null
precision: mixed_float16
resnet_depth: 50
second_lr_drop_epoch: 250.0
separable_conv: true
skip_crowd_during_training: true
strategy: null
survival_prob: null
target_size: null
testdev_dir: null
train_scale_max: 2.0
train_scale_min: 0.1
use_augmix: false
use_native_resize_op: true
use_tpu: false
val_json_file: null
var_exclude_expr: .*/class-predict/.*
var_freeze_expr: null
weight_decay: 4.0e-05

I0601 05:38:46.441061 140233166403456 efficientnet_builder.py:224] global_params= GlobalParams(batch_norm_momentum=0.99, batch_norm_epsilon=0.001, dropout_rate=0.2, data_format='channels_last', num_classes=1000, width_coefficient=1.0, depth_coefficient=1.0, depth_divisor=8, min_depth=None, survival_prob=0.0, relu_fn=functools.partial(<function activation_fn at 0x7f8a3d432d90>, act_type='swish'), batch_norm=<class 'utils.BatchNormalization'>, use_se=True, local_pooling=None, condconv_num_experts=None, clip_projection_output=False, blocks_args=['r1_k3_s11_e1_i32_o16_se0.25', 'r2_k3_s22_e6_i16_o24_se0.25', 'r2_k5_s22_e6_i24_o40_se0.25', 'r3_k3_s22_e6_i40_o80_se0.25', 'r3_k5_s11_e6_i80_o112_se0.25', 'r4_k5_s22_e6_i112_o192_se0.25', 'r1_k3_s11_e6_i192_o320_se0.25'], fix_head_stem=None)
I0601 05:38:49.344394 140233166403456 api.py:587] Built stem layers with output shape: (8, 256, 256, 32)
I0601 05:38:51.915598 140233166403456 api.py:587] Block mb_conv_block  input shape: (8, 256, 256, 32)
I0601 05:38:51.955710 140233166403456 api.py:587] DWConv shape: (8, 256, 256, 32)
I0601 05:38:52.124546 140233166403456 api.py:587] Built Squeeze and Excitation with tensor shape: (8, 1, 1, 32)
I0601 05:38:52.166485 140233166403456 api.py:587] Project shape: (8, 256, 256, 16)
I0601 05:38:52.220312 140233166403456 api.py:587] Block mb_conv_block_1  input shape: (8, 256, 256, 16)
I0601 05:38:52.259614 140233166403456 api.py:587] Expand shape: (8, 256, 256, 96)
I0601 05:38:52.297866 140233166403456 api.py:587] DWConv shape: (8, 128, 128, 96)
I0601 05:38:52.334243 140233166403456 api.py:587] Built Squeeze and Excitation with tensor shape: (8, 1, 1, 96)
I0601 05:38:52.371809 140233166403456 api.py:587] Project shape: (8, 128, 128, 24)
I0601 05:38:52.381985 140233166403456 api.py:587] Block mb_conv_block_2  input shape: (8, 128, 128, 24)
I0601 05:38:52.426074 140233166403456 api.py:587] Expand shape: (8, 128, 128, 144)
I0601 05:38:52.464259 140233166403456 api.py:587] DWConv shape: (8, 128, 128, 144)
I0601 05:38:52.498655 140233166403456 api.py:587] Built Squeeze and Excitation with tensor shape: (8, 1, 1, 144)
I0601 05:38:52.537778 140233166403456 api.py:587] Project shape: (8, 128, 128, 24)
I0601 05:38:52.547262 140233166403456 api.py:587] Block mb_conv_block_3  input shape: (8, 128, 128, 24)
I0601 05:38:52.587199 140233166403456 api.py:587] Expand shape: (8, 128, 128, 144)
I0601 05:38:52.632583 140233166403456 api.py:587] DWConv shape: (8, 64, 64, 144)
I0601 05:38:52.671356 140233166403456 api.py:587] Built Squeeze and Excitation with tensor shape: (8, 1, 1, 144)
I0601 05:38:52.708576 140233166403456 api.py:587] Project shape: (8, 64, 64, 40)
I0601 05:38:52.720406 140233166403456 api.py:587] Block mb_conv_block_4  input shape: (8, 64, 64, 40)
I0601 05:38:52.758789 140233166403456 api.py:587] Expand shape: (8, 64, 64, 240)
I0601 05:38:52.800186 140233166403456 api.py:587] DWConv shape: (8, 64, 64, 240)
I0601 05:38:52.834159 140233166403456 api.py:587] Built Squeeze and Excitation with tensor shape: (8, 1, 1, 240)
I0601 05:38:52.871949 140233166403456 api.py:587] Project shape: (8, 64, 64, 40)
I0601 05:38:52.881962 140233166403456 api.py:587] Block mb_conv_block_5  input shape: (8, 64, 64, 40)
I0601 05:38:52.925455 140233166403456 api.py:587] Expand shape: (8, 64, 64, 240)
I0601 05:38:52.964854 140233166403456 api.py:587] DWConv shape: (8, 32, 32, 240)
I0601 05:38:52.999113 140233166403456 api.py:587] Built Squeeze and Excitation with tensor shape: (8, 1, 1, 240)
I0601 05:38:53.038467 140233166403456 api.py:587] Project shape: (8, 32, 32, 80)
I0601 05:38:53.048321 140233166403456 api.py:587] Block mb_conv_block_6  input shape: (8, 32, 32, 80)
I0601 05:38:53.089564 140233166403456 api.py:587] Expand shape: (8, 32, 32, 480)
I0601 05:38:53.134333 140233166403456 api.py:587] DWConv shape: (8, 32, 32, 480)
I0601 05:38:53.170628 140233166403456 api.py:587] Built Squeeze and Excitation with tensor shape: (8, 1, 1, 480)
I0601 05:38:53.214241 140233166403456 api.py:587] Project shape: (8, 32, 32, 80)
I0601 05:38:53.223829 140233166403456 api.py:587] Block mb_conv_block_7  input shape: (8, 32, 32, 80)
I0601 05:38:53.263554 140233166403456 api.py:587] Expand shape: (8, 32, 32, 480)
I0601 05:38:53.303238 140233166403456 api.py:587] DWConv shape: (8, 32, 32, 480)
I0601 05:38:53.341757 140233166403456 api.py:587] Built Squeeze and Excitation with tensor shape: (8, 1, 1, 480)
I0601 05:38:53.385471 140233166403456 api.py:587] Project shape: (8, 32, 32, 80)
I0601 05:38:53.397035 140233166403456 api.py:587] Block mb_conv_block_8  input shape: (8, 32, 32, 80)
I0601 05:38:53.440428 140233166403456 api.py:587] Expand shape: (8, 32, 32, 480)
I0601 05:38:53.479074 140233166403456 api.py:587] DWConv shape: (8, 32, 32, 480)
I0601 05:38:53.514025 140233166403456 api.py:587] Built Squeeze and Excitation with tensor shape: (8, 1, 1, 480)
I0601 05:38:53.553477 140233166403456 api.py:587] Project shape: (8, 32, 32, 112)
I0601 05:38:53.563084 140233166403456 api.py:587] Block mb_conv_block_9  input shape: (8, 32, 32, 112)
I0601 05:38:53.608322 140233166403456 api.py:587] Expand shape: (8, 32, 32, 672)
I0601 05:38:53.649888 140233166403456 api.py:587] DWConv shape: (8, 32, 32, 672)
I0601 05:38:53.687101 140233166403456 api.py:587] Built Squeeze and Excitation with tensor shape: (8, 1, 1, 672)
I0601 05:38:53.727589 140233166403456 api.py:587] Project shape: (8, 32, 32, 112)
I0601 05:38:53.737505 140233166403456 api.py:587] Block mb_conv_block_10  input shape: (8, 32, 32, 112)
I0601 05:38:53.776249 140233166403456 api.py:587] Expand shape: (8, 32, 32, 672)
I0601 05:38:53.821624 140233166403456 api.py:587] DWConv shape: (8, 32, 32, 672)
I0601 05:38:53.856718 140233166403456 api.py:587] Built Squeeze and Excitation with tensor shape: (8, 1, 1, 672)
I0601 05:38:53.894424 140233166403456 api.py:587] Project shape: (8, 32, 32, 112)
I0601 05:38:53.904421 140233166403456 api.py:587] Block mb_conv_block_11  input shape: (8, 32, 32, 112)
I0601 05:38:53.946480 140233166403456 api.py:587] Expand shape: (8, 32, 32, 672)
I0601 05:38:53.987478 140233166403456 api.py:587] DWConv shape: (8, 16, 16, 672)
I0601 05:38:54.027495 140233166403456 api.py:587] Built Squeeze and Excitation with tensor shape: (8, 1, 1, 672)
I0601 05:38:54.066024 140233166403456 api.py:587] Project shape: (8, 16, 16, 192)
I0601 05:38:54.075679 140233166403456 api.py:587] Block mb_conv_block_12  input shape: (8, 16, 16, 192)
I0601 05:38:54.123987 140233166403456 api.py:587] Expand shape: (8, 16, 16, 1152)
I0601 05:38:54.170768 140233166403456 api.py:587] DWConv shape: (8, 16, 16, 1152)
I0601 05:38:54.212948 140233166403456 api.py:587] Built Squeeze and Excitation with tensor shape: (8, 1, 1, 1152)
I0601 05:38:54.251599 140233166403456 api.py:587] Project shape: (8, 16, 16, 192)
I0601 05:38:54.261630 140233166403456 api.py:587] Block mb_conv_block_13  input shape: (8, 16, 16, 192)
I0601 05:38:54.305206 140233166403456 api.py:587] Expand shape: (8, 16, 16, 1152)
I0601 05:38:54.350567 140233166403456 api.py:587] DWConv shape: (8, 16, 16, 1152)
I0601 05:38:54.388493 140233166403456 api.py:587] Built Squeeze and Excitation with tensor shape: (8, 1, 1, 1152)
I0601 05:38:54.431553 140233166403456 api.py:587] Project shape: (8, 16, 16, 192)
I0601 05:38:54.441506 140233166403456 api.py:587] Block mb_conv_block_14  input shape: (8, 16, 16, 192)
I0601 05:38:54.485316 140233166403456 api.py:587] Expand shape: (8, 16, 16, 1152)
I0601 05:38:54.528457 140233166403456 api.py:587] DWConv shape: (8, 16, 16, 1152)
I0601 05:38:54.567310 140233166403456 api.py:587] Built Squeeze and Excitation with tensor shape: (8, 1, 1, 1152)
I0601 05:38:54.609861 140233166403456 api.py:587] Project shape: (8, 16, 16, 192)
I0601 05:38:54.619987 140233166403456 api.py:587] Block mb_conv_block_15  input shape: (8, 16, 16, 192)
I0601 05:38:54.669234 140233166403456 api.py:587] Expand shape: (8, 16, 16, 1152)
I0601 05:38:54.713435 140233166403456 api.py:587] DWConv shape: (8, 16, 16, 1152)
I0601 05:38:54.753134 140233166403456 api.py:587] Built Squeeze and Excitation with tensor shape: (8, 1, 1, 1152)
I0601 05:38:54.793503 140233166403456 api.py:587] Project shape: (8, 16, 16, 320)
WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/profiler/internal/flops_registry.py:142: tensor_shape_from_node_def_name (from tensorflow.python.framework.graph_util_impl) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.compat.v1.graph_util.tensor_shape_from_node_def_name`
W0601 05:38:54.803869 140233166403456 deprecation.py:323] From /usr/local/lib/python3.6/dist-packages/tensorflow/python/profiler/internal/flops_registry.py:142: tensor_shape_from_node_def_name (from tensorflow.python.framework.graph_util_impl) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.compat.v1.graph_util.tensor_shape_from_node_def_name`
Parsing Inputs...
I0601 05:38:55.171925 140233166403456 efficientdet_arch.py:725] backbone params/flops = 3.595388M, 15.443714065B
WARNING:tensorflow:From /content/automl/efficientdet/efficientdet_arch.py:138: conv2d (from tensorflow.python.layers.convolutional) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.keras.layers.Conv2D` instead.
W0601 05:38:55.172413 140233166403456 deprecation.py:323] From /content/automl/efficientdet/efficientdet_arch.py:138: conv2d (from tensorflow.python.layers.convolutional) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.keras.layers.Conv2D` instead.
WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/layers/convolutional.py:424: Layer.apply (from tensorflow.python.keras.engine.base_layer_v1) is deprecated and will be removed in a future version.
Instructions for updating:
Please use `layer.__call__` method instead.
W0601 05:38:55.175734 140233166403456 deprecation.py:323] From /usr/local/lib/python3.6/dist-packages/tensorflow/python/layers/convolutional.py:424: Layer.apply (from tensorflow.python.keras.engine.base_layer_v1) is deprecated and will be removed in a future version.
Instructions for updating:
Please use `layer.__call__` method instead.
WARNING:tensorflow:From /content/automl/efficientdet/efficientdet_arch.py:164: max_pooling2d (from tensorflow.python.layers.pooling) is deprecated and will be removed in a future version.
Instructions for updating:
Use keras.layers.MaxPooling2D instead.
W0601 05:38:55.236947 140233166403456 deprecation.py:323] From /content/automl/efficientdet/efficientdet_arch.py:164: max_pooling2d (from tensorflow.python.layers.pooling) is deprecated and will be removed in a future version.
Instructions for updating:
Use keras.layers.MaxPooling2D instead.
I0601 05:38:55.243872 140233166403456 efficientdet_arch.py:471] building cell 0
I0601 05:38:55.244316 140233166403456 efficientdet_arch.py:648] fnode 0 : {'feat_level': 6, 'inputs_offsets': [3, 4]}
WARNING:tensorflow:From /content/automl/efficientdet/efficientdet_arch.py:685: separable_conv2d (from tensorflow.python.layers.convolutional) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.keras.layers.SeparableConv2D` instead.
W0601 05:38:55.258609 140233166403456 deprecation.py:323] From /content/automl/efficientdet/efficientdet_arch.py:685: separable_conv2d (from tensorflow.python.layers.convolutional) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.keras.layers.SeparableConv2D` instead.
I0601 05:38:55.329093 140233166403456 efficientdet_arch.py:648] fnode 1 : {'feat_level': 5, 'inputs_offsets': [2, 5]}
I0601 05:38:55.460398 140233166403456 efficientdet_arch.py:648] fnode 2 : {'feat_level': 4, 'inputs_offsets': [1, 6]}
I0601 05:38:55.598648 140233166403456 efficientdet_arch.py:648] fnode 3 : {'feat_level': 3, 'inputs_offsets': [0, 7]}
I0601 05:38:55.742610 140233166403456 efficientdet_arch.py:648] fnode 4 : {'feat_level': 4, 'inputs_offsets': [1, 7, 8]}
I0601 05:38:55.887705 140233166403456 efficientdet_arch.py:648] fnode 5 : {'feat_level': 5, 'inputs_offsets': [2, 6, 9]}
I0601 05:38:56.035905 140233166403456 efficientdet_arch.py:648] fnode 6 : {'feat_level': 6, 'inputs_offsets': [3, 5, 10]}
I0601 05:38:56.134672 140233166403456 efficientdet_arch.py:648] fnode 7 : {'feat_level': 7, 'inputs_offsets': [4, 11]}
I0601 05:38:56.335880 140233166403456 efficientdet_arch.py:471] building cell 1
I0601 05:38:56.336333 140233166403456 efficientdet_arch.py:648] fnode 0 : {'feat_level': 6, 'inputs_offsets': [3, 4]}
I0601 05:38:56.428193 140233166403456 efficientdet_arch.py:648] fnode 1 : {'feat_level': 5, 'inputs_offsets': [2, 5]}
I0601 05:38:56.522873 140233166403456 efficientdet_arch.py:648] fnode 2 : {'feat_level': 4, 'inputs_offsets': [1, 6]}
I0601 05:38:56.612578 140233166403456 efficientdet_arch.py:648] fnode 3 : {'feat_level': 3, 'inputs_offsets': [0, 7]}
I0601 05:38:56.722730 140233166403456 efficientdet_arch.py:648] fnode 4 : {'feat_level': 4, 'inputs_offsets': [1, 7, 8]}
I0601 05:38:56.823106 140233166403456 efficientdet_arch.py:648] fnode 5 : {'feat_level': 5, 'inputs_offsets': [2, 6, 9]}
I0601 05:38:56.918842 140233166403456 efficientdet_arch.py:648] fnode 6 : {'feat_level': 6, 'inputs_offsets': [3, 5, 10]}
I0601 05:38:57.021353 140233166403456 efficientdet_arch.py:648] fnode 7 : {'feat_level': 7, 'inputs_offsets': [4, 11]}
I0601 05:38:57.114783 140233166403456 efficientdet_arch.py:471] building cell 2
I0601 05:38:57.115130 140233166403456 efficientdet_arch.py:648] fnode 0 : {'feat_level': 6, 'inputs_offsets': [3, 4]}
I0601 05:38:57.213524 140233166403456 efficientdet_arch.py:648] fnode 1 : {'feat_level': 5, 'inputs_offsets': [2, 5]}
I0601 05:38:57.308053 140233166403456 efficientdet_arch.py:648] fnode 2 : {'feat_level': 4, 'inputs_offsets': [1, 6]}
I0601 05:38:57.408208 140233166403456 efficientdet_arch.py:648] fnode 3 : {'feat_level': 3, 'inputs_offsets': [0, 7]}
I0601 05:38:57.504937 140233166403456 efficientdet_arch.py:648] fnode 4 : {'feat_level': 4, 'inputs_offsets': [1, 7, 8]}
I0601 05:38:57.617605 140233166403456 efficientdet_arch.py:648] fnode 5 : {'feat_level': 5, 'inputs_offsets': [2, 6, 9]}
I0601 05:38:57.734111 140233166403456 efficientdet_arch.py:648] fnode 6 : {'feat_level': 6, 'inputs_offsets': [3, 5, 10]}
I0601 05:38:57.842484 140233166403456 efficientdet_arch.py:648] fnode 7 : {'feat_level': 7, 'inputs_offsets': [4, 11]}
Parsing Inputs...
I0601 05:38:58.557643 140233166403456 efficientdet_arch.py:730] backbone+fpn params/flops = 3.791669M, 16.584542043B
Parsing Inputs...
I0601 05:39:02.250997 140233166403456 efficientdet_arch.py:735] backbone+fpn+box params/flops = 3.839117M, 18.483276921B
I0601 05:39:02.251317 140233166403456 utils.py:595] use mixed precision policy name float32
I0601 05:39:02.255373 140233166403456 det_model_fn.py:97] LR schedule method: cosine
I0601 05:39:02.480796 140233166403456 utils.py:394] Adding scale summary ('lrn_rate', <tf.Tensor 'Select:0' shape=() dtype=float32>)
I0601 05:39:02.481896 140233166403456 utils.py:394] Adding scale summary ('trainloss/cls_loss', <tf.Tensor 'AddN:0' shape=() dtype=float32>)
I0601 05:39:02.482783 140233166403456 utils.py:394] Adding scale summary ('trainloss/box_loss', <tf.Tensor 'AddN_1:0' shape=() dtype=float32>)
I0601 05:39:02.483683 140233166403456 utils.py:394] Adding scale summary ('trainloss/det_loss', <tf.Tensor 'add_4:0' shape=() dtype=float32>)
I0601 05:39:02.484507 140233166403456 utils.py:394] Adding scale summary ('trainloss/reg_l2_loss', <tf.Tensor 'mul_14:0' shape=() dtype=float32>)
I0601 05:39:02.485368 140233166403456 utils.py:394] Adding scale summary ('trainloss/loss', <tf.Tensor 'add_5:0' shape=() dtype=float32>)
I0601 05:39:02.486211 140233166403456 det_model_fn.py:537] clip gradients norm by 10.000000
I0601 05:39:07.175653 140233166403456 utils.py:394] Adding scale summary ('gnorm', <tf.Tensor 'clip/global_norm/global_norm:0' shape=() dtype=float32>)
WARNING:tensorflow:From /content/automl/efficientdet/det_model_fn.py:659: The name tf.estimator.tpu.TPUEstimatorSpec is deprecated. Please use tf.compat.v1.estimator.tpu.TPUEstimatorSpec instead.

W0601 05:39:09.331756 140233166403456 module_wrapper.py:138] From /content/automl/efficientdet/det_model_fn.py:659: The name tf.estimator.tpu.TPUEstimatorSpec is deprecated. Please use tf.compat.v1.estimator.tpu.TPUEstimatorSpec instead.

