EfficientNet-PyTorch
EfficientNet-PyTorch copied to clipboard
Published
20 hours ago •
lukemelas
lukemelas
Error when converting tf checkpoint to pytorch weights
Hi, I am running a EfficientNet model built in tensorflow to classify some categories.
The version of my TF is 2.7.0.
I have already tried to save the checkpoints like this:
#1 -------
model.save_weights(filepath="ckp/EffNet-b0.ckpt")
#2 -------
model.save(filepath="ckp2/EffNet-b0.ckpt")
#3 -------
checkpoint = tf.train.Checkpoint(model)
checkpoint.save('ckp3/EffNet-b0.ckpt')
#4 -------
with tf.Graph().as_default(), tf.compat.v1.Session() as sess:
sess.run(tf.compat.v1.global_variables_initializer())
saver = tf.compat.v1.train.Saver(allow_empty=True)
saver.save(sess, 'ckp4/EffNet-b0.ckpt')
But when running the script tf_to_pytorch to convert this checkpoint to pytorch weights I am facing this error message:
$ convert_tf_to_pt$ python load_tf_weights.py --model_name efficientnet-b0 --tf_checkpoint ../pretrained_tensorflow/ckp2_27/EffNet-b0.ckpt/variables/ --output_file ../pretrained_pytorch/efficientnet-b0.pth
2021-11-27 15:50:09.949242: W tensorflow/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library 'libcudart.so.11.0'; dlerror: libcudart.so.11.0: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: /usr/local/cuda/lib64:/usr/local/cuda-10.1/lib64:
2021-11-27 15:50:09.949261: I tensorflow/stream_executor/cuda/cudart_stub.cc:29] Ignore above cudart dlerror if you do not have a GPU set up on your machine.
WARNING:tensorflow:From /home/viplabgpu/anaconda3/envs/alan_keras/lib/python3.9/site-packages/tensorflow/python/compat/v2_compat.py:111: disable_resource_variables (from tensorflow.python.ops.variable_scope) is deprecated and will be removed in a future version.
Instructions for updating:
non-resource variables are not supported in the long term
../pretrained_tensorflow/ckp2_27/EffNet-b0.ckpt/variables/
2021-11-27 15:50:14.354573: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:939] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2021-11-27 15:50:14.356128: W tensorflow/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library 'libcudart.so.11.0'; dlerror: libcudart.so.11.0: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: /usr/local/cuda/lib64:/usr/local/cuda-10.1/lib64:
2021-11-27 15:50:14.356395: W tensorflow/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library 'libcublas.so.11'; dlerror: libcublas.so.11: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: /usr/local/cuda/lib64:/usr/local/cuda-10.1/lib