I0601 05:39:09.375637 140233166403456 det_model_fn.py:638] restore variables from efficientdet-d0
I0601 05:39:09.375814 140233166403456 utils.py:73] Init model from checkpoint efficientdet-d0
I0601 05:39:09.394657 140233166403456 utils.py:108] Init global_step from ckpt var global_step
I0601 05:39:09.394810 140233166403456 utils.py:105] skip current_loss_scale (current_loss_scale) -- not in ckpt
I0601 05:39:09.394902 140233166403456 utils.py:105] skip good_steps (good_steps) -- not in ckpt
I0601 05:39:09.394985 140233166403456 utils.py:108] Init efficientnet-b0/stem/conv2d/kernel from ckpt var efficientnet-b0/stem/conv2d/kernel
I0601 05:39:09.395078 140233166403456 utils.py:108] Init efficientnet-b0/stem/tpu_batch_normalization/gamma from ckpt var efficientnet-b0/stem/tpu_batch_normalization/gamma
I0601 05:39:09.395157 140233166403456 utils.py:108] Init efficientnet-b0/stem/tpu_batch_normalization/beta from ckpt var efficientnet-b0/stem/tpu_batch_normalization/beta
I0601 05:39:09.395245 140233166403456 utils.py:108] Init efficientnet-b0/stem/tpu_batch_normalization/moving_mean from ckpt var efficientnet-b0/stem/tpu_batch_normalization/moving_mean
I0601 05:39:09.395364 140233166403456 utils.py:108] Init efficientnet-b0/stem/tpu_batch_normalization/moving_variance from ckpt var efficientnet-b0/stem/tpu_batch_normalization/moving_variance
I0601 05:39:09.395446 140233166403456 utils.py:108] Init efficientnet-b0/blocks_0/depthwise_conv2d/depthwise_kernel from ckpt var efficientnet-b0/blocks_0/depthwise_conv2d/depthwise_kernel
I0601 05:39:09.395530 140233166403456 utils.py:108] Init efficientnet-b0/blocks_0/tpu_batch_normalization/gamma from ckpt var efficientnet-b0/blocks_0/tpu_batch_normalization/gamma
I0601 05:39:09.395612 140233166403456 utils.py:108] Init efficientnet-b0/blocks_0/tpu_batch_normalization/beta from ckpt var efficientnet-b0/blocks_0/tpu_batch_normalization/beta
I0601 05:39:09.395702 140233166403456 utils.py:108] Init efficientnet-b0/blocks_0/tpu_batch_normalization/moving_mean from ckpt var efficientnet-b0/blocks_0/tpu_batch_normalization/moving_mean
I0601 05:39:09.395785 140233166403456 utils.py:108] Init efficientnet-b0/blocks_0/tpu_batch_normalization/moving_variance from ckpt var efficientnet-b0/blocks_0/tpu_batch_normalization/moving_variance
I0601 05:39:09.395869 140233166403456 utils.py:108] Init efficientnet-b0/blocks_0/se/conv2d/kernel from ckpt var efficientnet-b0/blocks_0/se/conv2d/kernel
I0601 05:39:09.395951 140233166403456 utils.py:108] Init efficientnet-b0/blocks_0/se/conv2d/bias from ckpt var efficientnet-b0/blocks_0/se/conv2d/bias
I0601 05:39:09.396034 140233166403456 utils.py:108] Init efficientnet-b0/blocks_0/se/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_0/se/conv2d_1/kernel
I0601 05:39:09.396115 140233166403456 utils.py:108] Init efficientnet-b0/blocks_0/se/conv2d_1/bias from ckpt var efficientnet-b0/blocks_0/se/conv2d_1/bias
I0601 05:39:09.396197 140233166403456 utils.py:108] Init efficientnet-b0/blocks_0/conv2d/kernel from ckpt var efficientnet-b0/blocks_0/conv2d/kernel
I0601 05:39:09.396296 140233166403456 utils.py:108] Init efficientnet-b0/blocks_0/tpu_batch_normalization_1/gamma from ckpt var efficientnet-b0/blocks_0/tpu_batch_normalization_1/gamma
I0601 05:39:09.396387 140233166403456 utils.py:108] Init efficientnet-b0/blocks_0/tpu_batch_normalization_1/beta from ckpt var efficientnet-b0/blocks_0/tpu_batch_normalization_1/beta
I0601 05:39:09.396468 140233166403456 utils.py:108] Init efficientnet-b0/blocks_0/tpu_batch_normalization_1/moving_mean from ckpt var efficientnet-b0/blocks_0/tpu_batch_normalization_1/moving_mean
I0601 05:39:09.396551 140233166403456 utils.py:108] Init efficientnet-b0/blocks_0/tpu_batch_normalization_1/moving_variance from ckpt var efficientnet-b0/blocks_0/tpu_batch_normalization_1/moving_variance
I0601 05:39:09.396643 140233166403456 utils.py:108] Init efficientnet-b0/blocks_1/conv2d/kernel from ckpt var efficientnet-b0/blocks_1/conv2d/kernel
I0601 05:39:09.396720 140233166403456 utils.py:108] Init efficientnet-b0/blocks_1/tpu_batch_normalization/gamma from ckpt var efficientnet-b0/blocks_1/tpu_batch_normalization/gamma
I0601 05:39:09.396797 140233166403456 utils.py:108] Init efficientnet-b0/blocks_1/tpu_batch_normalization/beta from ckpt var efficientnet-b0/blocks_1/tpu_batch_normalization/beta
I0601 05:39:09.396875 140233166403456 utils.py:108] Init efficientnet-b0/blocks_1/tpu_batch_normalization/moving_mean from ckpt var efficientnet-b0/blocks_1/tpu_batch_normalization/moving_mean
I0601 05:39:09.396951 140233166403456 utils.py:108] Init efficientnet-b0/blocks_1/tpu_batch_normalization/moving_variance from ckpt var efficientnet-b0/blocks_1/tpu_batch_normalization/moving_variance
I0601 05:39:09.397027 140233166403456 utils.py:108] Init efficientnet-b0/blocks_1/depthwise_conv2d/depthwise_kernel from ckpt var efficientnet-b0/blocks_1/depthwise_conv2d/depthwise_kernel
I0601 05:39:09.397111 140233166403456 utils.py:108] Init efficientnet-b0/blocks_1/tpu_batch_normalization_1/gamma from ckpt var efficientnet-b0/blocks_1/tpu_batch_normalization_1/gamma
I0601 05:39:09.397184 140233166403456 utils.py:108] Init efficientnet-b0/blocks_1/tpu_batch_normalization_1/beta from ckpt var efficientnet-b0/blocks_1/tpu_batch_normalization_1/beta
I0601 05:39:09.397257 140233166403456 utils.py:108] Init efficientnet-b0/blocks_1/tpu_batch_normalization_1/moving_mean from ckpt var efficientnet-b0/blocks_1/tpu_batch_normalization_1/moving_mean
I0601 05:39:09.397348 140233166403456 utils.py:108] Init efficientnet-b0/blocks_1/tpu_batch_normalization_1/moving_variance from ckpt var efficientnet-b0/blocks_1/tpu_batch_normalization_1/moving_variance
I0601 05:39:09.397421 140233166403456 utils.py:108] Init efficientnet-b0/blocks_1/se/conv2d/kernel from ckpt var efficientnet-b0/blocks_1/se/conv2d/kernel
I0601 05:39:09.397493 140233166403456 utils.py:108] Init efficientnet-b0/blocks_1/se/conv2d/bias from ckpt var efficientnet-b0/blocks_1/se/conv2d/bias
I0601 05:39:09.397565 140233166403456 utils.py:108] Init efficientnet-b0/blocks_1/se/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_1/se/conv2d_1/kernel
I0601 05:39:09.397644 140233166403456 utils.py:108] Init efficientnet-b0/blocks_1/se/conv2d_1/bias from ckpt var efficientnet-b0/blocks_1/se/conv2d_1/bias
I0601 05:39:09.397715 140233166403456 utils.py:108] Init efficientnet-b0/blocks_1/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_1/conv2d_1/kernel
I0601 05:39:09.397787 140233166403456 utils.py:108] Init efficientnet-b0/blocks_1/tpu_batch_normalization_2/gamma from ckpt var efficientnet-b0/blocks_1/tpu_batch_normalization_2/gamma
I0601 05:39:09.397858 140233166403456 utils.py:108] Init efficientnet-b0/blocks_1/tpu_batch_normalization_2/beta from ckpt var efficientnet-b0/blocks_1/tpu_batch_normalization_2/beta
I0601 05:39:09.397932 140233166403456 utils.py:108] Init efficientnet-b0/blocks_1/tpu_batch_normalization_2/moving_mean from ckpt var efficientnet-b0/blocks_1/tpu_batch_normalization_2/moving_mean
I0601 05:39:09.398004 140233166403456 utils.py:108] Init efficientnet-b0/blocks_1/tpu_batch_normalization_2/moving_variance from ckpt var efficientnet-b0/blocks_1/tpu_batch_normalization_2/moving_variance
I0601 05:39:09.398075 140233166403456 utils.py:108] Init efficientnet-b0/blocks_2/conv2d/kernel from ckpt var efficientnet-b0/blocks_2/conv2d/kernel
I0601 05:39:09.398149 140233166403456 utils.py:108] Init efficientnet-b0/blocks_2/tpu_batch_normalization/gamma from ckpt var efficientnet-b0/blocks_2/tpu_batch_normalization/gamma
I0601 05:39:09.398220 140233166403456 utils.py:108] Init efficientnet-b0/blocks_2/tpu_batch_normalization/beta from ckpt var efficientnet-b0/blocks_2/tpu_batch_normalization/beta
I0601 05:39:09.398308 140233166403456 utils.py:108] Init efficientnet-b0/blocks_2/tpu_batch_normalization/moving_mean from ckpt var efficientnet-b0/blocks_2/tpu_batch_normalization/moving_mean
I0601 05:39:09.398386 140233166403456 utils.py:108] Init efficientnet-b0/blocks_2/tpu_batch_normalization/moving_variance from ckpt var efficientnet-b0/blocks_2/tpu_batch_normalization/moving_variance
I0601 05:39:09.398461 140233166403456 utils.py:108] Init efficientnet-b0/blocks_2/depthwise_conv2d/depthwise_kernel from ckpt var efficientnet-b0/blocks_2/depthwise_conv2d/depthwise_kernel
I0601 05:39:09.398534 140233166403456 utils.py:108] Init efficientnet-b0/blocks_2/tpu_batch_normalization_1/gamma from ckpt var efficientnet-b0/blocks_2/tpu_batch_normalization_1/gamma
I0601 05:39:09.398606 140233166403456 utils.py:108] Init efficientnet-b0/blocks_2/tpu_batch_normalization_1/beta from ckpt var efficientnet-b0/blocks_2/tpu_batch_normalization_1/beta
I0601 05:39:09.398686 140233166403456 utils.py:108] Init efficientnet-b0/blocks_2/tpu_batch_normalization_1/moving_mean from ckpt var efficientnet-b0/blocks_2/tpu_batch_normalization_1/moving_mean
I0601 05:39:09.398759 140233166403456 utils.py:108] Init efficientnet-b0/blocks_2/tpu_batch_normalization_1/moving_variance from ckpt var efficientnet-b0/blocks_2/tpu_batch_normalization_1/moving_variance
I0601 05:39:09.398832 140233166403456 utils.py:108] Init efficientnet-b0/blocks_2/se/conv2d/kernel from ckpt var efficientnet-b0/blocks_2/se/conv2d/kernel
I0601 05:39:09.398904 140233166403456 utils.py:108] Init efficientnet-b0/blocks_2/se/conv2d/bias from ckpt var efficientnet-b0/blocks_2/se/conv2d/bias
I0601 05:39:09.398976 140233166403456 utils.py:108] Init efficientnet-b0/blocks_2/se/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_2/se/conv2d_1/kernel
I0601 05:39:09.399054 140233166403456 utils.py:108] Init efficientnet-b0/blocks_2/se/conv2d_1/bias from ckpt var efficientnet-b0/blocks_2/se/conv2d_1/bias
I0601 05:39:09.399127 140233166403456 utils.py:108] Init efficientnet-b0/blocks_2/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_2/conv2d_1/kernel
I0601 05:39:09.399202 140233166403456 utils.py:108] Init efficientnet-b0/blocks_2/tpu_batch_normalization_2/gamma from ckpt var efficientnet-b0/blocks_2/tpu_batch_normalization_2/gamma
I0601 05:39:09.399291 140233166403456 utils.py:108] Init efficientnet-b0/blocks_2/tpu_batch_normalization_2/beta from ckpt var efficientnet-b0/blocks_2/tpu_batch_normalization_2/beta
I0601 05:39:09.399364 140233166403456 utils.py:108] Init efficientnet-b0/blocks_2/tpu_batch_normalization_2/moving_mean from ckpt var efficientnet-b0/blocks_2/tpu_batch_normalization_2/moving_mean
I0601 05:39:09.399437 140233166403456 utils.py:108] Init efficientnet-b0/blocks_2/tpu_batch_normalization_2/moving_variance from ckpt var efficientnet-b0/blocks_2/tpu_batch_normalization_2/moving_variance
I0601 05:39:09.399510 140233166403456 utils.py:108] Init efficientnet-b0/blocks_3/conv2d/kernel from ckpt var efficientnet-b0/blocks_3/conv2d/kernel
I0601 05:39:09.399582 140233166403456 utils.py:108] Init efficientnet-b0/blocks_3/tpu_batch_normalization/gamma from ckpt var efficientnet-b0/blocks_3/tpu_batch_normalization/gamma
I0601 05:39:09.399658 140233166403456 utils.py:108] Init efficientnet-b0/blocks_3/tpu_batch_normalization/beta from ckpt var efficientnet-b0/blocks_3/tpu_batch_normalization/beta
I0601 05:39:09.399731 140233166403456 utils.py:108] Init efficientnet-b0/blocks_3/tpu_batch_normalization/moving_mean from ckpt var efficientnet-b0/blocks_3/tpu_batch_normalization/moving_mean
I0601 05:39:09.399805 140233166403456 utils.py:108] Init efficientnet-b0/blocks_3/tpu_batch_normalization/moving_variance from ckpt var efficientnet-b0/blocks_3/tpu_batch_normalization/moving_variance
I0601 05:39:09.399877 140233166403456 utils.py:108] Init efficientnet-b0/blocks_3/depthwise_conv2d/depthwise_kernel from ckpt var efficientnet-b0/blocks_3/depthwise_conv2d/depthwise_kernel
I0601 05:39:09.399950 140233166403456 utils.py:108] Init efficientnet-b0/blocks_3/tpu_batch_normalization_1/gamma from ckpt var efficientnet-b0/blocks_3/tpu_batch_normalization_1/gamma
I0601 05:39:09.400024 140233166403456 utils.py:108] Init efficientnet-b0/blocks_3/tpu_batch_normalization_1/beta from ckpt var efficientnet-b0/blocks_3/tpu_batch_normalization_1/beta
I0601 05:39:09.400097 140233166403456 utils.py:108] Init efficientnet-b0/blocks_3/tpu_batch_normalization_1/moving_mean from ckpt var efficientnet-b0/blocks_3/tpu_batch_normalization_1/moving_mean
I0601 05:39:09.400170 140233166403456 utils.py:108] Init efficientnet-b0/blocks_3/tpu_batch_normalization_1/moving_variance from ckpt var efficientnet-b0/blocks_3/tpu_batch_normalization_1/moving_variance
I0601 05:39:09.400242 140233166403456 utils.py:108] Init efficientnet-b0/blocks_3/se/conv2d/kernel from ckpt var efficientnet-b0/blocks_3/se/conv2d/kernel
I0601 05:39:09.400327 140233166403456 utils.py:108] Init efficientnet-b0/blocks_3/se/conv2d/bias from ckpt var efficientnet-b0/blocks_3/se/conv2d/bias
I0601 05:39:09.400401 140233166403456 utils.py:108] Init efficientnet-b0/blocks_3/se/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_3/se/conv2d_1/kernel
I0601 05:39:09.400473 140233166403456 utils.py:108] Init efficientnet-b0/blocks_3/se/conv2d_1/bias from ckpt var efficientnet-b0/blocks_3/se/conv2d_1/bias
I0601 05:39:09.400545 140233166403456 utils.py:108] Init efficientnet-b0/blocks_3/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_3/conv2d_1/kernel
I0601 05:39:09.400617 140233166403456 utils.py:108] Init efficientnet-b0/blocks_3/tpu_batch_normalization_2/gamma from ckpt var efficientnet-b0/blocks_3/tpu_batch_normalization_2/gamma
I0601 05:39:09.400696 140233166403456 utils.py:108] Init efficientnet-b0/blocks_3/tpu_batch_normalization_2/beta from ckpt var efficientnet-b0/blocks_3/tpu_batch_normalization_2/beta
I0601 05:39:09.400772 140233166403456 utils.py:108] Init efficientnet-b0/blocks_3/tpu_batch_normalization_2/moving_mean from ckpt var efficientnet-b0/blocks_3/tpu_batch_normalization_2/moving_mean
I0601 05:39:09.400850 140233166403456 utils.py:108] Init efficientnet-b0/blocks_3/tpu_batch_normalization_2/moving_variance from ckpt var efficientnet-b0/blocks_3/tpu_batch_normalization_2/moving_variance
I0601 05:39:09.400922 140233166403456 utils.py:108] Init efficientnet-b0/blocks_4/conv2d/kernel from ckpt var efficientnet-b0/blocks_4/conv2d/kernel
I0601 05:39:09.400992 140233166403456 utils.py:108] Init efficientnet-b0/blocks_4/tpu_batch_normalization/gamma from ckpt var efficientnet-b0/blocks_4/tpu_batch_normalization/gamma
I0601 05:39:09.401064 140233166403456 utils.py:108] Init efficientnet-b0/blocks_4/tpu_batch_normalization/beta from ckpt var efficientnet-b0/blocks_4/tpu_batch_normalization/beta
I0601 05:39:09.401135 140233166403456 utils.py:108] Init efficientnet-b0/blocks_4/tpu_batch_normalization/moving_mean from ckpt var efficientnet-b0/blocks_4/tpu_batch_normalization/moving_mean
I0601 05:39:09.401207 140233166403456 utils.py:108] Init efficientnet-b0/blocks_4/tpu_batch_normalization/moving_variance from ckpt var efficientnet-b0/blocks_4/tpu_batch_normalization/moving_variance
I0601 05:39:09.401290 140233166403456 utils.py:108] Init efficientnet-b0/blocks_4/depthwise_conv2d/depthwise_kernel from ckpt var efficientnet-b0/blocks_4/depthwise_conv2d/depthwise_kernel
I0601 05:39:09.401365 140233166403456 utils.py:108] Init efficientnet-b0/blocks_4/tpu_batch_normalization_1/gamma from ckpt var efficientnet-b0/blocks_4/tpu_batch_normalization_1/gamma
I0601 05:39:09.401436 140233166403456 utils.py:108] Init efficientnet-b0/blocks_4/tpu_batch_normalization_1/beta from ckpt var efficientnet-b0/blocks_4/tpu_batch_normalization_1/beta
I0601 05:39:09.401507 140233166403456 utils.py:108] Init efficientnet-b0/blocks_4/tpu_batch_normalization_1/moving_mean from ckpt var efficientnet-b0/blocks_4/tpu_batch_normalization_1/moving_mean
I0601 05:39:09.401582 140233166403456 utils.py:108] Init efficientnet-b0/blocks_4/tpu_batch_normalization_1/moving_variance from ckpt var efficientnet-b0/blocks_4/tpu_batch_normalization_1/moving_variance
I0601 05:39:09.401659 140233166403456 utils.py:108] Init efficientnet-b0/blocks_4/se/conv2d/kernel from ckpt var efficientnet-b0/blocks_4/se/conv2d/kernel
I0601 05:39:09.401734 140233166403456 utils.py:108] Init efficientnet-b0/blocks_4/se/conv2d/bias from ckpt var efficientnet-b0/blocks_4/se/conv2d/bias
I0601 05:39:09.401811 140233166403456 utils.py:108] Init efficientnet-b0/blocks_4/se/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_4/se/conv2d_1/kernel
I0601 05:39:09.401884 140233166403456 utils.py:108] Init efficientnet-b0/blocks_4/se/conv2d_1/bias from ckpt var efficientnet-b0/blocks_4/se/conv2d_1/bias
I0601 05:39:09.401956 140233166403456 utils.py:108] Init efficientnet-b0/blocks_4/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_4/conv2d_1/kernel
I0601 05:39:09.402033 140233166403456 utils.py:108] Init efficientnet-b0/blocks_4/tpu_batch_normalization_2/gamma from ckpt var efficientnet-b0/blocks_4/tpu_batch_normalization_2/gamma
I0601 05:39:09.402105 140233166403456 utils.py:108] Init efficientnet-b0/blocks_4/tpu_batch_normalization_2/beta from ckpt var efficientnet-b0/blocks_4/tpu_batch_normalization_2/beta
I0601 05:39:09.402178 140233166403456 utils.py:108] Init efficientnet-b0/blocks_4/tpu_batch_normalization_2/moving_mean from ckpt var efficientnet-b0/blocks_4/tpu_batch_normalization_2/moving_mean
I0601 05:39:09.402253 140233166403456 utils.py:108] Init efficientnet-b0/blocks_4/tpu_batch_normalization_2/moving_variance from ckpt var efficientnet-b0/blocks_4/tpu_batch_normalization_2/moving_variance
I0601 05:39:09.478443 140233166403456 utils.py:108] Init efficientnet-b0/blocks_5/conv2d/kernel from ckpt var efficientnet-b0/blocks_5/conv2d/kernel
I0601 05:39:09.478590 140233166403456 utils.py:108] Init efficientnet-b0/blocks_5/tpu_batch_normalization/gamma from ckpt var efficientnet-b0/blocks_5/tpu_batch_normalization/gamma
I0601 05:39:09.478709 140233166403456 utils.py:108] Init efficientnet-b0/blocks_5/tpu_batch_normalization/beta from ckpt var efficientnet-b0/blocks_5/tpu_batch_normalization/beta
I0601 05:39:09.478804 140233166403456 utils.py:108] Init efficientnet-b0/blocks_5/tpu_batch_normalization/moving_mean from ckpt var efficientnet-b0/blocks_5/tpu_batch_normalization/moving_mean
I0601 05:39:09.478899 140233166403456 utils.py:108] Init efficientnet-b0/blocks_5/tpu_batch_normalization/moving_variance from ckpt var efficientnet-b0/blocks_5/tpu_batch_normalization/moving_variance
I0601 05:39:09.478995 140233166403456 utils.py:108] Init efficientnet-b0/blocks_5/depthwise_conv2d/depthwise_kernel from ckpt var efficientnet-b0/blocks_5/depthwise_conv2d/depthwise_kernel
I0601 05:39:09.479170 140233166403456 utils.py:108] Init efficientnet-b0/blocks_5/tpu_batch_normalization_1/gamma from ckpt var efficientnet-b0/blocks_5/tpu_batch_normalization_1/gamma
I0601 05:39:09.479353 140233166403456 utils.py:108] Init efficientnet-b0/blocks_5/tpu_batch_normalization_1/beta from ckpt var efficientnet-b0/blocks_5/tpu_batch_normalization_1/beta
I0601 05:39:09.479465 140233166403456 utils.py:108] Init efficientnet-b0/blocks_5/tpu_batch_normalization_1/moving_mean from ckpt var efficientnet-b0/blocks_5/tpu_batch_normalization_1/moving_mean
I0601 05:39:09.479566 140233166403456 utils.py:108] Init efficientnet-b0/blocks_5/tpu_batch_normalization_1/moving_variance from ckpt var efficientnet-b0/blocks_5/tpu_batch_normalization_1/moving_variance
I0601 05:39:09.479683 140233166403456 utils.py:108] Init efficientnet-b0/blocks_5/se/conv2d/kernel from ckpt var efficientnet-b0/blocks_5/se/conv2d/kernel
I0601 05:39:09.479765 140233166403456 utils.py:108] Init efficientnet-b0/blocks_5/se/conv2d/bias from ckpt var efficientnet-b0/blocks_5/se/conv2d/bias
I0601 05:39:09.479843 140233166403456 utils.py:108] Init efficientnet-b0/blocks_5/se/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_5/se/conv2d_1/kernel
I0601 05:39:09.479918 140233166403456 utils.py:108] Init efficientnet-b0/blocks_5/se/conv2d_1/bias from ckpt var efficientnet-b0/blocks_5/se/conv2d_1/bias
I0601 05:39:09.479994 140233166403456 utils.py:108] Init efficientnet-b0/blocks_5/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_5/conv2d_1/kernel
I0601 05:39:09.480067 140233166403456 utils.py:108] Init efficientnet-b0/blocks_5/tpu_batch_normalization_2/gamma from ckpt var efficientnet-b0/blocks_5/tpu_batch_normalization_2/gamma
I0601 05:39:09.480148 140233166403456 utils.py:108] Init efficientnet-b0/blocks_5/tpu_batch_normalization_2/beta from ckpt var efficientnet-b0/blocks_5/tpu_batch_normalization_2/beta
I0601 05:39:09.480223 140233166403456 utils.py:108] Init efficientnet-b0/blocks_5/tpu_batch_normalization_2/moving_mean from ckpt var efficientnet-b0/blocks_5/tpu_batch_normalization_2/moving_mean
I0601 05:39:09.480322 140233166403456 utils.py:108] Init efficientnet-b0/blocks_5/tpu_batch_normalization_2/moving_variance from ckpt var efficientnet-b0/blocks_5/tpu_batch_normalization_2/moving_variance
I0601 05:39:09.480395 140233166403456 utils.py:108] Init efficientnet-b0/blocks_6/conv2d/kernel from ckpt var efficientnet-b0/blocks_6/conv2d/kernel
I0601 05:39:09.480469 140233166403456 utils.py:108] Init efficientnet-b0/blocks_6/tpu_batch_normalization/gamma from ckpt var efficientnet-b0/blocks_6/tpu_batch_normalization/gamma
I0601 05:39:09.480541 140233166403456 utils.py:108] Init efficientnet-b0/blocks_6/tpu_batch_normalization/beta from ckpt var efficientnet-b0/blocks_6/tpu_batch_normalization/beta
I0601 05:39:09.480614 140233166403456 utils.py:108] Init efficientnet-b0/blocks_6/tpu_batch_normalization/moving_mean from ckpt var efficientnet-b0/blocks_6/tpu_batch_normalization/moving_mean
I0601 05:39:09.480693 140233166403456 utils.py:108] Init efficientnet-b0/blocks_6/tpu_batch_normalization/moving_variance from ckpt var efficientnet-b0/blocks_6/tpu_batch_normalization/moving_variance
I0601 05:39:09.480766 140233166403456 utils.py:108] Init efficientnet-b0/blocks_6/depthwise_conv2d/depthwise_kernel from ckpt var efficientnet-b0/blocks_6/depthwise_conv2d/depthwise_kernel
I0601 05:39:09.480838 140233166403456 utils.py:108] Init efficientnet-b0/blocks_6/tpu_batch_normalization_1/gamma from ckpt var efficientnet-b0/blocks_6/tpu_batch_normalization_1/gamma
I0601 05:39:09.480909 140233166403456 utils.py:108] Init efficientnet-b0/blocks_6/tpu_batch_normalization_1/beta from ckpt var efficientnet-b0/blocks_6/tpu_batch_normalization_1/beta
I0601 05:39:09.480981 140233166403456 utils.py:108] Init efficientnet-b0/blocks_6/tpu_batch_normalization_1/moving_mean from ckpt var efficientnet-b0/blocks_6/tpu_batch_normalization_1/moving_mean
I0601 05:39:09.481052 140233166403456 utils.py:108] Init efficientnet-b0/blocks_6/tpu_batch_normalization_1/moving_variance from ckpt var efficientnet-b0/blocks_6/tpu_batch_normalization_1/moving_variance
I0601 05:39:09.481129 140233166403456 utils.py:108] Init efficientnet-b0/blocks_6/se/conv2d/kernel from ckpt var efficientnet-b0/blocks_6/se/conv2d/kernel
I0601 05:39:09.481203 140233166403456 utils.py:108] Init efficientnet-b0/blocks_6/se/conv2d/bias from ckpt var efficientnet-b0/blocks_6/se/conv2d/bias
I0601 05:39:09.481291 140233166403456 utils.py:108] Init efficientnet-b0/blocks_6/se/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_6/se/conv2d_1/kernel
I0601 05:39:09.481364 140233166403456 utils.py:108] Init efficientnet-b0/blocks_6/se/conv2d_1/bias from ckpt var efficientnet-b0/blocks_6/se/conv2d_1/bias
I0601 05:39:09.481438 140233166403456 utils.py:108] Init efficientnet-b0/blocks_6/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_6/conv2d_1/kernel
I0601 05:39:09.481511 140233166403456 utils.py:108] Init efficientnet-b0/blocks_6/tpu_batch_normalization_2/gamma from ckpt var efficientnet-b0/blocks_6/tpu_batch_normalization_2/gamma
I0601 05:39:09.481583 140233166403456 utils.py:108] Init efficientnet-b0/blocks_6/tpu_batch_normalization_2/beta from ckpt var efficientnet-b0/blocks_6/tpu_batch_normalization_2/beta
I0601 05:39:09.481656 140233166403456 utils.py:108] Init efficientnet-b0/blocks_6/tpu_batch_normalization_2/moving_mean from ckpt var efficientnet-b0/blocks_6/tpu_batch_normalization_2/moving_mean
I0601 05:39:09.481727 140233166403456 utils.py:108] Init efficientnet-b0/blocks_6/tpu_batch_normalization_2/moving_variance from ckpt var efficientnet-b0/blocks_6/tpu_batch_normalization_2/moving_variance
I0601 05:39:09.481799 140233166403456 utils.py:108] Init efficientnet-b0/blocks_7/conv2d/kernel from ckpt var efficientnet-b0/blocks_7/conv2d/kernel
I0601 05:39:09.481874 140233166403456 utils.py:108] Init efficientnet-b0/blocks_7/tpu_batch_normalization/gamma from ckpt var efficientnet-b0/blocks_7/tpu_batch_normalization/gamma
I0601 05:39:09.481947 140233166403456 utils.py:108] Init efficientnet-b0/blocks_7/tpu_batch_normalization/beta from ckpt var efficientnet-b0/blocks_7/tpu_batch_normalization/beta
I0601 05:39:09.482020 140233166403456 utils.py:108] Init efficientnet-b0/blocks_7/tpu_batch_normalization/moving_mean from ckpt var efficientnet-b0/blocks_7/tpu_batch_normalization/moving_mean
I0601 05:39:09.482092 140233166403456 utils.py:108] Init efficientnet-b0/blocks_7/tpu_batch_normalization/moving_variance from ckpt var efficientnet-b0/blocks_7/tpu_batch_normalization/moving_variance
I0601 05:39:09.482172 140233166403456 utils.py:108] Init efficientnet-b0/blocks_7/depthwise_conv2d/depthwise_kernel from ckpt var efficientnet-b0/blocks_7/depthwise_conv2d/depthwise_kernel
I0601 05:39:09.482246 140233166403456 utils.py:108] Init efficientnet-b0/blocks_7/tpu_batch_normalization_1/gamma from ckpt var efficientnet-b0/blocks_7/tpu_batch_normalization_1/gamma
I0601 05:39:09.482330 140233166403456 utils.py:108] Init efficientnet-b0/blocks_7/tpu_batch_normalization_1/beta from ckpt var efficientnet-b0/blocks_7/tpu_batch_normalization_1/beta
I0601 05:39:09.482403 140233166403456 utils.py:108] Init efficientnet-b0/blocks_7/tpu_batch_normalization_1/moving_mean from ckpt var efficientnet-b0/blocks_7/tpu_batch_normalization_1/moving_mean
I0601 05:39:09.482478 140233166403456 utils.py:108] Init efficientnet-b0/blocks_7/tpu_batch_normalization_1/moving_variance from ckpt var efficientnet-b0/blocks_7/tpu_batch_normalization_1/moving_variance
I0601 05:39:09.482550 140233166403456 utils.py:108] Init efficientnet-b0/blocks_7/se/conv2d/kernel from ckpt var efficientnet-b0/blocks_7/se/conv2d/kernel
I0601 05:39:09.482622 140233166403456 utils.py:108] Init efficientnet-b0/blocks_7/se/conv2d/bias from ckpt var efficientnet-b0/blocks_7/se/conv2d/bias
I0601 05:39:09.482696 140233166403456 utils.py:108] Init efficientnet-b0/blocks_7/se/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_7/se/conv2d_1/kernel
I0601 05:39:09.482769 140233166403456 utils.py:108] Init efficientnet-b0/blocks_7/se/conv2d_1/bias from ckpt var efficientnet-b0/blocks_7/se/conv2d_1/bias
I0601 05:39:09.482841 140233166403456 utils.py:108] Init efficientnet-b0/blocks_7/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_7/conv2d_1/kernel
I0601 05:39:09.482913 140233166403456 utils.py:108] Init efficientnet-b0/blocks_7/tpu_batch_normalization_2/gamma from ckpt var efficientnet-b0/blocks_7/tpu_batch_normalization_2/gamma
I0601 05:39:09.482984 140233166403456 utils.py:108] Init efficientnet-b0/blocks_7/tpu_batch_normalization_2/beta from ckpt var efficientnet-b0/blocks_7/tpu_batch_normalization_2/beta
I0601 05:39:09.483054 140233166403456 utils.py:108] Init efficientnet-b0/blocks_7/tpu_batch_normalization_2/moving_mean from ckpt var efficientnet-b0/blocks_7/tpu_batch_normalization_2/moving_mean
I0601 05:39:09.483131 140233166403456 utils.py:108] Init efficientnet-b0/blocks_7/tpu_batch_normalization_2/moving_variance from ckpt var efficientnet-b0/blocks_7/tpu_batch_normalization_2/moving_variance
I0601 05:39:09.483201 140233166403456 utils.py:108] Init efficientnet-b0/blocks_8/conv2d/kernel from ckpt var efficientnet-b0/blocks_8/conv2d/kernel
I0601 05:39:09.483285 140233166403456 utils.py:108] Init efficientnet-b0/blocks_8/tpu_batch_normalization/gamma from ckpt var efficientnet-b0/blocks_8/tpu_batch_normalization/gamma
I0601 05:39:09.483355 140233166403456 utils.py:108] Init efficientnet-b0/blocks_8/tpu_batch_normalization/beta from ckpt var efficientnet-b0/blocks_8/tpu_batch_normalization/beta
I0601 05:39:09.483428 140233166403456 utils.py:108] Init efficientnet-b0/blocks_8/tpu_batch_normalization/moving_mean from ckpt var efficientnet-b0/blocks_8/tpu_batch_normalization/moving_mean
I0601 05:39:09.483501 140233166403456 utils.py:108] Init efficientnet-b0/blocks_8/tpu_batch_normalization/moving_variance from ckpt var efficientnet-b0/blocks_8/tpu_batch_normalization/moving_variance
I0601 05:39:09.483573 140233166403456 utils.py:108] Init efficientnet-b0/blocks_8/depthwise_conv2d/depthwise_kernel from ckpt var efficientnet-b0/blocks_8/depthwise_conv2d/depthwise_kernel
I0601 05:39:09.483645 140233166403456 utils.py:108] Init efficientnet-b0/blocks_8/tpu_batch_normalization_1/gamma from ckpt var efficientnet-b0/blocks_8/tpu_batch_normalization_1/gamma
I0601 05:39:09.483716 140233166403456 utils.py:108] Init efficientnet-b0/blocks_8/tpu_batch_normalization_1/beta from ckpt var efficientnet-b0/blocks_8/tpu_batch_normalization_1/beta
I0601 05:39:09.483786 140233166403456 utils.py:108] Init efficientnet-b0/blocks_8/tpu_batch_normalization_1/moving_mean from ckpt var efficientnet-b0/blocks_8/tpu_batch_normalization_1/moving_mean
I0601 05:39:09.483860 140233166403456 utils.py:108] Init efficientnet-b0/blocks_8/tpu_batch_normalization_1/moving_variance from ckpt var efficientnet-b0/blocks_8/tpu_batch_normalization_1/moving_variance
I0601 05:39:09.483931 140233166403456 utils.py:108] Init efficientnet-b0/blocks_8/se/conv2d/kernel from ckpt var efficientnet-b0/blocks_8/se/conv2d/kernel
I0601 05:39:09.484006 140233166403456 utils.py:108] Init efficientnet-b0/blocks_8/se/conv2d/bias from ckpt var efficientnet-b0/blocks_8/se/conv2d/bias
I0601 05:39:09.484081 140233166403456 utils.py:108] Init efficientnet-b0/blocks_8/se/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_8/se/conv2d_1/kernel
I0601 05:39:09.484156 140233166403456 utils.py:108] Init efficientnet-b0/blocks_8/se/conv2d_1/bias from ckpt var efficientnet-b0/blocks_8/se/conv2d_1/bias
I0601 05:39:09.484230 140233166403456 utils.py:108] Init efficientnet-b0/blocks_8/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_8/conv2d_1/kernel
I0601 05:39:09.484314 140233166403456 utils.py:108] Init efficientnet-b0/blocks_8/tpu_batch_normalization_2/gamma from ckpt var efficientnet-b0/blocks_8/tpu_batch_normalization_2/gamma
I0601 05:39:09.484388 140233166403456 utils.py:108] Init efficientnet-b0/blocks_8/tpu_batch_normalization_2/beta from ckpt var efficientnet-b0/blocks_8/tpu_batch_normalization_2/beta
I0601 05:39:09.484469 140233166403456 utils.py:108] Init efficientnet-b0/blocks_8/tpu_batch_normalization_2/moving_mean from ckpt var efficientnet-b0/blocks_8/tpu_batch_normalization_2/moving_mean
I0601 05:39:09.484544 140233166403456 utils.py:108] Init efficientnet-b0/blocks_8/tpu_batch_normalization_2/moving_variance from ckpt var efficientnet-b0/blocks_8/tpu_batch_normalization_2/moving_variance
I0601 05:39:09.484617 140233166403456 utils.py:108] Init efficientnet-b0/blocks_9/conv2d/kernel from ckpt var efficientnet-b0/blocks_9/conv2d/kernel
I0601 05:39:09.484691 140233166403456 utils.py:108] Init efficientnet-b0/blocks_9/tpu_batch_normalization/gamma from ckpt var efficientnet-b0/blocks_9/tpu_batch_normalization/gamma
I0601 05:39:09.484762 140233166403456 utils.py:108] Init efficientnet-b0/blocks_9/tpu_batch_normalization/beta from ckpt var efficientnet-b0/blocks_9/tpu_batch_normalization/beta
I0601 05:39:09.484836 140233166403456 utils.py:108] Init efficientnet-b0/blocks_9/tpu_batch_normalization/moving_mean from ckpt var efficientnet-b0/blocks_9/tpu_batch_normalization/moving_mean
I0601 05:39:09.484909 140233166403456 utils.py:108] Init efficientnet-b0/blocks_9/tpu_batch_normalization/moving_variance from ckpt var efficientnet-b0/blocks_9/tpu_batch_normalization/moving_variance
I0601 05:39:09.484979 140233166403456 utils.py:108] Init efficientnet-b0/blocks_9/depthwise_conv2d/depthwise_kernel from ckpt var efficientnet-b0/blocks_9/depthwise_conv2d/depthwise_kernel
I0601 05:39:09.485051 140233166403456 utils.py:108] Init efficientnet-b0/blocks_9/tpu_batch_normalization_1/gamma from ckpt var efficientnet-b0/blocks_9/tpu_batch_normalization_1/gamma
I0601 05:39:09.485147 140233166403456 utils.py:108] Init efficientnet-b0/blocks_9/tpu_batch_normalization_1/beta from ckpt var efficientnet-b0/blocks_9/tpu_batch_normalization_1/beta
I0601 05:39:09.485226 140233166403456 utils.py:108] Init efficientnet-b0/blocks_9/tpu_batch_normalization_1/moving_mean from ckpt var efficientnet-b0/blocks_9/tpu_batch_normalization_1/moving_mean
I0601 05:39:09.485315 140233166403456 utils.py:108] Init efficientnet-b0/blocks_9/tpu_batch_normalization_1/moving_variance from ckpt var efficientnet-b0/blocks_9/tpu_batch_normalization_1/moving_variance
I0601 05:39:09.485393 140233166403456 utils.py:108] Init efficientnet-b0/blocks_9/se/conv2d/kernel from ckpt var efficientnet-b0/blocks_9/se/conv2d/kernel
I0601 05:39:09.485471 140233166403456 utils.py:108] Init efficientnet-b0/blocks_9/se/conv2d/bias from ckpt var efficientnet-b0/blocks_9/se/conv2d/bias
I0601 05:39:09.485548 140233166403456 utils.py:108] Init efficientnet-b0/blocks_9/se/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_9/se/conv2d_1/kernel
I0601 05:39:09.485626 140233166403456 utils.py:108] Init efficientnet-b0/blocks_9/se/conv2d_1/bias from ckpt var efficientnet-b0/blocks_9/se/conv2d_1/bias
I0601 05:39:09.485703 140233166403456 utils.py:108] Init efficientnet-b0/blocks_9/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_9/conv2d_1/kernel
I0601 05:39:09.485780 140233166403456 utils.py:108] Init efficientnet-b0/blocks_9/tpu_batch_normalization_2/gamma from ckpt var efficientnet-b0/blocks_9/tpu_batch_normalization_2/gamma
I0601 05:39:09.485867 140233166403456 utils.py:108] Init efficientnet-b0/blocks_9/tpu_batch_normalization_2/beta from ckpt var efficientnet-b0/blocks_9/tpu_batch_normalization_2/beta
I0601 05:39:09.485958 140233166403456 utils.py:108] Init efficientnet-b0/blocks_9/tpu_batch_normalization_2/moving_mean from ckpt var efficientnet-b0/blocks_9/tpu_batch_normalization_2/moving_mean
I0601 05:39:09.486035 140233166403456 utils.py:108] Init efficientnet-b0/blocks_9/tpu_batch_normalization_2/moving_variance from ckpt var efficientnet-b0/blocks_9/tpu_batch_normalization_2/moving_variance
I0601 05:39:09.486113 140233166403456 utils.py:108] Init efficientnet-b0/blocks_10/conv2d/kernel from ckpt var efficientnet-b0/blocks_10/conv2d/kernel
I0601 05:39:09.486196 140233166403456 utils.py:108] Init efficientnet-b0/blocks_10/tpu_batch_normalization/gamma from ckpt var efficientnet-b0/blocks_10/tpu_batch_normalization/gamma
I0601 05:39:09.486286 140233166403456 utils.py:108] Init efficientnet-b0/blocks_10/tpu_batch_normalization/beta from ckpt var efficientnet-b0/blocks_10/tpu_batch_normalization/beta
I0601 05:39:09.486371 140233166403456 utils.py:108] Init efficientnet-b0/blocks_10/tpu_batch_normalization/moving_mean from ckpt var efficientnet-b0/blocks_10/tpu_batch_normalization/moving_mean
I0601 05:39:09.486443 140233166403456 utils.py:108] Init efficientnet-b0/blocks_10/tpu_batch_normalization/moving_variance from ckpt var efficientnet-b0/blocks_10/tpu_batch_normalization/moving_variance
I0601 05:39:09.486515 140233166403456 utils.py:108] Init efficientnet-b0/blocks_10/depthwise_conv2d/depthwise_kernel from ckpt var efficientnet-b0/blocks_10/depthwise_conv2d/depthwise_kernel
I0601 05:39:09.486588 140233166403456 utils.py:108] Init efficientnet-b0/blocks_10/tpu_batch_normalization_1/gamma from ckpt var efficientnet-b0/blocks_10/tpu_batch_normalization_1/gamma
I0601 05:39:09.486662 140233166403456 utils.py:108] Init efficientnet-b0/blocks_10/tpu_batch_normalization_1/beta from ckpt var efficientnet-b0/blocks_10/tpu_batch_normalization_1/beta
I0601 05:39:09.579752 140233166403456 utils.py:108] Init efficientnet-b0/blocks_10/tpu_batch_normalization_1/moving_mean from ckpt var efficientnet-b0/blocks_10/tpu_batch_normalization_1/moving_mean
I0601 05:39:09.579918 140233166403456 utils.py:108] Init efficientnet-b0/blocks_10/tpu_batch_normalization_1/moving_variance from ckpt var efficientnet-b0/blocks_10/tpu_batch_normalization_1/moving_variance
I0601 05:39:09.580045 140233166403456 utils.py:108] Init efficientnet-b0/blocks_10/se/conv2d/kernel from ckpt var efficientnet-b0/blocks_10/se/conv2d/kernel
I0601 05:39:09.580178 140233166403456 utils.py:108] Init efficientnet-b0/blocks_10/se/conv2d/bias from ckpt var efficientnet-b0/blocks_10/se/conv2d/bias
I0601 05:39:09.580321 140233166403456 utils.py:108] Init efficientnet-b0/blocks_10/se/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_10/se/conv2d_1/kernel
I0601 05:39:09.580432 140233166403456 utils.py:108] Init efficientnet-b0/blocks_10/se/conv2d_1/bias from ckpt var efficientnet-b0/blocks_10/se/conv2d_1/bias
I0601 05:39:09.580530 140233166403456 utils.py:108] Init efficientnet-b0/blocks_10/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_10/conv2d_1/kernel
I0601 05:39:09.580643 140233166403456 utils.py:108] Init efficientnet-b0/blocks_10/tpu_batch_normalization_2/gamma from ckpt var efficientnet-b0/blocks_10/tpu_batch_normalization_2/gamma
I0601 05:39:09.580740 140233166403456 utils.py:108] Init efficientnet-b0/blocks_10/tpu_batch_normalization_2/beta from ckpt var efficientnet-b0/blocks_10/tpu_batch_normalization_2/beta
I0601 05:39:09.580813 140233166403456 utils.py:108] Init efficientnet-b0/blocks_10/tpu_batch_normalization_2/moving_mean from ckpt var efficientnet-b0/blocks_10/tpu_batch_normalization_2/moving_mean
I0601 05:39:09.580882 140233166403456 utils.py:108] Init efficientnet-b0/blocks_10/tpu_batch_normalization_2/moving_variance from ckpt var efficientnet-b0/blocks_10/tpu_batch_normalization_2/moving_variance
I0601 05:39:09.580953 140233166403456 utils.py:108] Init efficientnet-b0/blocks_11/conv2d/kernel from ckpt var efficientnet-b0/blocks_11/conv2d/kernel
I0601 05:39:09.581029 140233166403456 utils.py:108] Init efficientnet-b0/blocks_11/tpu_batch_normalization/gamma from ckpt var efficientnet-b0/blocks_11/tpu_batch_normalization/gamma
I0601 05:39:09.581105 140233166403456 utils.py:108] Init efficientnet-b0/blocks_11/tpu_batch_normalization/beta from ckpt var efficientnet-b0/blocks_11/tpu_batch_normalization/beta
I0601 05:39:09.581208 140233166403456 utils.py:108] Init efficientnet-b0/blocks_11/tpu_batch_normalization/moving_mean from ckpt var efficientnet-b0/blocks_11/tpu_batch_normalization/moving_mean
I0601 05:39:09.581321 140233166403456 utils.py:108] Init efficientnet-b0/blocks_11/tpu_batch_normalization/moving_variance from ckpt var efficientnet-b0/blocks_11/tpu_batch_normalization/moving_variance
I0601 05:39:09.581416 140233166403456 utils.py:108] Init efficientnet-b0/blocks_11/depthwise_conv2d/depthwise_kernel from ckpt var efficientnet-b0/blocks_11/depthwise_conv2d/depthwise_kernel
I0601 05:39:09.581508 140233166403456 utils.py:108] Init efficientnet-b0/blocks_11/tpu_batch_normalization_1/gamma from ckpt var efficientnet-b0/blocks_11/tpu_batch_normalization_1/gamma
I0601 05:39:09.581592 140233166403456 utils.py:108] Init efficientnet-b0/blocks_11/tpu_batch_normalization_1/beta from ckpt var efficientnet-b0/blocks_11/tpu_batch_normalization_1/beta
I0601 05:39:09.581684 140233166403456 utils.py:108] Init efficientnet-b0/blocks_11/tpu_batch_normalization_1/moving_mean from ckpt var efficientnet-b0/blocks_11/tpu_batch_normalization_1/moving_mean
I0601 05:39:09.581779 140233166403456 utils.py:108] Init efficientnet-b0/blocks_11/tpu_batch_normalization_1/moving_variance from ckpt var efficientnet-b0/blocks_11/tpu_batch_normalization_1/moving_variance
I0601 05:39:09.581890 140233166403456 utils.py:108] Init efficientnet-b0/blocks_11/se/conv2d/kernel from ckpt var efficientnet-b0/blocks_11/se/conv2d/kernel
I0601 05:39:09.581973 140233166403456 utils.py:108] Init efficientnet-b0/blocks_11/se/conv2d/bias from ckpt var efficientnet-b0/blocks_11/se/conv2d/bias
I0601 05:39:09.582057 140233166403456 utils.py:108] Init efficientnet-b0/blocks_11/se/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_11/se/conv2d_1/kernel
I0601 05:39:09.582149 140233166403456 utils.py:108] Init efficientnet-b0/blocks_11/se/conv2d_1/bias from ckpt var efficientnet-b0/blocks_11/se/conv2d_1/bias
I0601 05:39:09.582243 140233166403456 utils.py:108] Init efficientnet-b0/blocks_11/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_11/conv2d_1/kernel
I0601 05:39:09.582370 140233166403456 utils.py:108] Init efficientnet-b0/blocks_11/tpu_batch_normalization_2/gamma from ckpt var efficientnet-b0/blocks_11/tpu_batch_normalization_2/gamma
I0601 05:39:09.582453 140233166403456 utils.py:108] Init efficientnet-b0/blocks_11/tpu_batch_normalization_2/beta from ckpt var efficientnet-b0/blocks_11/tpu_batch_normalization_2/beta
I0601 05:39:09.582535 140233166403456 utils.py:108] Init efficientnet-b0/blocks_11/tpu_batch_normalization_2/moving_mean from ckpt var efficientnet-b0/blocks_11/tpu_batch_normalization_2/moving_mean
I0601 05:39:09.582617 140233166403456 utils.py:108] Init efficientnet-b0/blocks_11/tpu_batch_normalization_2/moving_variance from ckpt var efficientnet-b0/blocks_11/tpu_batch_normalization_2/moving_variance
I0601 05:39:09.582708 140233166403456 utils.py:108] Init efficientnet-b0/blocks_12/conv2d/kernel from ckpt var efficientnet-b0/blocks_12/conv2d/kernel
I0601 05:39:09.582797 140233166403456 utils.py:108] Init efficientnet-b0/blocks_12/tpu_batch_normalization/gamma from ckpt var efficientnet-b0/blocks_12/tpu_batch_normalization/gamma
I0601 05:39:09.582885 140233166403456 utils.py:108] Init efficientnet-b0/blocks_12/tpu_batch_normalization/beta from ckpt var efficientnet-b0/blocks_12/tpu_batch_normalization/beta
I0601 05:39:09.582967 140233166403456 utils.py:108] Init efficientnet-b0/blocks_12/tpu_batch_normalization/moving_mean from ckpt var efficientnet-b0/blocks_12/tpu_batch_normalization/moving_mean
I0601 05:39:09.583068 140233166403456 utils.py:108] Init efficientnet-b0/blocks_12/tpu_batch_normalization/moving_variance from ckpt var efficientnet-b0/blocks_12/tpu_batch_normalization/moving_variance
I0601 05:39:09.583152 140233166403456 utils.py:108] Init efficientnet-b0/blocks_12/depthwise_conv2d/depthwise_kernel from ckpt var efficientnet-b0/blocks_12/depthwise_conv2d/depthwise_kernel
I0601 05:39:09.583225 140233166403456 utils.py:108] Init efficientnet-b0/blocks_12/tpu_batch_normalization_1/gamma from ckpt var efficientnet-b0/blocks_12/tpu_batch_normalization_1/gamma
I0601 05:39:09.583319 140233166403456 utils.py:108] Init efficientnet-b0/blocks_12/tpu_batch_normalization_1/beta from ckpt var efficientnet-b0/blocks_12/tpu_batch_normalization_1/beta
I0601 05:39:09.583429 140233166403456 utils.py:108] Init efficientnet-b0/blocks_12/tpu_batch_normalization_1/moving_mean from ckpt var efficientnet-b0/blocks_12/tpu_batch_normalization_1/moving_mean
I0601 05:39:09.583520 140233166403456 utils.py:108] Init efficientnet-b0/blocks_12/tpu_batch_normalization_1/moving_variance from ckpt var efficientnet-b0/blocks_12/tpu_batch_normalization_1/moving_variance
I0601 05:39:09.583604 140233166403456 utils.py:108] Init efficientnet-b0/blocks_12/se/conv2d/kernel from ckpt var efficientnet-b0/blocks_12/se/conv2d/kernel
I0601 05:39:09.583690 140233166403456 utils.py:108] Init efficientnet-b0/blocks_12/se/conv2d/bias from ckpt var efficientnet-b0/blocks_12/se/conv2d/bias
I0601 05:39:09.583773 140233166403456 utils.py:108] Init efficientnet-b0/blocks_12/se/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_12/se/conv2d_1/kernel
I0601 05:39:09.583862 140233166403456 utils.py:108] Init efficientnet-b0/blocks_12/se/conv2d_1/bias from ckpt var efficientnet-b0/blocks_12/se/conv2d_1/bias
I0601 05:39:09.583948 140233166403456 utils.py:108] Init efficientnet-b0/blocks_12/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_12/conv2d_1/kernel
I0601 05:39:09.584049 140233166403456 utils.py:108] Init efficientnet-b0/blocks_12/tpu_batch_normalization_2/gamma from ckpt var efficientnet-b0/blocks_12/tpu_batch_normalization_2/gamma
I0601 05:39:09.584135 140233166403456 utils.py:108] Init efficientnet-b0/blocks_12/tpu_batch_normalization_2/beta from ckpt var efficientnet-b0/blocks_12/tpu_batch_normalization_2/beta
I0601 05:39:09.584223 140233166403456 utils.py:108] Init efficientnet-b0/blocks_12/tpu_batch_normalization_2/moving_mean from ckpt var efficientnet-b0/blocks_12/tpu_batch_normalization_2/moving_mean
I0601 05:39:09.585918 140233166403456 utils.py:108] Init efficientnet-b0/blocks_12/tpu_batch_normalization_2/moving_variance from ckpt var efficientnet-b0/blocks_12/tpu_batch_normalization_2/moving_variance
I0601 05:39:09.586033 140233166403456 utils.py:108] Init efficientnet-b0/blocks_13/conv2d/kernel from ckpt var efficientnet-b0/blocks_13/conv2d/kernel
I0601 05:39:09.586118 140233166403456 utils.py:108] Init efficientnet-b0/blocks_13/tpu_batch_normalization/gamma from ckpt var efficientnet-b0/blocks_13/tpu_batch_normalization/gamma
I0601 05:39:09.586199 140233166403456 utils.py:108] Init efficientnet-b0/blocks_13/tpu_batch_normalization/beta from ckpt var efficientnet-b0/blocks_13/tpu_batch_normalization/beta
I0601 05:39:09.586302 140233166403456 utils.py:108] Init efficientnet-b0/blocks_13/tpu_batch_normalization/moving_mean from ckpt var efficientnet-b0/blocks_13/tpu_batch_normalization/moving_mean
I0601 05:39:09.586386 140233166403456 utils.py:108] Init efficientnet-b0/blocks_13/tpu_batch_normalization/moving_variance from ckpt var efficientnet-b0/blocks_13/tpu_batch_normalization/moving_variance
I0601 05:39:09.586467 140233166403456 utils.py:108] Init efficientnet-b0/blocks_13/depthwise_conv2d/depthwise_kernel from ckpt var efficientnet-b0/blocks_13/depthwise_conv2d/depthwise_kernel
I0601 05:39:09.586549 140233166403456 utils.py:108] Init efficientnet-b0/blocks_13/tpu_batch_normalization_1/gamma from ckpt var efficientnet-b0/blocks_13/tpu_batch_normalization_1/gamma
I0601 05:39:09.586637 140233166403456 utils.py:108] Init efficientnet-b0/blocks_13/tpu_batch_normalization_1/beta from ckpt var efficientnet-b0/blocks_13/tpu_batch_normalization_1/beta
I0601 05:39:09.586729 140233166403456 utils.py:108] Init efficientnet-b0/blocks_13/tpu_batch_normalization_1/moving_mean from ckpt var efficientnet-b0/blocks_13/tpu_batch_normalization_1/moving_mean
I0601 05:39:09.586802 140233166403456 utils.py:108] Init efficientnet-b0/blocks_13/tpu_batch_normalization_1/moving_variance from ckpt var efficientnet-b0/blocks_13/tpu_batch_normalization_1/moving_variance
I0601 05:39:09.586876 140233166403456 utils.py:108] Init efficientnet-b0/blocks_13/se/conv2d/kernel from ckpt var efficientnet-b0/blocks_13/se/conv2d/kernel
I0601 05:39:09.586953 140233166403456 utils.py:108] Init efficientnet-b0/blocks_13/se/conv2d/bias from ckpt var efficientnet-b0/blocks_13/se/conv2d/bias
I0601 05:39:09.587029 140233166403456 utils.py:108] Init efficientnet-b0/blocks_13/se/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_13/se/conv2d_1/kernel
I0601 05:39:09.587103 140233166403456 utils.py:108] Init efficientnet-b0/blocks_13/se/conv2d_1/bias from ckpt var efficientnet-b0/blocks_13/se/conv2d_1/bias
I0601 05:39:09.587177 140233166403456 utils.py:108] Init efficientnet-b0/blocks_13/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_13/conv2d_1/kernel
I0601 05:39:09.587250 140233166403456 utils.py:108] Init efficientnet-b0/blocks_13/tpu_batch_normalization_2/gamma from ckpt var efficientnet-b0/blocks_13/tpu_batch_normalization_2/gamma
I0601 05:39:09.587341 140233166403456 utils.py:108] Init efficientnet-b0/blocks_13/tpu_batch_normalization_2/beta from ckpt var efficientnet-b0/blocks_13/tpu_batch_normalization_2/beta
I0601 05:39:09.587430 140233166403456 utils.py:108] Init efficientnet-b0/blocks_13/tpu_batch_normalization_2/moving_mean from ckpt var efficientnet-b0/blocks_13/tpu_batch_normalization_2/moving_mean
I0601 05:39:09.587511 140233166403456 utils.py:108] Init efficientnet-b0/blocks_13/tpu_batch_normalization_2/moving_variance from ckpt var efficientnet-b0/blocks_13/tpu_batch_normalization_2/moving_variance
I0601 05:39:09.587595 140233166403456 utils.py:108] Init efficientnet-b0/blocks_14/conv2d/kernel from ckpt var efficientnet-b0/blocks_14/conv2d/kernel
I0601 05:39:09.587682 140233166403456 utils.py:108] Init efficientnet-b0/blocks_14/tpu_batch_normalization/gamma from ckpt var efficientnet-b0/blocks_14/tpu_batch_normalization/gamma
I0601 05:39:09.587758 140233166403456 utils.py:108] Init efficientnet-b0/blocks_14/tpu_batch_normalization/beta from ckpt var efficientnet-b0/blocks_14/tpu_batch_normalization/beta
I0601 05:39:09.587831 140233166403456 utils.py:108] Init efficientnet-b0/blocks_14/tpu_batch_normalization/moving_mean from ckpt var efficientnet-b0/blocks_14/tpu_batch_normalization/moving_mean
I0601 05:39:09.587917 140233166403456 utils.py:108] Init efficientnet-b0/blocks_14/tpu_batch_normalization/moving_variance from ckpt var efficientnet-b0/blocks_14/tpu_batch_normalization/moving_variance
I0601 05:39:09.588001 140233166403456 utils.py:108] Init efficientnet-b0/blocks_14/depthwise_conv2d/depthwise_kernel from ckpt var efficientnet-b0/blocks_14/depthwise_conv2d/depthwise_kernel
I0601 05:39:09.588090 140233166403456 utils.py:108] Init efficientnet-b0/blocks_14/tpu_batch_normalization_1/gamma from ckpt var efficientnet-b0/blocks_14/tpu_batch_normalization_1/gamma
I0601 05:39:09.588164 140233166403456 utils.py:108] Init efficientnet-b0/blocks_14/tpu_batch_normalization_1/beta from ckpt var efficientnet-b0/blocks_14/tpu_batch_normalization_1/beta
I0601 05:39:09.588239 140233166403456 utils.py:108] Init efficientnet-b0/blocks_14/tpu_batch_normalization_1/moving_mean from ckpt var efficientnet-b0/blocks_14/tpu_batch_normalization_1/moving_mean
I0601 05:39:09.588338 140233166403456 utils.py:108] Init efficientnet-b0/blocks_14/tpu_batch_normalization_1/moving_variance from ckpt var efficientnet-b0/blocks_14/tpu_batch_normalization_1/moving_variance
I0601 05:39:09.588409 140233166403456 utils.py:108] Init efficientnet-b0/blocks_14/se/conv2d/kernel from ckpt var efficientnet-b0/blocks_14/se/conv2d/kernel
I0601 05:39:09.588497 140233166403456 utils.py:108] Init efficientnet-b0/blocks_14/se/conv2d/bias from ckpt var efficientnet-b0/blocks_14/se/conv2d/bias
I0601 05:39:09.588569 140233166403456 utils.py:108] Init efficientnet-b0/blocks_14/se/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_14/se/conv2d_1/kernel
I0601 05:39:09.588649 140233166403456 utils.py:108] Init efficientnet-b0/blocks_14/se/conv2d_1/bias from ckpt var efficientnet-b0/blocks_14/se/conv2d_1/bias
I0601 05:39:09.588721 140233166403456 utils.py:108] Init efficientnet-b0/blocks_14/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_14/conv2d_1/kernel
I0601 05:39:09.588803 140233166403456 utils.py:108] Init efficientnet-b0/blocks_14/tpu_batch_normalization_2/gamma from ckpt var efficientnet-b0/blocks_14/tpu_batch_normalization_2/gamma
I0601 05:39:09.588892 140233166403456 utils.py:108] Init efficientnet-b0/blocks_14/tpu_batch_normalization_2/beta from ckpt var efficientnet-b0/blocks_14/tpu_batch_normalization_2/beta
I0601 05:39:09.588983 140233166403456 utils.py:108] Init efficientnet-b0/blocks_14/tpu_batch_normalization_2/moving_mean from ckpt var efficientnet-b0/blocks_14/tpu_batch_normalization_2/moving_mean
I0601 05:39:09.589067 140233166403456 utils.py:108] Init efficientnet-b0/blocks_14/tpu_batch_normalization_2/moving_variance from ckpt var efficientnet-b0/blocks_14/tpu_batch_normalization_2/moving_variance
I0601 05:39:09.589147 140233166403456 utils.py:108] Init efficientnet-b0/blocks_15/conv2d/kernel from ckpt var efficientnet-b0/blocks_15/conv2d/kernel
I0601 05:39:09.589230 140233166403456 utils.py:108] Init efficientnet-b0/blocks_15/tpu_batch_normalization/gamma from ckpt var efficientnet-b0/blocks_15/tpu_batch_normalization/gamma
I0601 05:39:09.589327 140233166403456 utils.py:108] Init efficientnet-b0/blocks_15/tpu_batch_normalization/beta from ckpt var efficientnet-b0/blocks_15/tpu_batch_normalization/beta
I0601 05:39:09.589406 140233166403456 utils.py:108] Init efficientnet-b0/blocks_15/tpu_batch_normalization/moving_mean from ckpt var efficientnet-b0/blocks_15/tpu_batch_normalization/moving_mean
I0601 05:39:09.589492 140233166403456 utils.py:108] Init efficientnet-b0/blocks_15/tpu_batch_normalization/moving_variance from ckpt var efficientnet-b0/blocks_15/tpu_batch_normalization/moving_variance
I0601 05:39:09.680724 140233166403456 utils.py:108] Init efficientnet-b0/blocks_15/depthwise_conv2d/depthwise_kernel from ckpt var efficientnet-b0/blocks_15/depthwise_conv2d/depthwise_kernel
I0601 05:39:09.680886 140233166403456 utils.py:108] Init efficientnet-b0/blocks_15/tpu_batch_normalization_1/gamma from ckpt var efficientnet-b0/blocks_15/tpu_batch_normalization_1/gamma
I0601 05:39:09.681005 140233166403456 utils.py:108] Init efficientnet-b0/blocks_15/tpu_batch_normalization_1/beta from ckpt var efficientnet-b0/blocks_15/tpu_batch_normalization_1/beta
I0601 05:39:09.681105 140233166403456 utils.py:108] Init efficientnet-b0/blocks_15/tpu_batch_normalization_1/moving_mean from ckpt var efficientnet-b0/blocks_15/tpu_batch_normalization_1/moving_mean
I0601 05:39:09.681201 140233166403456 utils.py:108] Init efficientnet-b0/blocks_15/tpu_batch_normalization_1/moving_variance from ckpt var efficientnet-b0/blocks_15/tpu_batch_normalization_1/moving_variance
I0601 05:39:09.681341 140233166403456 utils.py:108] Init efficientnet-b0/blocks_15/se/conv2d/kernel from ckpt var efficientnet-b0/blocks_15/se/conv2d/kernel
I0601 05:39:09.681472 140233166403456 utils.py:108] Init efficientnet-b0/blocks_15/se/conv2d/bias from ckpt var efficientnet-b0/blocks_15/se/conv2d/bias
I0601 05:39:09.681543 140233166403456 utils.py:108] Init efficientnet-b0/blocks_15/se/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_15/se/conv2d_1/kernel
I0601 05:39:09.681612 140233166403456 utils.py:108] Init efficientnet-b0/blocks_15/se/conv2d_1/bias from ckpt var efficientnet-b0/blocks_15/se/conv2d_1/bias
I0601 05:39:09.681701 140233166403456 utils.py:108] Init efficientnet-b0/blocks_15/conv2d_1/kernel from ckpt var efficientnet-b0/blocks_15/conv2d_1/kernel
I0601 05:39:09.681772 140233166403456 utils.py:108] Init efficientnet-b0/blocks_15/tpu_batch_normalization_2/gamma from ckpt var efficientnet-b0/blocks_15/tpu_batch_normalization_2/gamma
I0601 05:39:09.681840 140233166403456 utils.py:108] Init efficientnet-b0/blocks_15/tpu_batch_normalization_2/beta from ckpt var efficientnet-b0/blocks_15/tpu_batch_normalization_2/beta
I0601 05:39:09.681908 140233166403456 utils.py:108] Init efficientnet-b0/blocks_15/tpu_batch_normalization_2/moving_mean from ckpt var efficientnet-b0/blocks_15/tpu_batch_normalization_2/moving_mean
I0601 05:39:09.681974 140233166403456 utils.py:108] Init efficientnet-b0/blocks_15/tpu_batch_normalization_2/moving_variance from ckpt var efficientnet-b0/blocks_15/tpu_batch_normalization_2/moving_variance
I0601 05:39:09.682042 140233166403456 utils.py:108] Init resample_p6/conv2d/kernel from ckpt var resample_p6/conv2d/kernel
I0601 05:39:09.682110 140233166403456 utils.py:108] Init resample_p6/conv2d/bias from ckpt var resample_p6/conv2d/bias
I0601 05:39:09.682184 140233166403456 utils.py:108] Init resample_p6/bn/gamma from ckpt var resample_p6/bn/gamma
I0601 05:39:09.682249 140233166403456 utils.py:108] Init resample_p6/bn/beta from ckpt var resample_p6/bn/beta
I0601 05:39:09.682337 140233166403456 utils.py:108] Init resample_p6/bn/moving_mean from ckpt var resample_p6/bn/moving_mean
I0601 05:39:09.682409 140233166403456 utils.py:108] Init resample_p6/bn/moving_variance from ckpt var resample_p6/bn/moving_variance
I0601 05:39:09.682477 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode0/WSM from ckpt var fpn_cells/cell_0/fnode0/WSM
I0601 05:39:09.682542 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode0/WSM_1 from ckpt var fpn_cells/cell_0/fnode0/WSM_1
I0601 05:39:09.682608 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode0/op_after_combine5/conv/depthwise_kernel from ckpt var fpn_cells/cell_0/fnode0/op_after_combine5/conv/depthwise_kernel
I0601 05:39:09.682686 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode0/op_after_combine5/conv/pointwise_kernel from ckpt var fpn_cells/cell_0/fnode0/op_after_combine5/conv/pointwise_kernel
I0601 05:39:09.682752 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode0/op_after_combine5/conv/bias from ckpt var fpn_cells/cell_0/fnode0/op_after_combine5/conv/bias
I0601 05:39:09.682821 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode0/op_after_combine5/bn/gamma from ckpt var fpn_cells/cell_0/fnode0/op_after_combine5/bn/gamma
I0601 05:39:09.682886 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode0/op_after_combine5/bn/beta from ckpt var fpn_cells/cell_0/fnode0/op_after_combine5/bn/beta
I0601 05:39:09.682978 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode0/op_after_combine5/bn/moving_mean from ckpt var fpn_cells/cell_0/fnode0/op_after_combine5/bn/moving_mean
I0601 05:39:09.683043 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode0/op_after_combine5/bn/moving_variance from ckpt var fpn_cells/cell_0/fnode0/op_after_combine5/bn/moving_variance
I0601 05:39:09.683118 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode1/resample_0_2_6/conv2d/kernel from ckpt var fpn_cells/cell_0/fnode1/resample_0_2_6/conv2d/kernel
I0601 05:39:09.683178 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode1/resample_0_2_6/conv2d/bias from ckpt var fpn_cells/cell_0/fnode1/resample_0_2_6/conv2d/bias
I0601 05:39:09.683238 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode1/resample_0_2_6/bn/gamma from ckpt var fpn_cells/cell_0/fnode1/resample_0_2_6/bn/gamma
I0601 05:39:09.683312 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode1/resample_0_2_6/bn/beta from ckpt var fpn_cells/cell_0/fnode1/resample_0_2_6/bn/beta
I0601 05:39:09.683375 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode1/resample_0_2_6/bn/moving_mean from ckpt var fpn_cells/cell_0/fnode1/resample_0_2_6/bn/moving_mean
I0601 05:39:09.683435 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode1/resample_0_2_6/bn/moving_variance from ckpt var fpn_cells/cell_0/fnode1/resample_0_2_6/bn/moving_variance
I0601 05:39:09.683495 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode1/WSM from ckpt var fpn_cells/cell_0/fnode1/WSM
I0601 05:39:09.683554 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode1/WSM_1 from ckpt var fpn_cells/cell_0/fnode1/WSM_1
I0601 05:39:09.683614 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode1/op_after_combine6/conv/depthwise_kernel from ckpt var fpn_cells/cell_0/fnode1/op_after_combine6/conv/depthwise_kernel
I0601 05:39:09.683681 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode1/op_after_combine6/conv/pointwise_kernel from ckpt var fpn_cells/cell_0/fnode1/op_after_combine6/conv/pointwise_kernel
I0601 05:39:09.683743 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode1/op_after_combine6/conv/bias from ckpt var fpn_cells/cell_0/fnode1/op_after_combine6/conv/bias
I0601 05:39:09.683811 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode1/op_after_combine6/bn/gamma from ckpt var fpn_cells/cell_0/fnode1/op_after_combine6/bn/gamma
I0601 05:39:09.683873 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode1/op_after_combine6/bn/beta from ckpt var fpn_cells/cell_0/fnode1/op_after_combine6/bn/beta
I0601 05:39:09.683935 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode1/op_after_combine6/bn/moving_mean from ckpt var fpn_cells/cell_0/fnode1/op_after_combine6/bn/moving_mean
I0601 05:39:09.684015 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode1/op_after_combine6/bn/moving_variance from ckpt var fpn_cells/cell_0/fnode1/op_after_combine6/bn/moving_variance
I0601 05:39:09.684082 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode2/resample_0_1_7/conv2d/kernel from ckpt var fpn_cells/cell_0/fnode2/resample_0_1_7/conv2d/kernel
I0601 05:39:09.684148 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode2/resample_0_1_7/conv2d/bias from ckpt var fpn_cells/cell_0/fnode2/resample_0_1_7/conv2d/bias
I0601 05:39:09.684214 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode2/resample_0_1_7/bn/gamma from ckpt var fpn_cells/cell_0/fnode2/resample_0_1_7/bn/gamma
I0601 05:39:09.684289 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode2/resample_0_1_7/bn/beta from ckpt var fpn_cells/cell_0/fnode2/resample_0_1_7/bn/beta
I0601 05:39:09.684356 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode2/resample_0_1_7/bn/moving_mean from ckpt var fpn_cells/cell_0/fnode2/resample_0_1_7/bn/moving_mean
I0601 05:39:09.684422 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode2/resample_0_1_7/bn/moving_variance from ckpt var fpn_cells/cell_0/fnode2/resample_0_1_7/bn/moving_variance
I0601 05:39:09.684489 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode2/WSM from ckpt var fpn_cells/cell_0/fnode2/WSM
I0601 05:39:09.684562 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode2/WSM_1 from ckpt var fpn_cells/cell_0/fnode2/WSM_1
I0601 05:39:09.684633 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode2/op_after_combine7/conv/depthwise_kernel from ckpt var fpn_cells/cell_0/fnode2/op_after_combine7/conv/depthwise_kernel
I0601 05:39:09.684700 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode2/op_after_combine7/conv/pointwise_kernel from ckpt var fpn_cells/cell_0/fnode2/op_after_combine7/conv/pointwise_kernel
I0601 05:39:09.684767 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode2/op_after_combine7/conv/bias from ckpt var fpn_cells/cell_0/fnode2/op_after_combine7/conv/bias
I0601 05:39:09.684833 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode2/op_after_combine7/bn/gamma from ckpt var fpn_cells/cell_0/fnode2/op_after_combine7/bn/gamma
I0601 05:39:09.684899 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode2/op_after_combine7/bn/beta from ckpt var fpn_cells/cell_0/fnode2/op_after_combine7/bn/beta
I0601 05:39:09.684964 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode2/op_after_combine7/bn/moving_mean from ckpt var fpn_cells/cell_0/fnode2/op_after_combine7/bn/moving_mean
I0601 05:39:09.685030 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode2/op_after_combine7/bn/moving_variance from ckpt var fpn_cells/cell_0/fnode2/op_after_combine7/bn/moving_variance
I0601 05:39:09.685096 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode3/resample_0_0_8/conv2d/kernel from ckpt var fpn_cells/cell_0/fnode3/resample_0_0_8/conv2d/kernel
I0601 05:39:09.685162 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode3/resample_0_0_8/conv2d/bias from ckpt var fpn_cells/cell_0/fnode3/resample_0_0_8/conv2d/bias
I0601 05:39:09.685228 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode3/resample_0_0_8/bn/gamma from ckpt var fpn_cells/cell_0/fnode3/resample_0_0_8/bn/gamma
I0601 05:39:09.685305 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode3/resample_0_0_8/bn/beta from ckpt var fpn_cells/cell_0/fnode3/resample_0_0_8/bn/beta
I0601 05:39:09.685372 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode3/resample_0_0_8/bn/moving_mean from ckpt var fpn_cells/cell_0/fnode3/resample_0_0_8/bn/moving_mean
I0601 05:39:09.685436 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode3/resample_0_0_8/bn/moving_variance from ckpt var fpn_cells/cell_0/fnode3/resample_0_0_8/bn/moving_variance
I0601 05:39:09.685500 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode3/WSM from ckpt var fpn_cells/cell_0/fnode3/WSM
I0601 05:39:09.685568 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode3/WSM_1 from ckpt var fpn_cells/cell_0/fnode3/WSM_1
I0601 05:39:09.685638 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode3/op_after_combine8/conv/depthwise_kernel from ckpt var fpn_cells/cell_0/fnode3/op_after_combine8/conv/depthwise_kernel
I0601 05:39:09.685704 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode3/op_after_combine8/conv/pointwise_kernel from ckpt var fpn_cells/cell_0/fnode3/op_after_combine8/conv/pointwise_kernel
I0601 05:39:09.685769 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode3/op_after_combine8/conv/bias from ckpt var fpn_cells/cell_0/fnode3/op_after_combine8/conv/bias
I0601 05:39:09.685832 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode3/op_after_combine8/bn/gamma from ckpt var fpn_cells/cell_0/fnode3/op_after_combine8/bn/gamma
I0601 05:39:09.685894 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode3/op_after_combine8/bn/beta from ckpt var fpn_cells/cell_0/fnode3/op_after_combine8/bn/beta
I0601 05:39:09.685956 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode3/op_after_combine8/bn/moving_mean from ckpt var fpn_cells/cell_0/fnode3/op_after_combine8/bn/moving_mean
I0601 05:39:09.686019 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode3/op_after_combine8/bn/moving_variance from ckpt var fpn_cells/cell_0/fnode3/op_after_combine8/bn/moving_variance
I0601 05:39:09.686087 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode4/resample_0_1_9/conv2d/kernel from ckpt var fpn_cells/cell_0/fnode4/resample_0_1_9/conv2d/kernel
I0601 05:39:09.686151 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode4/resample_0_1_9/conv2d/bias from ckpt var fpn_cells/cell_0/fnode4/resample_0_1_9/conv2d/bias
I0601 05:39:09.686214 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode4/resample_0_1_9/bn/gamma from ckpt var fpn_cells/cell_0/fnode4/resample_0_1_9/bn/gamma
I0601 05:39:09.686291 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode4/resample_0_1_9/bn/beta from ckpt var fpn_cells/cell_0/fnode4/resample_0_1_9/bn/beta
I0601 05:39:09.686357 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode4/resample_0_1_9/bn/moving_mean from ckpt var fpn_cells/cell_0/fnode4/resample_0_1_9/bn/moving_mean
I0601 05:39:09.686423 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode4/resample_0_1_9/bn/moving_variance from ckpt var fpn_cells/cell_0/fnode4/resample_0_1_9/bn/moving_variance
I0601 05:39:09.686486 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode4/WSM from ckpt var fpn_cells/cell_0/fnode4/WSM
I0601 05:39:09.686549 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode4/WSM_1 from ckpt var fpn_cells/cell_0/fnode4/WSM_1
I0601 05:39:09.686622 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode4/WSM_2 from ckpt var fpn_cells/cell_0/fnode4/WSM_2
I0601 05:39:09.686692 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode4/op_after_combine9/conv/depthwise_kernel from ckpt var fpn_cells/cell_0/fnode4/op_after_combine9/conv/depthwise_kernel
I0601 05:39:09.686753 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode4/op_after_combine9/conv/pointwise_kernel from ckpt var fpn_cells/cell_0/fnode4/op_after_combine9/conv/pointwise_kernel
I0601 05:39:09.686812 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode4/op_after_combine9/conv/bias from ckpt var fpn_cells/cell_0/fnode4/op_after_combine9/conv/bias
I0601 05:39:09.686872 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode4/op_after_combine9/bn/gamma from ckpt var fpn_cells/cell_0/fnode4/op_after_combine9/bn/gamma
I0601 05:39:09.686930 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode4/op_after_combine9/bn/beta from ckpt var fpn_cells/cell_0/fnode4/op_after_combine9/bn/beta
I0601 05:39:09.686988 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode4/op_after_combine9/bn/moving_mean from ckpt var fpn_cells/cell_0/fnode4/op_after_combine9/bn/moving_mean
I0601 05:39:09.687047 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode4/op_after_combine9/bn/moving_variance from ckpt var fpn_cells/cell_0/fnode4/op_after_combine9/bn/moving_variance
I0601 05:39:09.687106 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode5/resample_0_2_10/conv2d/kernel from ckpt var fpn_cells/cell_0/fnode5/resample_0_2_10/conv2d/kernel
I0601 05:39:09.687166 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode5/resample_0_2_10/conv2d/bias from ckpt var fpn_cells/cell_0/fnode5/resample_0_2_10/conv2d/bias
I0601 05:39:09.687225 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode5/resample_0_2_10/bn/gamma from ckpt var fpn_cells/cell_0/fnode5/resample_0_2_10/bn/gamma
I0601 05:39:09.687295 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode5/resample_0_2_10/bn/beta from ckpt var fpn_cells/cell_0/fnode5/resample_0_2_10/bn/beta
I0601 05:39:09.687374 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode5/resample_0_2_10/bn/moving_mean from ckpt var fpn_cells/cell_0/fnode5/resample_0_2_10/bn/moving_mean
I0601 05:39:09.687442 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode5/resample_0_2_10/bn/moving_variance from ckpt var fpn_cells/cell_0/fnode5/resample_0_2_10/bn/moving_variance
I0601 05:39:09.687508 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode5/WSM from ckpt var fpn_cells/cell_0/fnode5/WSM
I0601 05:39:09.687572 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode5/WSM_1 from ckpt var fpn_cells/cell_0/fnode5/WSM_1
I0601 05:39:09.687639 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode5/WSM_2 from ckpt var fpn_cells/cell_0/fnode5/WSM_2
I0601 05:39:09.687706 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode5/op_after_combine10/conv/depthwise_kernel from ckpt var fpn_cells/cell_0/fnode5/op_after_combine10/conv/depthwise_kernel
I0601 05:39:09.687773 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode5/op_after_combine10/conv/pointwise_kernel from ckpt var fpn_cells/cell_0/fnode5/op_after_combine10/conv/pointwise_kernel
I0601 05:39:09.687840 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode5/op_after_combine10/conv/bias from ckpt var fpn_cells/cell_0/fnode5/op_after_combine10/conv/bias
I0601 05:39:09.781375 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode5/op_after_combine10/bn/gamma from ckpt var fpn_cells/cell_0/fnode5/op_after_combine10/bn/gamma
I0601 05:39:09.781520 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode5/op_after_combine10/bn/beta from ckpt var fpn_cells/cell_0/fnode5/op_after_combine10/bn/beta
I0601 05:39:09.781668 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode5/op_after_combine10/bn/moving_mean from ckpt var fpn_cells/cell_0/fnode5/op_after_combine10/bn/moving_mean
I0601 05:39:09.781780 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode5/op_after_combine10/bn/moving_variance from ckpt var fpn_cells/cell_0/fnode5/op_after_combine10/bn/moving_variance
I0601 05:39:09.781880 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode6/WSM from ckpt var fpn_cells/cell_0/fnode6/WSM
I0601 05:39:09.781976 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode6/WSM_1 from ckpt var fpn_cells/cell_0/fnode6/WSM_1
I0601 05:39:09.782070 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode6/WSM_2 from ckpt var fpn_cells/cell_0/fnode6/WSM_2
I0601 05:39:09.782170 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode6/op_after_combine11/conv/depthwise_kernel from ckpt var fpn_cells/cell_0/fnode6/op_after_combine11/conv/depthwise_kernel
I0601 05:39:09.782253 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode6/op_after_combine11/conv/pointwise_kernel from ckpt var fpn_cells/cell_0/fnode6/op_after_combine11/conv/pointwise_kernel
I0601 05:39:09.782346 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode6/op_after_combine11/conv/bias from ckpt var fpn_cells/cell_0/fnode6/op_after_combine11/conv/bias
I0601 05:39:09.782412 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode6/op_after_combine11/bn/gamma from ckpt var fpn_cells/cell_0/fnode6/op_after_combine11/bn/gamma
I0601 05:39:09.782474 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode6/op_after_combine11/bn/beta from ckpt var fpn_cells/cell_0/fnode6/op_after_combine11/bn/beta
I0601 05:39:09.782537 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode6/op_after_combine11/bn/moving_mean from ckpt var fpn_cells/cell_0/fnode6/op_after_combine11/bn/moving_mean
I0601 05:39:09.782600 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode6/op_after_combine11/bn/moving_variance from ckpt var fpn_cells/cell_0/fnode6/op_after_combine11/bn/moving_variance
I0601 05:39:09.782676 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode7/WSM from ckpt var fpn_cells/cell_0/fnode7/WSM
I0601 05:39:09.782739 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode7/WSM_1 from ckpt var fpn_cells/cell_0/fnode7/WSM_1
I0601 05:39:09.782803 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode7/op_after_combine12/conv/depthwise_kernel from ckpt var fpn_cells/cell_0/fnode7/op_after_combine12/conv/depthwise_kernel
I0601 05:39:09.782864 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode7/op_after_combine12/conv/pointwise_kernel from ckpt var fpn_cells/cell_0/fnode7/op_after_combine12/conv/pointwise_kernel
I0601 05:39:09.782925 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode7/op_after_combine12/conv/bias from ckpt var fpn_cells/cell_0/fnode7/op_after_combine12/conv/bias
I0601 05:39:09.782986 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode7/op_after_combine12/bn/gamma from ckpt var fpn_cells/cell_0/fnode7/op_after_combine12/bn/gamma
I0601 05:39:09.783047 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode7/op_after_combine12/bn/beta from ckpt var fpn_cells/cell_0/fnode7/op_after_combine12/bn/beta
I0601 05:39:09.783108 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode7/op_after_combine12/bn/moving_mean from ckpt var fpn_cells/cell_0/fnode7/op_after_combine12/bn/moving_mean
I0601 05:39:09.783169 140233166403456 utils.py:108] Init fpn_cells/cell_0/fnode7/op_after_combine12/bn/moving_variance from ckpt var fpn_cells/cell_0/fnode7/op_after_combine12/bn/moving_variance
I0601 05:39:09.783231 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode0/WSM from ckpt var fpn_cells/cell_1/fnode0/WSM
I0601 05:39:09.783308 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode0/WSM_1 from ckpt var fpn_cells/cell_1/fnode0/WSM_1
I0601 05:39:09.783375 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode0/op_after_combine5/conv/depthwise_kernel from ckpt var fpn_cells/cell_1/fnode0/op_after_combine5/conv/depthwise_kernel
I0601 05:39:09.783438 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode0/op_after_combine5/conv/pointwise_kernel from ckpt var fpn_cells/cell_1/fnode0/op_after_combine5/conv/pointwise_kernel
I0601 05:39:09.783498 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode0/op_after_combine5/conv/bias from ckpt var fpn_cells/cell_1/fnode0/op_after_combine5/conv/bias
I0601 05:39:09.783558 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode0/op_after_combine5/bn/gamma from ckpt var fpn_cells/cell_1/fnode0/op_after_combine5/bn/gamma
I0601 05:39:09.783617 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode0/op_after_combine5/bn/beta from ckpt var fpn_cells/cell_1/fnode0/op_after_combine5/bn/beta
I0601 05:39:09.783686 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode0/op_after_combine5/bn/moving_mean from ckpt var fpn_cells/cell_1/fnode0/op_after_combine5/bn/moving_mean
I0601 05:39:09.783748 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode0/op_after_combine5/bn/moving_variance from ckpt var fpn_cells/cell_1/fnode0/op_after_combine5/bn/moving_variance
I0601 05:39:09.783813 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode1/WSM from ckpt var fpn_cells/cell_1/fnode1/WSM
I0601 05:39:09.783874 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode1/WSM_1 from ckpt var fpn_cells/cell_1/fnode1/WSM_1
I0601 05:39:09.783934 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode1/op_after_combine6/conv/depthwise_kernel from ckpt var fpn_cells/cell_1/fnode1/op_after_combine6/conv/depthwise_kernel
I0601 05:39:09.783993 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode1/op_after_combine6/conv/pointwise_kernel from ckpt var fpn_cells/cell_1/fnode1/op_after_combine6/conv/pointwise_kernel
I0601 05:39:09.784075 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode1/op_after_combine6/conv/bias from ckpt var fpn_cells/cell_1/fnode1/op_after_combine6/conv/bias
I0601 05:39:09.784140 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode1/op_after_combine6/bn/gamma from ckpt var fpn_cells/cell_1/fnode1/op_after_combine6/bn/gamma
I0601 05:39:09.784207 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode1/op_after_combine6/bn/beta from ckpt var fpn_cells/cell_1/fnode1/op_after_combine6/bn/beta
I0601 05:39:09.784284 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode1/op_after_combine6/bn/moving_mean from ckpt var fpn_cells/cell_1/fnode1/op_after_combine6/bn/moving_mean
I0601 05:39:09.784351 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode1/op_after_combine6/bn/moving_variance from ckpt var fpn_cells/cell_1/fnode1/op_after_combine6/bn/moving_variance
I0601 05:39:09.784428 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode2/WSM from ckpt var fpn_cells/cell_1/fnode2/WSM
I0601 05:39:09.784506 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode2/WSM_1 from ckpt var fpn_cells/cell_1/fnode2/WSM_1
I0601 05:39:09.784582 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode2/op_after_combine7/conv/depthwise_kernel from ckpt var fpn_cells/cell_1/fnode2/op_after_combine7/conv/depthwise_kernel
I0601 05:39:09.784648 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode2/op_after_combine7/conv/pointwise_kernel from ckpt var fpn_cells/cell_1/fnode2/op_after_combine7/conv/pointwise_kernel
I0601 05:39:09.784710 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode2/op_after_combine7/conv/bias from ckpt var fpn_cells/cell_1/fnode2/op_after_combine7/conv/bias
I0601 05:39:09.784774 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode2/op_after_combine7/bn/gamma from ckpt var fpn_cells/cell_1/fnode2/op_after_combine7/bn/gamma
I0601 05:39:09.784835 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode2/op_after_combine7/bn/beta from ckpt var fpn_cells/cell_1/fnode2/op_after_combine7/bn/beta
I0601 05:39:09.784895 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode2/op_after_combine7/bn/moving_mean from ckpt var fpn_cells/cell_1/fnode2/op_after_combine7/bn/moving_mean
I0601 05:39:09.784955 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode2/op_after_combine7/bn/moving_variance from ckpt var fpn_cells/cell_1/fnode2/op_after_combine7/bn/moving_variance
I0601 05:39:09.785014 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode3/WSM from ckpt var fpn_cells/cell_1/fnode3/WSM
I0601 05:39:09.785072 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode3/WSM_1 from ckpt var fpn_cells/cell_1/fnode3/WSM_1
I0601 05:39:09.785131 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode3/op_after_combine8/conv/depthwise_kernel from ckpt var fpn_cells/cell_1/fnode3/op_after_combine8/conv/depthwise_kernel
I0601 05:39:09.785192 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode3/op_after_combine8/conv/pointwise_kernel from ckpt var fpn_cells/cell_1/fnode3/op_after_combine8/conv/pointwise_kernel
I0601 05:39:09.785251 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode3/op_after_combine8/conv/bias from ckpt var fpn_cells/cell_1/fnode3/op_after_combine8/conv/bias
I0601 05:39:09.785324 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode3/op_after_combine8/bn/gamma from ckpt var fpn_cells/cell_1/fnode3/op_after_combine8/bn/gamma
I0601 05:39:09.785387 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode3/op_after_combine8/bn/beta from ckpt var fpn_cells/cell_1/fnode3/op_after_combine8/bn/beta
I0601 05:39:09.785446 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode3/op_after_combine8/bn/moving_mean from ckpt var fpn_cells/cell_1/fnode3/op_after_combine8/bn/moving_mean
I0601 05:39:09.785504 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode3/op_after_combine8/bn/moving_variance from ckpt var fpn_cells/cell_1/fnode3/op_after_combine8/bn/moving_variance
I0601 05:39:09.785562 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode4/WSM from ckpt var fpn_cells/cell_1/fnode4/WSM
I0601 05:39:09.785621 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode4/WSM_1 from ckpt var fpn_cells/cell_1/fnode4/WSM_1
I0601 05:39:09.785688 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode4/WSM_2 from ckpt var fpn_cells/cell_1/fnode4/WSM_2
I0601 05:39:09.785750 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode4/op_after_combine9/conv/depthwise_kernel from ckpt var fpn_cells/cell_1/fnode4/op_after_combine9/conv/depthwise_kernel
I0601 05:39:09.785813 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode4/op_after_combine9/conv/pointwise_kernel from ckpt var fpn_cells/cell_1/fnode4/op_after_combine9/conv/pointwise_kernel
I0601 05:39:09.785874 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode4/op_after_combine9/conv/bias from ckpt var fpn_cells/cell_1/fnode4/op_after_combine9/conv/bias
I0601 05:39:09.785933 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode4/op_after_combine9/bn/gamma from ckpt var fpn_cells/cell_1/fnode4/op_after_combine9/bn/gamma
I0601 05:39:09.785992 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode4/op_after_combine9/bn/beta from ckpt var fpn_cells/cell_1/fnode4/op_after_combine9/bn/beta
I0601 05:39:09.786051 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode4/op_after_combine9/bn/moving_mean from ckpt var fpn_cells/cell_1/fnode4/op_after_combine9/bn/moving_mean
I0601 05:39:09.786109 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode4/op_after_combine9/bn/moving_variance from ckpt var fpn_cells/cell_1/fnode4/op_after_combine9/bn/moving_variance
I0601 05:39:09.786168 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode5/WSM from ckpt var fpn_cells/cell_1/fnode5/WSM
I0601 05:39:09.786227 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode5/WSM_1 from ckpt var fpn_cells/cell_1/fnode5/WSM_1
I0601 05:39:09.786297 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode5/WSM_2 from ckpt var fpn_cells/cell_1/fnode5/WSM_2
I0601 05:39:09.786358 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode5/op_after_combine10/conv/depthwise_kernel from ckpt var fpn_cells/cell_1/fnode5/op_after_combine10/conv/depthwise_kernel
I0601 05:39:09.786420 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode5/op_after_combine10/conv/pointwise_kernel from ckpt var fpn_cells/cell_1/fnode5/op_after_combine10/conv/pointwise_kernel
I0601 05:39:09.786479 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode5/op_after_combine10/conv/bias from ckpt var fpn_cells/cell_1/fnode5/op_after_combine10/conv/bias
I0601 05:39:09.786538 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode5/op_after_combine10/bn/gamma from ckpt var fpn_cells/cell_1/fnode5/op_after_combine10/bn/gamma
I0601 05:39:09.786597 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode5/op_after_combine10/bn/beta from ckpt var fpn_cells/cell_1/fnode5/op_after_combine10/bn/beta
I0601 05:39:09.786663 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode5/op_after_combine10/bn/moving_mean from ckpt var fpn_cells/cell_1/fnode5/op_after_combine10/bn/moving_mean
I0601 05:39:09.786725 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode5/op_after_combine10/bn/moving_variance from ckpt var fpn_cells/cell_1/fnode5/op_after_combine10/bn/moving_variance
I0601 05:39:09.786785 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode6/WSM from ckpt var fpn_cells/cell_1/fnode6/WSM
I0601 05:39:09.786845 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode6/WSM_1 from ckpt var fpn_cells/cell_1/fnode6/WSM_1
I0601 05:39:09.786904 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode6/WSM_2 from ckpt var fpn_cells/cell_1/fnode6/WSM_2
I0601 05:39:09.786964 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode6/op_after_combine11/conv/depthwise_kernel from ckpt var fpn_cells/cell_1/fnode6/op_after_combine11/conv/depthwise_kernel
I0601 05:39:09.787023 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode6/op_after_combine11/conv/pointwise_kernel from ckpt var fpn_cells/cell_1/fnode6/op_after_combine11/conv/pointwise_kernel
I0601 05:39:09.787081 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode6/op_after_combine11/conv/bias from ckpt var fpn_cells/cell_1/fnode6/op_after_combine11/conv/bias
I0601 05:39:09.787141 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode6/op_after_combine11/bn/gamma from ckpt var fpn_cells/cell_1/fnode6/op_after_combine11/bn/gamma
I0601 05:39:09.787205 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode6/op_after_combine11/bn/beta from ckpt var fpn_cells/cell_1/fnode6/op_after_combine11/bn/beta
I0601 05:39:09.787276 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode6/op_after_combine11/bn/moving_mean from ckpt var fpn_cells/cell_1/fnode6/op_after_combine11/bn/moving_mean
I0601 05:39:09.787339 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode6/op_after_combine11/bn/moving_variance from ckpt var fpn_cells/cell_1/fnode6/op_after_combine11/bn/moving_variance
I0601 05:39:09.787401 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode7/WSM from ckpt var fpn_cells/cell_1/fnode7/WSM
I0601 05:39:09.787460 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode7/WSM_1 from ckpt var fpn_cells/cell_1/fnode7/WSM_1
I0601 05:39:09.787520 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode7/op_after_combine12/conv/depthwise_kernel from ckpt var fpn_cells/cell_1/fnode7/op_after_combine12/conv/depthwise_kernel
I0601 05:39:09.787579 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode7/op_after_combine12/conv/pointwise_kernel from ckpt var fpn_cells/cell_1/fnode7/op_after_combine12/conv/pointwise_kernel
I0601 05:39:09.787642 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode7/op_after_combine12/conv/bias from ckpt var fpn_cells/cell_1/fnode7/op_after_combine12/conv/bias
I0601 05:39:09.787704 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode7/op_after_combine12/bn/gamma from ckpt var fpn_cells/cell_1/fnode7/op_after_combine12/bn/gamma
I0601 05:39:09.787765 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode7/op_after_combine12/bn/beta from ckpt var fpn_cells/cell_1/fnode7/op_after_combine12/bn/beta
I0601 05:39:09.787825 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode7/op_after_combine12/bn/moving_mean from ckpt var fpn_cells/cell_1/fnode7/op_after_combine12/bn/moving_mean
I0601 05:39:09.787887 140233166403456 utils.py:108] Init fpn_cells/cell_1/fnode7/op_after_combine12/bn/moving_variance from ckpt var fpn_cells/cell_1/fnode7/op_after_combine12/bn/moving_variance
I0601 05:39:09.787950 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode0/WSM from ckpt var fpn_cells/cell_2/fnode0/WSM
I0601 05:39:09.788008 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode0/WSM_1 from ckpt var fpn_cells/cell_2/fnode0/WSM_1
I0601 05:39:09.788068 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode0/op_after_combine5/conv/depthwise_kernel from ckpt var fpn_cells/cell_2/fnode0/op_after_combine5/conv/depthwise_kernel
I0601 05:39:09.883184 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode0/op_after_combine5/conv/pointwise_kernel from ckpt var fpn_cells/cell_2/fnode0/op_after_combine5/conv/pointwise_kernel
I0601 05:39:09.883306 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode0/op_after_combine5/conv/bias from ckpt var fpn_cells/cell_2/fnode0/op_after_combine5/conv/bias
I0601 05:39:09.883397 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode0/op_after_combine5/bn/gamma from ckpt var fpn_cells/cell_2/fnode0/op_after_combine5/bn/gamma
I0601 05:39:09.883474 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode0/op_after_combine5/bn/beta from ckpt var fpn_cells/cell_2/fnode0/op_after_combine5/bn/beta
I0601 05:39:09.883541 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode0/op_after_combine5/bn/moving_mean from ckpt var fpn_cells/cell_2/fnode0/op_after_combine5/bn/moving_mean
I0601 05:39:09.883614 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode0/op_after_combine5/bn/moving_variance from ckpt var fpn_cells/cell_2/fnode0/op_after_combine5/bn/moving_variance
I0601 05:39:09.883682 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode1/WSM from ckpt var fpn_cells/cell_2/fnode1/WSM
I0601 05:39:09.883745 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode1/WSM_1 from ckpt var fpn_cells/cell_2/fnode1/WSM_1
I0601 05:39:09.883815 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode1/op_after_combine6/conv/depthwise_kernel from ckpt var fpn_cells/cell_2/fnode1/op_after_combine6/conv/depthwise_kernel
I0601 05:39:09.883885 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode1/op_after_combine6/conv/pointwise_kernel from ckpt var fpn_cells/cell_2/fnode1/op_after_combine6/conv/pointwise_kernel
I0601 05:39:09.883952 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode1/op_after_combine6/conv/bias from ckpt var fpn_cells/cell_2/fnode1/op_after_combine6/conv/bias
I0601 05:39:09.884037 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode1/op_after_combine6/bn/gamma from ckpt var fpn_cells/cell_2/fnode1/op_after_combine6/bn/gamma
I0601 05:39:09.884108 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode1/op_after_combine6/bn/beta from ckpt var fpn_cells/cell_2/fnode1/op_after_combine6/bn/beta
I0601 05:39:09.884201 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode1/op_after_combine6/bn/moving_mean from ckpt var fpn_cells/cell_2/fnode1/op_after_combine6/bn/moving_mean
I0601 05:39:09.884292 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode1/op_after_combine6/bn/moving_variance from ckpt var fpn_cells/cell_2/fnode1/op_after_combine6/bn/moving_variance
I0601 05:39:09.884384 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode2/WSM from ckpt var fpn_cells/cell_2/fnode2/WSM
I0601 05:39:09.884450 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode2/WSM_1 from ckpt var fpn_cells/cell_2/fnode2/WSM_1
I0601 05:39:09.884507 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode2/op_after_combine7/conv/depthwise_kernel from ckpt var fpn_cells/cell_2/fnode2/op_after_combine7/conv/depthwise_kernel
I0601 05:39:09.884561 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode2/op_after_combine7/conv/pointwise_kernel from ckpt var fpn_cells/cell_2/fnode2/op_after_combine7/conv/pointwise_kernel
I0601 05:39:09.884613 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode2/op_after_combine7/conv/bias from ckpt var fpn_cells/cell_2/fnode2/op_after_combine7/conv/bias
I0601 05:39:09.884668 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode2/op_after_combine7/bn/gamma from ckpt var fpn_cells/cell_2/fnode2/op_after_combine7/bn/gamma
I0601 05:39:09.884725 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode2/op_after_combine7/bn/beta from ckpt var fpn_cells/cell_2/fnode2/op_after_combine7/bn/beta
I0601 05:39:09.884793 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode2/op_after_combine7/bn/moving_mean from ckpt var fpn_cells/cell_2/fnode2/op_after_combine7/bn/moving_mean
I0601 05:39:09.884858 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode2/op_after_combine7/bn/moving_variance from ckpt var fpn_cells/cell_2/fnode2/op_after_combine7/bn/moving_variance
I0601 05:39:09.884924 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode3/WSM from ckpt var fpn_cells/cell_2/fnode3/WSM
I0601 05:39:09.884983 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode3/WSM_1 from ckpt var fpn_cells/cell_2/fnode3/WSM_1
I0601 05:39:09.885045 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode3/op_after_combine8/conv/depthwise_kernel from ckpt var fpn_cells/cell_2/fnode3/op_after_combine8/conv/depthwise_kernel
I0601 05:39:09.885109 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode3/op_after_combine8/conv/pointwise_kernel from ckpt var fpn_cells/cell_2/fnode3/op_after_combine8/conv/pointwise_kernel
I0601 05:39:09.885179 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode3/op_after_combine8/conv/bias from ckpt var fpn_cells/cell_2/fnode3/op_after_combine8/conv/bias
I0601 05:39:09.885231 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode3/op_after_combine8/bn/gamma from ckpt var fpn_cells/cell_2/fnode3/op_after_combine8/bn/gamma
I0601 05:39:09.885309 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode3/op_after_combine8/bn/beta from ckpt var fpn_cells/cell_2/fnode3/op_after_combine8/bn/beta
I0601 05:39:09.885375 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode3/op_after_combine8/bn/moving_mean from ckpt var fpn_cells/cell_2/fnode3/op_after_combine8/bn/moving_mean
I0601 05:39:09.885443 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode3/op_after_combine8/bn/moving_variance from ckpt var fpn_cells/cell_2/fnode3/op_after_combine8/bn/moving_variance
I0601 05:39:09.885506 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode4/WSM from ckpt var fpn_cells/cell_2/fnode4/WSM
I0601 05:39:09.885568 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode4/WSM_1 from ckpt var fpn_cells/cell_2/fnode4/WSM_1
I0601 05:39:09.885658 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode4/WSM_2 from ckpt var fpn_cells/cell_2/fnode4/WSM_2
I0601 05:39:09.885725 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode4/op_after_combine9/conv/depthwise_kernel from ckpt var fpn_cells/cell_2/fnode4/op_after_combine9/conv/depthwise_kernel
I0601 05:39:09.885786 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode4/op_after_combine9/conv/pointwise_kernel from ckpt var fpn_cells/cell_2/fnode4/op_after_combine9/conv/pointwise_kernel
I0601 05:39:09.885845 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode4/op_after_combine9/conv/bias from ckpt var fpn_cells/cell_2/fnode4/op_after_combine9/conv/bias
I0601 05:39:09.885913 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode4/op_after_combine9/bn/gamma from ckpt var fpn_cells/cell_2/fnode4/op_after_combine9/bn/gamma
I0601 05:39:09.885987 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode4/op_after_combine9/bn/beta from ckpt var fpn_cells/cell_2/fnode4/op_after_combine9/bn/beta
I0601 05:39:09.886046 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode4/op_after_combine9/bn/moving_mean from ckpt var fpn_cells/cell_2/fnode4/op_after_combine9/bn/moving_mean
I0601 05:39:09.886107 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode4/op_after_combine9/bn/moving_variance from ckpt var fpn_cells/cell_2/fnode4/op_after_combine9/bn/moving_variance
I0601 05:39:09.886177 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode5/WSM from ckpt var fpn_cells/cell_2/fnode5/WSM
I0601 05:39:09.886237 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode5/WSM_1 from ckpt var fpn_cells/cell_2/fnode5/WSM_1
I0601 05:39:09.886321 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode5/WSM_2 from ckpt var fpn_cells/cell_2/fnode5/WSM_2
I0601 05:39:09.886381 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode5/op_after_combine10/conv/depthwise_kernel from ckpt var fpn_cells/cell_2/fnode5/op_after_combine10/conv/depthwise_kernel
I0601 05:39:09.886441 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode5/op_after_combine10/conv/pointwise_kernel from ckpt var fpn_cells/cell_2/fnode5/op_after_combine10/conv/pointwise_kernel
I0601 05:39:09.886497 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode5/op_after_combine10/conv/bias from ckpt var fpn_cells/cell_2/fnode5/op_after_combine10/conv/bias
I0601 05:39:09.886555 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode5/op_after_combine10/bn/gamma from ckpt var fpn_cells/cell_2/fnode5/op_after_combine10/bn/gamma
I0601 05:39:09.886611 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode5/op_after_combine10/bn/beta from ckpt var fpn_cells/cell_2/fnode5/op_after_combine10/bn/beta
I0601 05:39:09.886669 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode5/op_after_combine10/bn/moving_mean from ckpt var fpn_cells/cell_2/fnode5/op_after_combine10/bn/moving_mean
I0601 05:39:09.886727 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode5/op_after_combine10/bn/moving_variance from ckpt var fpn_cells/cell_2/fnode5/op_after_combine10/bn/moving_variance
I0601 05:39:09.886785 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode6/WSM from ckpt var fpn_cells/cell_2/fnode6/WSM
I0601 05:39:09.886841 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode6/WSM_1 from ckpt var fpn_cells/cell_2/fnode6/WSM_1
I0601 05:39:09.886898 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode6/WSM_2 from ckpt var fpn_cells/cell_2/fnode6/WSM_2
I0601 05:39:09.886963 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode6/op_after_combine11/conv/depthwise_kernel from ckpt var fpn_cells/cell_2/fnode6/op_after_combine11/conv/depthwise_kernel
I0601 05:39:09.887020 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode6/op_after_combine11/conv/pointwise_kernel from ckpt var fpn_cells/cell_2/fnode6/op_after_combine11/conv/pointwise_kernel
I0601 05:39:09.887078 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode6/op_after_combine11/conv/bias from ckpt var fpn_cells/cell_2/fnode6/op_after_combine11/conv/bias
I0601 05:39:09.887142 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode6/op_after_combine11/bn/gamma from ckpt var fpn_cells/cell_2/fnode6/op_after_combine11/bn/gamma
I0601 05:39:09.887200 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode6/op_after_combine11/bn/beta from ckpt var fpn_cells/cell_2/fnode6/op_after_combine11/bn/beta
I0601 05:39:09.887255 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode6/op_after_combine11/bn/moving_mean from ckpt var fpn_cells/cell_2/fnode6/op_after_combine11/bn/moving_mean
I0601 05:39:09.887362 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode6/op_after_combine11/bn/moving_variance from ckpt var fpn_cells/cell_2/fnode6/op_after_combine11/bn/moving_variance
I0601 05:39:09.887431 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode7/WSM from ckpt var fpn_cells/cell_2/fnode7/WSM
I0601 05:39:09.887496 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode7/WSM_1 from ckpt var fpn_cells/cell_2/fnode7/WSM_1
I0601 05:39:09.887563 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode7/op_after_combine12/conv/depthwise_kernel from ckpt var fpn_cells/cell_2/fnode7/op_after_combine12/conv/depthwise_kernel
I0601 05:39:09.887628 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode7/op_after_combine12/conv/pointwise_kernel from ckpt var fpn_cells/cell_2/fnode7/op_after_combine12/conv/pointwise_kernel
I0601 05:39:09.887694 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode7/op_after_combine12/conv/bias from ckpt var fpn_cells/cell_2/fnode7/op_after_combine12/conv/bias
I0601 05:39:09.887764 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode7/op_after_combine12/bn/gamma from ckpt var fpn_cells/cell_2/fnode7/op_after_combine12/bn/gamma
I0601 05:39:09.887830 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode7/op_after_combine12/bn/beta from ckpt var fpn_cells/cell_2/fnode7/op_after_combine12/bn/beta
I0601 05:39:09.887897 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode7/op_after_combine12/bn/moving_mean from ckpt var fpn_cells/cell_2/fnode7/op_after_combine12/bn/moving_mean
I0601 05:39:09.887961 140233166403456 utils.py:108] Init fpn_cells/cell_2/fnode7/op_after_combine12/bn/moving_variance from ckpt var fpn_cells/cell_2/fnode7/op_after_combine12/bn/moving_variance
I0601 05:39:09.888028 140233166403456 utils.py:108] Init class_net/class-0/depthwise_kernel from ckpt var class_net/class-0/depthwise_kernel
I0601 05:39:09.888116 140233166403456 utils.py:108] Init class_net/class-0/pointwise_kernel from ckpt var class_net/class-0/pointwise_kernel
I0601 05:39:09.888230 140233166403456 utils.py:108] Init class_net/class-0/bias from ckpt var class_net/class-0/bias
I0601 05:39:09.888341 140233166403456 utils.py:108] Init class_net/class-0-bn-3/gamma from ckpt var class_net/class-0-bn-3/gamma
I0601 05:39:09.888427 140233166403456 utils.py:108] Init class_net/class-0-bn-3/beta from ckpt var class_net/class-0-bn-3/beta
I0601 05:39:09.888499 140233166403456 utils.py:108] Init class_net/class-0-bn-3/moving_mean from ckpt var class_net/class-0-bn-3/moving_mean
I0601 05:39:09.888574 140233166403456 utils.py:108] Init class_net/class-0-bn-3/moving_variance from ckpt var class_net/class-0-bn-3/moving_variance
I0601 05:39:09.888659 140233166403456 utils.py:108] Init class_net/class-1/depthwise_kernel from ckpt var class_net/class-1/depthwise_kernel
I0601 05:39:09.888747 140233166403456 utils.py:108] Init class_net/class-1/pointwise_kernel from ckpt var class_net/class-1/pointwise_kernel
I0601 05:39:09.888813 140233166403456 utils.py:108] Init class_net/class-1/bias from ckpt var class_net/class-1/bias
I0601 05:39:09.888877 140233166403456 utils.py:108] Init class_net/class-1-bn-3/gamma from ckpt var class_net/class-1-bn-3/gamma
I0601 05:39:09.888990 140233166403456 utils.py:108] Init class_net/class-1-bn-3/beta from ckpt var class_net/class-1-bn-3/beta
I0601 05:39:09.889225 140233166403456 utils.py:108] Init class_net/class-1-bn-3/moving_mean from ckpt var class_net/class-1-bn-3/moving_mean
I0601 05:39:09.889338 140233166403456 utils.py:108] Init class_net/class-1-bn-3/moving_variance from ckpt var class_net/class-1-bn-3/moving_variance
I0601 05:39:09.889415 140233166403456 utils.py:108] Init class_net/class-2/depthwise_kernel from ckpt var class_net/class-2/depthwise_kernel
I0601 05:39:09.889478 140233166403456 utils.py:108] Init class_net/class-2/pointwise_kernel from ckpt var class_net/class-2/pointwise_kernel
I0601 05:39:09.889540 140233166403456 utils.py:108] Init class_net/class-2/bias from ckpt var class_net/class-2/bias
I0601 05:39:09.889602 140233166403456 utils.py:108] Init class_net/class-2-bn-3/gamma from ckpt var class_net/class-2-bn-3/gamma
I0601 05:39:09.889662 140233166403456 utils.py:108] Init class_net/class-2-bn-3/beta from ckpt var class_net/class-2-bn-3/beta
I0601 05:39:09.889726 140233166403456 utils.py:108] Init class_net/class-2-bn-3/moving_mean from ckpt var class_net/class-2-bn-3/moving_mean
I0601 05:39:09.889790 140233166403456 utils.py:108] Init class_net/class-2-bn-3/moving_variance from ckpt var class_net/class-2-bn-3/moving_variance
I0601 05:39:09.890433 140233166403456 utils.py:91] skip class_net/class-predict/depthwise_kernel -- excluded by .*/class-predict/.*
I0601 05:39:09.890554 140233166403456 utils.py:91] skip class_net/class-predict/pointwise_kernel -- excluded by .*/class-predict/.*
I0601 05:39:09.890639 140233166403456 utils.py:91] skip class_net/class-predict/bias -- excluded by .*/class-predict/.*
I0601 05:39:09.890724 140233166403456 utils.py:108] Init class_net/class-0-bn-4/gamma from ckpt var class_net/class-0-bn-4/gamma
I0601 05:39:09.890801 140233166403456 utils.py:108] Init class_net/class-0-bn-4/beta from ckpt var class_net/class-0-bn-4/beta
I0601 05:39:09.890890 140233166403456 utils.py:108] Init class_net/class-0-bn-4/moving_mean from ckpt var class_net/class-0-bn-4/moving_mean
I0601 05:39:09.890956 140233166403456 utils.py:108] Init class_net/class-0-bn-4/moving_variance from ckpt var class_net/class-0-bn-4/moving_variance
I0601 05:39:09.891025 140233166403456 utils.py:108] Init class_net/class-1-bn-4/gamma from ckpt var class_net/class-1-bn-4/gamma
I0601 05:39:09.891092 140233166403456 utils.py:108] Init class_net/class-1-bn-4/beta from ckpt var class_net/class-1-bn-4/beta
I0601 05:39:09.891165 140233166403456 utils.py:108] Init class_net/class-1-bn-4/moving_mean from ckpt var class_net/class-1-bn-4/moving_mean
I0601 05:39:09.891230 140233166403456 utils.py:108] Init class_net/class-1-bn-4/moving_variance from ckpt var class_net/class-1-bn-4/moving_variance
I0601 05:39:09.983481 140233166403456 utils.py:108] Init class_net/class-2-bn-4/gamma from ckpt var class_net/class-2-bn-4/gamma
I0601 05:39:09.983674 140233166403456 utils.py:108] Init class_net/class-2-bn-4/beta from ckpt var class_net/class-2-bn-4/beta
I0601 05:39:09.983794 140233166403456 utils.py:108] Init class_net/class-2-bn-4/moving_mean from ckpt var class_net/class-2-bn-4/moving_mean
I0601 05:39:09.983893 140233166403456 utils.py:108] Init class_net/class-2-bn-4/moving_variance from ckpt var class_net/class-2-bn-4/moving_variance
I0601 05:39:09.983987 140233166403456 utils.py:108] Init class_net/class-0-bn-5/gamma from ckpt var class_net/class-0-bn-5/gamma
I0601 05:39:09.984071 140233166403456 utils.py:108] Init class_net/class-0-bn-5/beta from ckpt var class_net/class-0-bn-5/beta
I0601 05:39:09.984158 140233166403456 utils.py:108] Init class_net/class-0-bn-5/moving_mean from ckpt var class_net/class-0-bn-5/moving_mean
I0601 05:39:09.984228 140233166403456 utils.py:108] Init class_net/class-0-bn-5/moving_variance from ckpt var class_net/class-0-bn-5/moving_variance
I0601 05:39:09.984346 140233166403456 utils.py:108] Init class_net/class-1-bn-5/gamma from ckpt var class_net/class-1-bn-5/gamma
I0601 05:39:09.984415 140233166403456 utils.py:108] Init class_net/class-1-bn-5/beta from ckpt var class_net/class-1-bn-5/beta
I0601 05:39:09.984481 140233166403456 utils.py:108] Init class_net/class-1-bn-5/moving_mean from ckpt var class_net/class-1-bn-5/moving_mean
I0601 05:39:09.984550 140233166403456 utils.py:108] Init class_net/class-1-bn-5/moving_variance from ckpt var class_net/class-1-bn-5/moving_variance
I0601 05:39:09.984613 140233166403456 utils.py:108] Init class_net/class-2-bn-5/gamma from ckpt var class_net/class-2-bn-5/gamma
I0601 05:39:09.984683 140233166403456 utils.py:108] Init class_net/class-2-bn-5/beta from ckpt var class_net/class-2-bn-5/beta
I0601 05:39:09.984744 140233166403456 utils.py:108] Init class_net/class-2-bn-5/moving_mean from ckpt var class_net/class-2-bn-5/moving_mean
I0601 05:39:09.984804 140233166403456 utils.py:108] Init class_net/class-2-bn-5/moving_variance from ckpt var class_net/class-2-bn-5/moving_variance
I0601 05:39:09.984863 140233166403456 utils.py:108] Init class_net/class-0-bn-6/gamma from ckpt var class_net/class-0-bn-6/gamma
I0601 05:39:09.984923 140233166403456 utils.py:108] Init class_net/class-0-bn-6/beta from ckpt var class_net/class-0-bn-6/beta
I0601 05:39:09.984981 140233166403456 utils.py:108] Init class_net/class-0-bn-6/moving_mean from ckpt var class_net/class-0-bn-6/moving_mean
I0601 05:39:09.985039 140233166403456 utils.py:108] Init class_net/class-0-bn-6/moving_variance from ckpt var class_net/class-0-bn-6/moving_variance
I0601 05:39:09.985130 140233166403456 utils.py:108] Init class_net/class-1-bn-6/gamma from ckpt var class_net/class-1-bn-6/gamma
I0601 05:39:09.985195 140233166403456 utils.py:108] Init class_net/class-1-bn-6/beta from ckpt var class_net/class-1-bn-6/beta
I0601 05:39:09.985259 140233166403456 utils.py:108] Init class_net/class-1-bn-6/moving_mean from ckpt var class_net/class-1-bn-6/moving_mean
I0601 05:39:09.985340 140233166403456 utils.py:108] Init class_net/class-1-bn-6/moving_variance from ckpt var class_net/class-1-bn-6/moving_variance
I0601 05:39:09.985402 140233166403456 utils.py:108] Init class_net/class-2-bn-6/gamma from ckpt var class_net/class-2-bn-6/gamma
I0601 05:39:09.985461 140233166403456 utils.py:108] Init class_net/class-2-bn-6/beta from ckpt var class_net/class-2-bn-6/beta
I0601 05:39:09.985521 140233166403456 utils.py:108] Init class_net/class-2-bn-6/moving_mean from ckpt var class_net/class-2-bn-6/moving_mean
I0601 05:39:09.985580 140233166403456 utils.py:108] Init class_net/class-2-bn-6/moving_variance from ckpt var class_net/class-2-bn-6/moving_variance
I0601 05:39:09.985645 140233166403456 utils.py:108] Init class_net/class-0-bn-7/gamma from ckpt var class_net/class-0-bn-7/gamma
I0601 05:39:09.985706 140233166403456 utils.py:108] Init class_net/class-0-bn-7/beta from ckpt var class_net/class-0-bn-7/beta
I0601 05:39:09.985767 140233166403456 utils.py:108] Init class_net/class-0-bn-7/moving_mean from ckpt var class_net/class-0-bn-7/moving_mean
I0601 05:39:09.985827 140233166403456 utils.py:108] Init class_net/class-0-bn-7/moving_variance from ckpt var class_net/class-0-bn-7/moving_variance
I0601 05:39:09.985886 140233166403456 utils.py:108] Init class_net/class-1-bn-7/gamma from ckpt var class_net/class-1-bn-7/gamma
I0601 05:39:09.985946 140233166403456 utils.py:108] Init class_net/class-1-bn-7/beta from ckpt var class_net/class-1-bn-7/beta
I0601 05:39:09.986003 140233166403456 utils.py:108] Init class_net/class-1-bn-7/moving_mean from ckpt var class_net/class-1-bn-7/moving_mean
I0601 05:39:09.986062 140233166403456 utils.py:108] Init class_net/class-1-bn-7/moving_variance from ckpt var class_net/class-1-bn-7/moving_variance
I0601 05:39:09.986149 140233166403456 utils.py:108] Init class_net/class-2-bn-7/gamma from ckpt var class_net/class-2-bn-7/gamma
I0601 05:39:09.986224 140233166403456 utils.py:108] Init class_net/class-2-bn-7/beta from ckpt var class_net/class-2-bn-7/beta
I0601 05:39:09.986313 140233166403456 utils.py:108] Init class_net/class-2-bn-7/moving_mean from ckpt var class_net/class-2-bn-7/moving_mean
I0601 05:39:09.986379 140233166403456 utils.py:108] Init class_net/class-2-bn-7/moving_variance from ckpt var class_net/class-2-bn-7/moving_variance
I0601 05:39:09.986446 140233166403456 utils.py:108] Init box_net/box-0/depthwise_kernel from ckpt var box_net/box-0/depthwise_kernel
I0601 05:39:09.986513 140233166403456 utils.py:108] Init box_net/box-0/pointwise_kernel from ckpt var box_net/box-0/pointwise_kernel
I0601 05:39:09.986578 140233166403456 utils.py:108] Init box_net/box-0/bias from ckpt var box_net/box-0/bias
I0601 05:39:09.986651 140233166403456 utils.py:108] Init box_net/box-0-bn-3/gamma from ckpt var box_net/box-0-bn-3/gamma
I0601 05:39:09.986748 140233166403456 utils.py:108] Init box_net/box-0-bn-3/beta from ckpt var box_net/box-0-bn-3/beta
I0601 05:39:09.986815 140233166403456 utils.py:108] Init box_net/box-0-bn-3/moving_mean from ckpt var box_net/box-0-bn-3/moving_mean
I0601 05:39:09.986881 140233166403456 utils.py:108] Init box_net/box-0-bn-3/moving_variance from ckpt var box_net/box-0-bn-3/moving_variance
I0601 05:39:09.986946 140233166403456 utils.py:108] Init box_net/box-1/depthwise_kernel from ckpt var box_net/box-1/depthwise_kernel
I0601 05:39:09.987009 140233166403456 utils.py:108] Init box_net/box-1/pointwise_kernel from ckpt var box_net/box-1/pointwise_kernel
I0601 05:39:09.987073 140233166403456 utils.py:108] Init box_net/box-1/bias from ckpt var box_net/box-1/bias
I0601 05:39:09.987136 140233166403456 utils.py:108] Init box_net/box-1-bn-3/gamma from ckpt var box_net/box-1-bn-3/gamma
I0601 05:39:09.987200 140233166403456 utils.py:108] Init box_net/box-1-bn-3/beta from ckpt var box_net/box-1-bn-3/beta
I0601 05:39:09.987262 140233166403456 utils.py:108] Init box_net/box-1-bn-3/moving_mean from ckpt var box_net/box-1-bn-3/moving_mean
I0601 05:39:09.987340 140233166403456 utils.py:108] Init box_net/box-1-bn-3/moving_variance from ckpt var box_net/box-1-bn-3/moving_variance
I0601 05:39:09.987406 140233166403456 utils.py:108] Init box_net/box-2/depthwise_kernel from ckpt var box_net/box-2/depthwise_kernel
I0601 05:39:09.987472 140233166403456 utils.py:108] Init box_net/box-2/pointwise_kernel from ckpt var box_net/box-2/pointwise_kernel
I0601 05:39:09.987536 140233166403456 utils.py:108] Init box_net/box-2/bias from ckpt var box_net/box-2/bias
I0601 05:39:09.987598 140233166403456 utils.py:108] Init box_net/box-2-bn-3/gamma from ckpt var box_net/box-2-bn-3/gamma
I0601 05:39:09.987668 140233166403456 utils.py:108] Init box_net/box-2-bn-3/beta from ckpt var box_net/box-2-bn-3/beta
I0601 05:39:09.987733 140233166403456 utils.py:108] Init box_net/box-2-bn-3/moving_mean from ckpt var box_net/box-2-bn-3/moving_mean
I0601 05:39:09.987796 140233166403456 utils.py:108] Init box_net/box-2-bn-3/moving_variance from ckpt var box_net/box-2-bn-3/moving_variance
I0601 05:39:09.987859 140233166403456 utils.py:108] Init box_net/box-predict/depthwise_kernel from ckpt var box_net/box-predict/depthwise_kernel
I0601 05:39:09.987922 140233166403456 utils.py:108] Init box_net/box-predict/pointwise_kernel from ckpt var box_net/box-predict/pointwise_kernel
I0601 05:39:09.987985 140233166403456 utils.py:108] Init box_net/box-predict/bias from ckpt var box_net/box-predict/bias
I0601 05:39:09.988047 140233166403456 utils.py:108] Init box_net/box-0-bn-4/gamma from ckpt var box_net/box-0-bn-4/gamma
I0601 05:39:09.988109 140233166403456 utils.py:108] Init box_net/box-0-bn-4/beta from ckpt var box_net/box-0-bn-4/beta
I0601 05:39:09.988172 140233166403456 utils.py:108] Init box_net/box-0-bn-4/moving_mean from ckpt var box_net/box-0-bn-4/moving_mean
I0601 05:39:09.988252 140233166403456 utils.py:108] Init box_net/box-0-bn-4/moving_variance from ckpt var box_net/box-0-bn-4/moving_variance
I0601 05:39:09.988334 140233166403456 utils.py:108] Init box_net/box-1-bn-4/gamma from ckpt var box_net/box-1-bn-4/gamma
I0601 05:39:09.988414 140233166403456 utils.py:108] Init box_net/box-1-bn-4/beta from ckpt var box_net/box-1-bn-4/beta
I0601 05:39:09.988478 140233166403456 utils.py:108] Init box_net/box-1-bn-4/moving_mean from ckpt var box_net/box-1-bn-4/moving_mean
I0601 05:39:09.988542 140233166403456 utils.py:108] Init box_net/box-1-bn-4/moving_variance from ckpt var box_net/box-1-bn-4/moving_variance
I0601 05:39:09.988607 140233166403456 utils.py:108] Init box_net/box-2-bn-4/gamma from ckpt var box_net/box-2-bn-4/gamma
I0601 05:39:09.988676 140233166403456 utils.py:108] Init box_net/box-2-bn-4/beta from ckpt var box_net/box-2-bn-4/beta
I0601 05:39:09.988740 140233166403456 utils.py:108] Init box_net/box-2-bn-4/moving_mean from ckpt var box_net/box-2-bn-4/moving_mean
I0601 05:39:09.988821 140233166403456 utils.py:108] Init box_net/box-2-bn-4/moving_variance from ckpt var box_net/box-2-bn-4/moving_variance
I0601 05:39:09.988892 140233166403456 utils.py:108] Init box_net/box-0-bn-5/gamma from ckpt var box_net/box-0-bn-5/gamma
I0601 05:39:09.988961 140233166403456 utils.py:108] Init box_net/box-0-bn-5/beta from ckpt var box_net/box-0-bn-5/beta
I0601 05:39:09.989033 140233166403456 utils.py:108] Init box_net/box-0-bn-5/moving_mean from ckpt var box_net/box-0-bn-5/moving_mean
I0601 05:39:09.989109 140233166403456 utils.py:108] Init box_net/box-0-bn-5/moving_variance from ckpt var box_net/box-0-bn-5/moving_variance
I0601 05:39:09.989176 140233166403456 utils.py:108] Init box_net/box-1-bn-5/gamma from ckpt var box_net/box-1-bn-5/gamma
I0601 05:39:09.989240 140233166403456 utils.py:108] Init box_net/box-1-bn-5/beta from ckpt var box_net/box-1-bn-5/beta
I0601 05:39:09.989319 140233166403456 utils.py:108] Init box_net/box-1-bn-5/moving_mean from ckpt var box_net/box-1-bn-5/moving_mean
I0601 05:39:09.989387 140233166403456 utils.py:108] Init box_net/box-1-bn-5/moving_variance from ckpt var box_net/box-1-bn-5/moving_variance
I0601 05:39:09.989453 140233166403456 utils.py:108] Init box_net/box-2-bn-5/gamma from ckpt var box_net/box-2-bn-5/gamma
I0601 05:39:09.989518 140233166403456 utils.py:108] Init box_net/box-2-bn-5/beta from ckpt var box_net/box-2-bn-5/beta
I0601 05:39:09.989582 140233166403456 utils.py:108] Init box_net/box-2-bn-5/moving_mean from ckpt var box_net/box-2-bn-5/moving_mean
I0601 05:39:09.989650 140233166403456 utils.py:108] Init box_net/box-2-bn-5/moving_variance from ckpt var box_net/box-2-bn-5/moving_variance
I0601 05:39:09.989731 140233166403456 utils.py:108] Init box_net/box-0-bn-6/gamma from ckpt var box_net/box-0-bn-6/gamma
I0601 05:39:09.989798 140233166403456 utils.py:108] Init box_net/box-0-bn-6/beta from ckpt var box_net/box-0-bn-6/beta
I0601 05:39:09.989864 140233166403456 utils.py:108] Init box_net/box-0-bn-6/moving_mean from ckpt var box_net/box-0-bn-6/moving_mean
I0601 05:39:09.989939 140233166403456 utils.py:108] Init box_net/box-0-bn-6/moving_variance from ckpt var box_net/box-0-bn-6/moving_variance
I0601 05:39:09.990000 140233166403456 utils.py:108] Init box_net/box-1-bn-6/gamma from ckpt var box_net/box-1-bn-6/gamma
I0601 05:39:09.990062 140233166403456 utils.py:108] Init box_net/box-1-bn-6/beta from ckpt var box_net/box-1-bn-6/beta
I0601 05:39:09.990124 140233166403456 utils.py:108] Init box_net/box-1-bn-6/moving_mean from ckpt var box_net/box-1-bn-6/moving_mean
I0601 05:39:09.990205 140233166403456 utils.py:108] Init box_net/box-1-bn-6/moving_variance from ckpt var box_net/box-1-bn-6/moving_variance
I0601 05:39:09.990283 140233166403456 utils.py:108] Init box_net/box-2-bn-6/gamma from ckpt var box_net/box-2-bn-6/gamma
I0601 05:39:09.990348 140233166403456 utils.py:108] Init box_net/box-2-bn-6/beta from ckpt var box_net/box-2-bn-6/beta
I0601 05:39:09.990414 140233166403456 utils.py:108] Init box_net/box-2-bn-6/moving_mean from ckpt var box_net/box-2-bn-6/moving_mean
I0601 05:39:09.990480 140233166403456 utils.py:108] Init box_net/box-2-bn-6/moving_variance from ckpt var box_net/box-2-bn-6/moving_variance
I0601 05:39:09.990545 140233166403456 utils.py:108] Init box_net/box-0-bn-7/gamma from ckpt var box_net/box-0-bn-7/gamma
I0601 05:39:09.990610 140233166403456 utils.py:108] Init box_net/box-0-bn-7/beta from ckpt var box_net/box-0-bn-7/beta
I0601 05:39:09.990680 140233166403456 utils.py:108] Init box_net/box-0-bn-7/moving_mean from ckpt var box_net/box-0-bn-7/moving_mean
I0601 05:39:09.990744 140233166403456 utils.py:108] Init box_net/box-0-bn-7/moving_variance from ckpt var box_net/box-0-bn-7/moving_variance
I0601 05:39:09.990807 140233166403456 utils.py:108] Init box_net/box-1-bn-7/gamma from ckpt var box_net/box-1-bn-7/gamma
I0601 05:39:09.990874 140233166403456 utils.py:108] Init box_net/box-1-bn-7/beta from ckpt var box_net/box-1-bn-7/beta
I0601 05:39:09.990938 140233166403456 utils.py:108] Init box_net/box-1-bn-7/moving_mean from ckpt var box_net/box-1-bn-7/moving_mean
I0601 05:39:09.991002 140233166403456 utils.py:108] Init box_net/box-1-bn-7/moving_variance from ckpt var box_net/box-1-bn-7/moving_variance
I0601 05:39:09.991075 140233166403456 utils.py:108] Init box_net/box-2-bn-7/gamma from ckpt var box_net/box-2-bn-7/gamma
I0601 05:39:09.991136 140233166403456 utils.py:108] Init box_net/box-2-bn-7/beta from ckpt var box_net/box-2-bn-7/beta
I0601 05:39:09.991196 140233166403456 utils.py:108] Init box_net/box-2-bn-7/moving_mean from ckpt var box_net/box-2-bn-7/moving_mean
I0601 05:39:09.991255 140233166403456 utils.py:108] Init box_net/box-2-bn-7/moving_variance from ckpt var box_net/box-2-bn-7/moving_variance
I0601 05:39:09.993984 140233166403456 utils.py:91] skip class_net/class-predict/depthwise_kernel/Momentum -- excluded by .*/class-predict/.*
I0601 05:39:09.994096 140233166403456 utils.py:91] skip class_net/class-predict/pointwise_kernel/Momentum -- excluded by .*/class-predict/.*
I0601 05:39:09.994165 140233166403456 utils.py:91] skip class_net/class-predict/bias/Momentum -- excluded by .*/class-predict/.*
INFO:tensorflow:Done calling model_fn.
I0601 05:39:11.603600 140233166403456 estimator.py:1171] Done calling model_fn.
INFO:tensorflow:Create CheckpointSaverHook.
I0601 05:39:11.604779 140233166403456 basic_session_run_hooks.py:546] Create CheckpointSaverHook.
INFO:tensorflow:Graph was finalized.
I0601 05:39:15.392178 140233166403456 monitored_session.py:246] Graph was finalized.
2020-06-01 05:39:15.397315: I tensorflow/core/platform/profile_utils/cpu_utils.cc:102] CPU Frequency: 2300000000 Hz
2020-06-01 05:39:15.397531: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x313f100 initialized for platform Host (this does not guarantee that XLA will be used). Devices:
2020-06-01 05:39:15.397568: I tensorflow/compiler/xla/service/service.cc:176]   StreamExecutor device (0): Host, Default Version
2020-06-01 05:39:15.504852: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:39:15.505557: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x313f2c0 initialized for platform CUDA (this does not guarantee that XLA will be used). Devices:
2020-06-01 05:39:15.505594: I tensorflow/compiler/xla/service/service.cc:176]   StreamExecutor device (0): Tesla T4, Compute Capability 7.5
2020-06-01 05:39:15.505802: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:39:15.506529: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1561] Found device 0 with properties: 
pciBusID: 0000:00:04.0 name: Tesla T4 computeCapability: 7.5
coreClock: 1.59GHz coreCount: 40 deviceMemorySize: 14.73GiB deviceMemoryBandwidth: 298.08GiB/s
2020-06-01 05:39:15.506599: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.1
2020-06-01 05:39:15.506655: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcublas.so.10
2020-06-01 05:39:15.506677: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcufft.so.10
2020-06-01 05:39:15.506701: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcurand.so.10
2020-06-01 05:39:15.506720: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusolver.so.10
2020-06-01 05:39:15.506738: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusparse.so.10
2020-06-01 05:39:15.506757: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudnn.so.7
2020-06-01 05:39:15.506858: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:39:15.507460: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:39:15.507937: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1703] Adding visible gpu devices: 0
2020-06-01 05:39:15.507984: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.1
2020-06-01 05:39:16.020285: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1102] Device interconnect StreamExecutor with strength 1 edge matrix:
2020-06-01 05:39:16.020341: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1108]      0 
2020-06-01 05:39:16.020355: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1121] 0:   N 
2020-06-01 05:39:16.020572: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:39:16.021232: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:39:16.021838: W tensorflow/core/common_runtime/gpu/gpu_bfc_allocator.cc:39] Overriding allow_growth setting because the TF_FORCE_GPU_ALLOW_GROWTH environment variable is set. Original config value was 0.
2020-06-01 05:39:16.021902: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1247] Created TensorFlow device (/job:localhost/replica:0/task:0/device:GPU:0 with 13970 MB memory) -> physical GPU (device: 0, name: Tesla T4, pci bus id: 0000:00:04.0, compute capability: 7.5)
INFO:tensorflow:Running local_init_op.
I0601 05:39:20.487684 140233166403456 session_manager.py:505] Running local_init_op.
INFO:tensorflow:Done running local_init_op.
I0601 05:39:20.766017 140233166403456 session_manager.py:508] Done running local_init_op.
INFO:tensorflow:Calling checkpoint listeners before saving checkpoint 0...
I0601 05:39:30.224425 140233166403456 basic_session_run_hooks.py:614] Calling checkpoint listeners before saving checkpoint 0...
INFO:tensorflow:Saving checkpoints for 0 into /tmp/model_dir/efficientdet-d0-finetune/model.ckpt.
I0601 05:39:30.225207 140233166403456 basic_session_run_hooks.py:618] Saving checkpoints for 0 into /tmp/model_dir/efficientdet-d0-finetune/model.ckpt.
INFO:tensorflow:Calling checkpoint listeners after saving checkpoint 0...
I0601 05:39:32.620833 140233166403456 basic_session_run_hooks.py:626] Calling checkpoint listeners after saving checkpoint 0...
2020-06-01 05:39:42.789892: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudnn.so.7
2020-06-01 05:39:44.603517: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcublas.so.10
INFO:tensorflow:global_step/sec: 0.127531
I0601 05:39:53.424471 140233166403456 tpu_estimator.py:2350] global_step/sec: 0.127531
INFO:tensorflow:examples/sec: 1.02024
I0601 05:39:53.425666 140233166403456 tpu_estimator.py:2351] examples/sec: 1.02024
INFO:tensorflow:global_step/sec: 2.87711
I0601 05:39:53.771956 140233166403456 tpu_estimator.py:2350] global_step/sec: 2.87711
INFO:tensorflow:examples/sec: 23.0169
I0601 05:39:53.772370 140233166403456 tpu_estimator.py:2351] examples/sec: 23.0169
INFO:tensorflow:global_step/sec: 2.94437
I0601 05:39:54.111575 140233166403456 tpu_estimator.py:2350] global_step/sec: 2.94437
INFO:tensorflow:examples/sec: 23.5549
I0601 05:39:54.111993 140233166403456 tpu_estimator.py:2351] examples/sec: 23.5549
INFO:tensorflow:global_step/sec: 3.28482
I0601 05:39:54.415997 140233166403456 tpu_estimator.py:2350] global_step/sec: 3.28482
INFO:tensorflow:examples/sec: 26.2785
I0601 05:39:54.416417 140233166403456 tpu_estimator.py:2351] examples/sec: 26.2785
INFO:tensorflow:global_step/sec: 3.26759
I0601 05:39:54.722034 140233166403456 tpu_estimator.py:2350] global_step/sec: 3.26759
INFO:tensorflow:examples/sec: 26.1407
I0601 05:39:54.722304 140233166403456 tpu_estimator.py:2351] examples/sec: 26.1407
INFO:tensorflow:global_step/sec: 3.4836
I0601 05:39:55.009089 140233166403456 tpu_estimator.py:2350] global_step/sec: 3.4836
INFO:tensorflow:examples/sec: 27.8688
I0601 05:39:55.009364 140233166403456 tpu_estimator.py:2351] examples/sec: 27.8688
INFO:tensorflow:Calling checkpoint listeners before saving checkpoint 7...
I0601 05:39:55.009915 140233166403456 basic_session_run_hooks.py:614] Calling checkpoint listeners before saving checkpoint 7...
INFO:tensorflow:Saving checkpoints for 7 into /tmp/model_dir/efficientdet-d0-finetune/model.ckpt.
I0601 05:39:55.010076 140233166403456 basic_session_run_hooks.py:618] Saving checkpoints for 7 into /tmp/model_dir/efficientdet-d0-finetune/model.ckpt.
INFO:tensorflow:Calling checkpoint listeners after saving checkpoint 7...
I0601 05:39:56.579441 140233166403456 basic_session_run_hooks.py:626] Calling checkpoint listeners after saving checkpoint 7...
INFO:tensorflow:Loss for final step: 1.5810888.
I0601 05:39:57.246410 140233166403456 estimator.py:350] Loss for final step: 1.5810888.
INFO:tensorflow:training_loop marked as finished
I0601 05:39:57.247187 140233166403456 error_handling.py:115] training_loop marked as finished
I0601 05:39:57.247375 140233166403456 main.py:388] Starting evaluation cycle, epoch: 0.
INFO:tensorflow:Using config: {'_model_dir': '/tmp/model_dir/efficientdet-d0-finetune', '_tf_random_seed': None, '_save_summary_steps': 100, '_save_checkpoints_steps': None, '_save_checkpoints_secs': 600, '_session_config': allow_soft_placement: true
, '_keep_checkpoint_max': 5, '_keep_checkpoint_every_n_hours': 10000, '_log_step_count_steps': None, '_train_distribute': None, '_device_fn': None, '_protocol': None, '_eval_distribute': None, '_experimental_distribute': None, '_experimental_max_worker_delay_secs': None, '_session_creation_timeout_secs': 7200, '_service': None, '_cluster_spec': ClusterSpec({}), '_task_type': 'worker', '_task_id': 0, '_global_id_in_cluster': 0, '_master': '', '_evaluation_master': '', '_is_chief': True, '_num_ps_replicas': 0, '_num_worker_replicas': 1, '_tpu_config': TPUConfig(iterations_per_loop=100, num_shards=8, num_cores_per_replica=None, per_host_input_for_training=3, tpu_job_name=None, initial_infeed_sleep_secs=None, input_partition_dims=None, eval_training_input_configuration=2, experimental_host_call_every_n_steps=1), '_cluster': None}
I0601 05:39:57.248242 140233166403456 estimator.py:191] Using config: {'_model_dir': '/tmp/model_dir/efficientdet-d0-finetune', '_tf_random_seed': None, '_save_summary_steps': 100, '_save_checkpoints_steps': None, '_save_checkpoints_secs': 600, '_session_config': allow_soft_placement: true
, '_keep_checkpoint_max': 5, '_keep_checkpoint_every_n_hours': 10000, '_log_step_count_steps': None, '_train_distribute': None, '_device_fn': None, '_protocol': None, '_eval_distribute': None, '_experimental_distribute': None, '_experimental_max_worker_delay_secs': None, '_session_creation_timeout_secs': 7200, '_service': None, '_cluster_spec': ClusterSpec({}), '_task_type': 'worker', '_task_id': 0, '_global_id_in_cluster': 0, '_master': '', '_evaluation_master': '', '_is_chief': True, '_num_ps_replicas': 0, '_num_worker_replicas': 1, '_tpu_config': TPUConfig(iterations_per_loop=100, num_shards=8, num_cores_per_replica=None, per_host_input_for_training=3, tpu_job_name=None, initial_infeed_sleep_secs=None, input_partition_dims=None, eval_training_input_configuration=2, experimental_host_call_every_n_steps=1), '_cluster': None}
INFO:tensorflow:_TPUContext: eval_on_tpu True
I0601 05:39:57.248666 140233166403456 tpu_context.py:216] _TPUContext: eval_on_tpu True
WARNING:tensorflow:eval_on_tpu ignored because use_tpu is False.
W0601 05:39:57.248775 140233166403456 tpu_context.py:218] eval_on_tpu ignored because use_tpu is False.
INFO:tensorflow:Calling model_fn.
I0601 05:39:57.717588 140233166403456 estimator.py:1169] Calling model_fn.
INFO:tensorflow:Running eval on CPU/GPU
I0601 05:39:57.717878 140233166403456 tpu_estimator.py:3171] Running eval on CPU/GPU
I0601 05:39:57.718361 140233166403456 utils.py:595] use mixed precision policy name mixed_float16
I0601 05:39:57.719170 140233166403456 efficientdet_arch.py:720] act_type: swish
alpha: 0.25
anchor_scale: 4.0
apply_bn_for_resampling: true
aspect_ratios:
- !!python/tuple
    - 1.0
    - 1.0
- !!python/tuple
    - 1.4
    - 0.7
- !!python/tuple
    - 0.7
    - 1.4
augmix_params: !!python/tuple
- 3
- -1
- 1
autoaugment_policy: null
backbone_ckpt: ''
backbone_config: null
backbone_name: efficientnet-b0
batch_size: 8
box_class_repeats: 3
box_loss_weight: 50.0
ckpt: efficientdet-d0
ckpt_var_scope: null
clip_gradients_norm: 10.0
conv_after_downsample: false
conv_bn_act_pattern: false
data_format: channels_last
delta: 0.1
first_lr_drop_epoch: 200.0
fpn_cell_repeats: 3
fpn_config: null
fpn_name: null
fpn_num_filters: 64
fpn_weight_method: null
gamma: 1.5
image_size: !!python/tuple
- 512
- 512
img_summary_steps: null
input_rand_hflip: false
iou_loss_type: null
iou_loss_weight: 1.0
is_training_bn: false
iterations_per_loop: 100
label_id_mapping: null
learning_rate: 0.08
lr_decay_method: cosine
lr_warmup_epoch: 1.0
lr_warmup_init: 0.008
max_instances_per_image: 100
max_level: 7
min_level: 3
mode: train_and_eval
model_dir: /tmp/model_dir/efficientdet-d0-finetune
model_name: efficientdet-d0
momentum: 0.9
moving_average_decay: 0
name: efficientdet-d0
num_classes: 20
num_epochs: 1
num_examples_per_epoch: 56
num_scales: 3
num_shards: 8
optimizer: sgd
poly_lr_power: 0.9
pooling_type: null
precision: mixed_float16
resnet_depth: 50
second_lr_drop_epoch: 250.0
separable_conv: true
skip_crowd_during_training: true
strategy: null
survival_prob: null
target_size: null
testdev_dir: null
train_scale_max: 2.0
train_scale_min: 0.1
use_augmix: false
use_native_resize_op: true
use_tpu: false
val_json_file: null
var_exclude_expr: .*/class-predict/.*
var_freeze_expr: null
weight_decay: 4.0e-05

I0601 05:39:57.724439 140233166403456 efficientnet_builder.py:224] global_params= GlobalParams(batch_norm_momentum=0.99, batch_norm_epsilon=0.001, dropout_rate=0.2, data_format='channels_last', num_classes=1000, width_coefficient=1.0, depth_coefficient=1.0, depth_divisor=8, min_depth=None, survival_prob=0.0, relu_fn=functools.partial(<function activation_fn at 0x7f8a3d432d90>, act_type='swish'), batch_norm=<class 'utils.BatchNormalization'>, use_se=True, local_pooling=None, condconv_num_experts=None, clip_projection_output=False, blocks_args=['r1_k3_s11_e1_i32_o16_se0.25', 'r2_k3_s22_e6_i16_o24_se0.25', 'r2_k5_s22_e6_i24_o40_se0.25', 'r3_k3_s22_e6_i40_o80_se0.25', 'r3_k5_s11_e6_i80_o112_se0.25', 'r4_k5_s22_e6_i112_o192_se0.25', 'r1_k3_s11_e6_i192_o320_se0.25'], fix_head_stem=None)
I0601 05:39:57.994890 140233166403456 api.py:587] Built stem layers with output shape: (8, 256, 256, 32)
I0601 05:39:58.004813 140233166403456 api.py:587] Block mb_conv_block  input shape: (8, 256, 256, 32)
I0601 05:39:58.041763 140233166403456 api.py:587] DWConv shape: (8, 256, 256, 32)
I0601 05:39:58.082441 140233166403456 api.py:587] Built Squeeze and Excitation with tensor shape: (8, 1, 1, 32)
I0601 05:39:58.118283 140233166403456 api.py:587] Project shape: (8, 256, 256, 16)
I0601 05:39:58.129572 140233166403456 api.py:587] Block mb_conv_block_1  input shape: (8, 256, 256, 16)
I0601 05:39:58.167157 140233166403456 api.py:587] Expand shape: (8, 256, 256, 96)
I0601 05:39:58.200410 140233166403456 api.py:587] DWConv shape: (8, 128, 128, 96)
I0601 05:39:58.240649 140233166403456 api.py:587] Built Squeeze and Excitation with tensor shape: (8, 1, 1, 96)
I0601 05:39:58.273849 140233166403456 api.py:587] Project shape: (8, 128, 128, 24)
I0601 05:39:58.283733 140233166403456 api.py:587] Block mb_conv_block_2  input shape: (8, 128, 128, 24)
I0601 05:39:58.316170 140233166403456 api.py:587] Expand shape: (8, 128, 128, 144)
I0601 05:39:58.349213 140233166403456 api.py:587] DWConv shape: (8, 128, 128, 144)
I0601 05:39:58.393920 140233166403456 api.py:587] Built Squeeze and Excitation with tensor shape: (8, 1, 1, 144)
I0601 05:39:58.425131 140233166403456 api.py:587] Project shape: (8, 128, 128, 24)
I0601 05:39:58.439640 140233166403456 api.py:587] Block mb_conv_block_3  input shape: (8, 128, 128, 24)
I0601 05:39:58.480211 140233166403456 api.py:587] Expand shape: (8, 128, 128, 144)
I0601 05:39:58.512615 140233166403456 api.py:587] DWConv shape: (8, 64, 64, 144)
I0601 05:39:58.547719 140233166403456 api.py:587] Built Squeeze and Excitation with tensor shape: (8, 1, 1, 144)
I0601 05:39:58.581837 140233166403456 api.py:587] Project shape: (8, 64, 64, 40)
I0601 05:39:58.593011 140233166403456 api.py:587] Block mb_conv_block_4  input shape: (8, 64, 64, 40)
I0601 05:39:58.625525 140233166403456 api.py:587] Expand shape: (8, 64, 64, 240)
I0601 05:39:58.658681 140233166403456 api.py:587] DWConv shape: (8, 64, 64, 240)
I0601 05:39:58.694889 140233166403456 api.py:587] Built Squeeze and Excitation with tensor shape: (8, 1, 1, 240)
I0601 05:39:58.726229 140233166403456 api.py:587] Project shape: (8, 64, 64, 40)
I0601 05:39:58.739845 140233166403456 api.py:587] Block mb_conv_block_5  input shape: (8, 64, 64, 40)
I0601 05:39:58.774328 140233166403456 api.py:587] Expand shape: (8, 64, 64, 240)
I0601 05:39:58.809783 140233166403456 api.py:587] DWConv shape: (8, 32, 32, 240)
I0601 05:39:58.845743 140233166403456 api.py:587] Built Squeeze and Excitation with tensor shape: (8, 1, 1, 240)
I0601 05:39:58.878186 140233166403456 api.py:587] Project shape: (8, 32, 32, 80)
I0601 05:39:58.889413 140233166403456 api.py:587] Block mb_conv_block_6  input shape: (8, 32, 32, 80)
I0601 05:39:58.924906 140233166403456 api.py:587] Expand shape: (8, 32, 32, 480)
I0601 05:39:58.958735 140233166403456 api.py:587] DWConv shape: (8, 32, 32, 480)
I0601 05:39:59.002022 140233166403456 api.py:587] Built Squeeze and Excitation with tensor shape: (8, 1, 1, 480)
I0601 05:39:59.038743 140233166403456 api.py:587] Project shape: (8, 32, 32, 80)
I0601 05:39:59.051755 140233166403456 api.py:587] Block mb_conv_block_7  input shape: (8, 32, 32, 80)
I0601 05:39:59.085218 140233166403456 api.py:587] Expand shape: (8, 32, 32, 480)
I0601 05:39:59.118895 140233166403456 api.py:587] DWConv shape: (8, 32, 32, 480)
I0601 05:39:59.154775 140233166403456 api.py:587] Built Squeeze and Excitation with tensor shape: (8, 1, 1, 480)
I0601 05:39:59.191923 140233166403456 api.py:587] Project shape: (8, 32, 32, 80)
I0601 05:39:59.201802 140233166403456 api.py:587] Block mb_conv_block_8  input shape: (8, 32, 32, 80)
I0601 05:39:59.236343 140233166403456 api.py:587] Expand shape: (8, 32, 32, 480)
I0601 05:39:59.277680 140233166403456 api.py:587] DWConv shape: (8, 32, 32, 480)
I0601 05:39:59.316390 140233166403456 api.py:587] Built Squeeze and Excitation with tensor shape: (8, 1, 1, 480)
I0601 05:39:59.352003 140233166403456 api.py:587] Project shape: (8, 32, 32, 112)
I0601 05:39:59.361599 140233166403456 api.py:587] Block mb_conv_block_9  input shape: (8, 32, 32, 112)
I0601 05:39:59.403441 140233166403456 api.py:587] Expand shape: (8, 32, 32, 672)
I0601 05:39:59.437427 140233166403456 api.py:587] DWConv shape: (8, 32, 32, 672)
I0601 05:39:59.482208 140233166403456 api.py:587] Built Squeeze and Excitation with tensor shape: (8, 1, 1, 672)
I0601 05:39:59.514467 140233166403456 api.py:587] Project shape: (8, 32, 32, 112)
I0601 05:39:59.524105 140233166403456 api.py:587] Block mb_conv_block_10  input shape: (8, 32, 32, 112)
I0601 05:39:59.558211 140233166403456 api.py:587] Expand shape: (8, 32, 32, 672)
I0601 05:39:59.595371 140233166403456 api.py:587] DWConv shape: (8, 32, 32, 672)
I0601 05:39:59.630897 140233166403456 api.py:587] Built Squeeze and Excitation with tensor shape: (8, 1, 1, 672)
I0601 05:39:59.666866 140233166403456 api.py:587] Project shape: (8, 32, 32, 112)
I0601 05:39:59.676909 140233166403456 api.py:587] Block mb_conv_block_11  input shape: (8, 32, 32, 112)
I0601 05:39:59.711543 140233166403456 api.py:587] Expand shape: (8, 32, 32, 672)
I0601 05:39:59.745486 140233166403456 api.py:587] DWConv shape: (8, 16, 16, 672)
I0601 05:39:59.784313 140233166403456 api.py:587] Built Squeeze and Excitation with tensor shape: (8, 1, 1, 672)
I0601 05:39:59.815086 140233166403456 api.py:587] Project shape: (8, 16, 16, 192)
I0601 05:39:59.824952 140233166403456 api.py:587] Block mb_conv_block_12  input shape: (8, 16, 16, 192)
I0601 05:39:59.865648 140233166403456 api.py:587] Expand shape: (8, 16, 16, 1152)
I0601 05:39:59.903240 140233166403456 api.py:587] DWConv shape: (8, 16, 16, 1152)
I0601 05:39:59.940747 140233166403456 api.py:587] Built Squeeze and Excitation with tensor shape: (8, 1, 1, 1152)
I0601 05:39:59.988296 140233166403456 api.py:587] Project shape: (8, 16, 16, 192)
I0601 05:39:59.998693 140233166403456 api.py:587] Block mb_conv_block_13  input shape: (8, 16, 16, 192)
I0601 05:40:00.038145 140233166403456 api.py:587] Expand shape: (8, 16, 16, 1152)
I0601 05:40:00.081367 140233166403456 api.py:587] DWConv shape: (8, 16, 16, 1152)
I0601 05:40:00.118943 140233166403456 api.py:587] Built Squeeze and Excitation with tensor shape: (8, 1, 1, 1152)
I0601 05:40:00.155311 140233166403456 api.py:587] Project shape: (8, 16, 16, 192)
I0601 05:40:00.166164 140233166403456 api.py:587] Block mb_conv_block_14  input shape: (8, 16, 16, 192)
I0601 05:40:00.204564 140233166403456 api.py:587] Expand shape: (8, 16, 16, 1152)
I0601 05:40:00.248919 140233166403456 api.py:587] DWConv shape: (8, 16, 16, 1152)
I0601 05:40:00.289414 140233166403456 api.py:587] Built Squeeze and Excitation with tensor shape: (8, 1, 1, 1152)
I0601 05:40:00.327728 140233166403456 api.py:587] Project shape: (8, 16, 16, 192)
I0601 05:40:00.337415 140233166403456 api.py:587] Block mb_conv_block_15  input shape: (8, 16, 16, 192)
I0601 05:40:00.375041 140233166403456 api.py:587] Expand shape: (8, 16, 16, 1152)
I0601 05:40:00.420453 140233166403456 api.py:587] DWConv shape: (8, 16, 16, 1152)
I0601 05:40:00.460810 140233166403456 api.py:587] Built Squeeze and Excitation with tensor shape: (8, 1, 1, 1152)
I0601 05:40:00.500894 140233166403456 api.py:587] Project shape: (8, 16, 16, 320)
Parsing Inputs...
I0601 05:40:00.823563 140233166403456 efficientdet_arch.py:725] backbone params/flops = 3.595388M, 15.443674561B
I0601 05:40:00.884665 140233166403456 efficientdet_arch.py:471] building cell 0
I0601 05:40:00.885131 140233166403456 efficientdet_arch.py:648] fnode 0 : {'feat_level': 6, 'inputs_offsets': [3, 4]}
I0601 05:40:00.957454 140233166403456 efficientdet_arch.py:648] fnode 1 : {'feat_level': 5, 'inputs_offsets': [2, 5]}
I0601 05:40:01.079147 140233166403456 efficientdet_arch.py:648] fnode 2 : {'feat_level': 4, 'inputs_offsets': [1, 6]}
I0601 05:40:01.204365 140233166403456 efficientdet_arch.py:648] fnode 3 : {'feat_level': 3, 'inputs_offsets': [0, 7]}
I0601 05:40:01.331107 140233166403456 efficientdet_arch.py:648] fnode 4 : {'feat_level': 4, 'inputs_offsets': [1, 7, 8]}
I0601 05:40:01.475595 140233166403456 efficientdet_arch.py:648] fnode 5 : {'feat_level': 5, 'inputs_offsets': [2, 6, 9]}
I0601 05:40:01.613884 140233166403456 efficientdet_arch.py:648] fnode 6 : {'feat_level': 6, 'inputs_offsets': [3, 5, 10]}
I0601 05:40:01.698622 140233166403456 efficientdet_arch.py:648] fnode 7 : {'feat_level': 7, 'inputs_offsets': [4, 11]}
I0601 05:40:01.783592 140233166403456 efficientdet_arch.py:471] building cell 1
I0601 05:40:01.783952 140233166403456 efficientdet_arch.py:648] fnode 0 : {'feat_level': 6, 'inputs_offsets': [3, 4]}
I0601 05:40:01.864315 140233166403456 efficientdet_arch.py:648] fnode 1 : {'feat_level': 5, 'inputs_offsets': [2, 5]}
I0601 05:40:01.946357 140233166403456 efficientdet_arch.py:648] fnode 2 : {'feat_level': 4, 'inputs_offsets': [1, 6]}
I0601 05:40:02.038874 140233166403456 efficientdet_arch.py:648] fnode 3 : {'feat_level': 3, 'inputs_offsets': [0, 7]}
I0601 05:40:02.128052 140233166403456 efficientdet_arch.py:648] fnode 4 : {'feat_level': 4, 'inputs_offsets': [1, 7, 8]}
I0601 05:40:02.226508 140233166403456 efficientdet_arch.py:648] fnode 5 : {'feat_level': 5, 'inputs_offsets': [2, 6, 9]}
I0601 05:40:02.316256 140233166403456 efficientdet_arch.py:648] fnode 6 : {'feat_level': 6, 'inputs_offsets': [3, 5, 10]}
I0601 05:40:02.414864 140233166403456 efficientdet_arch.py:648] fnode 7 : {'feat_level': 7, 'inputs_offsets': [4, 11]}
I0601 05:40:02.514123 140233166403456 efficientdet_arch.py:471] building cell 2
I0601 05:40:02.514500 140233166403456 efficientdet_arch.py:648] fnode 0 : {'feat_level': 6, 'inputs_offsets': [3, 4]}
I0601 05:40:02.607001 140233166403456 efficientdet_arch.py:648] fnode 1 : {'feat_level': 5, 'inputs_offsets': [2, 5]}
I0601 05:40:02.694679 140233166403456 efficientdet_arch.py:648] fnode 2 : {'feat_level': 4, 'inputs_offsets': [1, 6]}
I0601 05:40:02.784893 140233166403456 efficientdet_arch.py:648] fnode 3 : {'feat_level': 3, 'inputs_offsets': [0, 7]}
I0601 05:40:02.871577 140233166403456 efficientdet_arch.py:648] fnode 4 : {'feat_level': 4, 'inputs_offsets': [1, 7, 8]}
I0601 05:40:02.971745 140233166403456 efficientdet_arch.py:648] fnode 5 : {'feat_level': 5, 'inputs_offsets': [2, 6, 9]}
I0601 05:40:03.074774 140233166403456 efficientdet_arch.py:648] fnode 6 : {'feat_level': 6, 'inputs_offsets': [3, 5, 10]}
I0601 05:40:03.176123 140233166403456 efficientdet_arch.py:648] fnode 7 : {'feat_level': 7, 'inputs_offsets': [4, 11]}
Parsing Inputs...
I0601 05:40:03.882147 140233166403456 efficientdet_arch.py:730] backbone+fpn params/flops = 3.791669M, 16.584498668B
Parsing Inputs...
I0601 05:40:07.417416 140233166403456 efficientdet_arch.py:735] backbone+fpn+box params/flops = 3.839117M, 18.483229676B
I0601 05:40:07.417771 140233166403456 utils.py:595] use mixed precision policy name float32
I0601 05:40:07.421836 140233166403456 det_model_fn.py:97] LR schedule method: cosine
I0601 05:40:07.972064 140233166403456 det_model_fn.py:589] Eval val with groudtruths None.
WARNING:tensorflow:From /content/automl/efficientdet/anchors.py:587: py_func (from tensorflow.python.ops.script_ops) is deprecated and will be removed in a future version.
Instructions for updating:
tf.py_func is deprecated in TF V2. Instead, there are two
    options available in V2.
    - tf.py_function takes a python function which manipulates tf eager
    tensors instead of numpy arrays. It's easy to convert a tf eager tensor to
    an ndarray (just call tensor.numpy()) but having access to eager tensors
    means `tf.py_function`s can use accelerators such as GPUs as well as
    being differentiable using a gradient tape.
    - tf.numpy_function maintains the semantics of the deprecated tf.py_func
    (it is not differentiable, and manipulates numpy arrays). It drops the
    stateful argument making all functions stateful.
    
W0601 05:40:07.981205 140233166403456 deprecation.py:323] From /content/automl/efficientdet/anchors.py:587: py_func (from tensorflow.python.ops.script_ops) is deprecated and will be removed in a future version.
Instructions for updating:
tf.py_func is deprecated in TF V2. Instead, there are two
    options available in V2.
    - tf.py_function takes a python function which manipulates tf eager
    tensors instead of numpy arrays. It's easy to convert a tf eager tensor to
    an ndarray (just call tensor.numpy()) but having access to eager tensors
    means `tf.py_function`s can use accelerators such as GPUs as well as
    being differentiable using a gradient tape.
    - tf.numpy_function maintains the semantics of the deprecated tf.py_func
    (it is not differentiable, and manipulates numpy arrays). It drops the
    stateful argument making all functions stateful.
    
INFO:tensorflow:Done calling model_fn.
I0601 05:40:08.068996 140233166403456 estimator.py:1171] Done calling model_fn.
INFO:tensorflow:Starting evaluation at 2020-06-01T05:40:08Z
I0601 05:40:08.085507 140233166403456 evaluation.py:255] Starting evaluation at 2020-06-01T05:40:08Z
INFO:tensorflow:Graph was finalized.
I0601 05:40:08.946110 140233166403456 monitored_session.py:246] Graph was finalized.
2020-06-01 05:40:08.946854: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:40:08.947376: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1561] Found device 0 with properties: 
pciBusID: 0000:00:04.0 name: Tesla T4 computeCapability: 7.5
coreClock: 1.59GHz coreCount: 40 deviceMemorySize: 14.73GiB deviceMemoryBandwidth: 298.08GiB/s
2020-06-01 05:40:08.947440: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.1
2020-06-01 05:40:08.947494: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcublas.so.10
2020-06-01 05:40:08.947518: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcufft.so.10
2020-06-01 05:40:08.947538: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcurand.so.10
2020-06-01 05:40:08.947557: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusolver.so.10
2020-06-01 05:40:08.947576: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusparse.so.10
2020-06-01 05:40:08.947596: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudnn.so.7
2020-06-01 05:40:08.947709: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:40:08.948242: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:40:08.948695: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1703] Adding visible gpu devices: 0
2020-06-01 05:40:08.948743: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1102] Device interconnect StreamExecutor with strength 1 edge matrix:
2020-06-01 05:40:08.948757: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1108]      0 
2020-06-01 05:40:08.948767: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1121] 0:   N 
2020-06-01 05:40:08.948898: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:40:08.949417: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-01 05:40:08.949866: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1247] Created TensorFlow device (/job:localhost/replica:0/task:0/device:GPU:0 with 13970 MB memory) -> physical GPU (device: 0, name: Tesla T4, pci bus id: 0000:00:04.0, compute capability: 7.5)
INFO:tensorflow:Restoring parameters from /tmp/model_dir/efficientdet-d0-finetune/model.ckpt-7
I0601 05:40:08.951066 140233166403456 saver.py:1293] Restoring parameters from /tmp/model_dir/efficientdet-d0-finetune/model.ckpt-7
INFO:tensorflow:Running local_init_op.
I0601 05:40:10.190663 140233166403456 session_manager.py:505] Running local_init_op.
INFO:tensorflow:Done running local_init_op.
I0601 05:40:10.302396 140233166403456 session_manager.py:508] Done running local_init_op.
INFO:tensorflow:Evaluation [1/7]
I0601 05:40:14.238775 140233166403456 evaluation.py:167] Evaluation [1/7]
INFO:tensorflow:Evaluation [2/7]
I0601 05:40:14.770026 140233166403456 evaluation.py:167] Evaluation [2/7]
INFO:tensorflow:Evaluation [3/7]
I0601 05:40:15.282811 140233166403456 evaluation.py:167] Evaluation [3/7]
INFO:tensorflow:Evaluation [4/7]
I0601 05:40:15.774128 140233166403456 evaluation.py:167] Evaluation [4/7]
INFO:tensorflow:Evaluation [5/7]
I0601 05:40:16.294507 140233166403456 evaluation.py:167] Evaluation [5/7]
INFO:tensorflow:Evaluation [6/7]
I0601 05:40:16.848151 140233166403456 evaluation.py:167] Evaluation [6/7]
INFO:tensorflow:Evaluation [7/7]
I0601 05:40:17.344349 140233166403456 evaluation.py:167] Evaluation [7/7]
creating index...
index created!
Loading and preparing results...
Converting ndarray to lists...
(5600, 7)
0/5600
DONE (t=0.02s)
creating index...
index created!
Running per image evaluation...
Evaluate annotation type *bbox*
DONE (t=0.36s).
Accumulating evaluation results...
DONE (t=0.13s).
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.021
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.030
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.021
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.172
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = -1.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = -1.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.146
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.270
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.270
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.270
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = -1.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = -1.000
INFO:tensorflow:Inference Time : 10.03779s
I0601 05:40:18.123470 140233166403456 evaluation.py:273] Inference Time : 10.03779s
INFO:tensorflow:Finished evaluation at 2020-06-01-05:40:18
I0601 05:40:18.123685 140233166403456 evaluation.py:276] Finished evaluation at 2020-06-01-05:40:18
INFO:tensorflow:Saving dict for global step 7: AP = 0.021243384, AP50 = 0.02968482, AP75 = 0.020640317, APl = -1.0, APm = -1.0, APs = 0.17150821, ARl = -1.0, ARm = -1.0, ARmax1 = 0.14580555, ARmax10 = 0.26992804, ARmax100 = 0.26992804, ARs = 0.26992804, box_loss = 0.0024008455, cls_loss = 1.2510093, global_step = 7, loss = 1.4653944
I0601 05:40:18.123841 140233166403456 estimator.py:2066] Saving dict for global step 7: AP = 0.021243384, AP50 = 0.02968482, AP75 = 0.020640317, APl = -1.0, APm = -1.0, APs = 0.17150821, ARl = -1.0, ARm = -1.0, ARmax1 = 0.14580555, ARmax10 = 0.26992804, ARmax100 = 0.26992804, ARs = 0.26992804, box_loss = 0.0024008455, cls_loss = 1.2510093, global_step = 7, loss = 1.4653944
INFO:tensorflow:Saving 'checkpoint_path' summary for global step 7: /tmp/model_dir/efficientdet-d0-finetune/model.ckpt-7
I0601 05:40:19.170074 140233166403456 estimator.py:2127] Saving 'checkpoint_path' summary for global step 7: /tmp/model_dir/efficientdet-d0-finetune/model.ckpt-7
INFO:tensorflow:evaluation_loop marked as finished
I0601 05:40:19.170791 140233166403456 error_handling.py:115] evaluation_loop marked as finished
I0601 05:40:19.170985 140233166403456 main.py:411] Evaluation results: {'AP': 0.021243384, 'AP50': 0.02968482, 'AP75': 0.020640317, 'APl': -1.0, 'APm': -1.0, 'APs': 0.17150821, 'ARl': -1.0, 'ARm': -1.0, 'ARmax1': 0.14580555, 'ARmax10': 0.26992804, 'ARmax100': 0.26992804, 'ARs': 0.26992804, 'box_loss': 0.0024008455, 'cls_loss': 1.2510093, 'loss': 1.4653944, 'global_step': 7}
INFO:tensorflow:/tmp/model_dir/efficientdet-d0-finetune/archive/model.ckpt-7 is not in all_model_checkpoint_paths. Manually adding it.
I0601 05:40:19.210081 140233166403456 checkpoint_management.py:102] /tmp/model_dir/efficientdet-d0-finetune/archive/model.ckpt-7 is not in all_model_checkpoint_paths. Manually adding it.
I0601 05:40:19.210609 140233166403456 utils.py:491] Copying checkpoint /tmp/model_dir/efficientdet-d0-finetune/model.ckpt-7 to /tmp/model_dir/efficientdet-d0-finetune/archive

4.4 View tensorboard for loss and accuracy.



In [0]:

    
%load_ext tensorboard
%tensorboard --logdir /tmp/model_dir/
# Notably, this is just a demo with almost zero accuracy due to very limited
# training steps, but we can see finetuning has smaller loss than training
# from scratch at the begining.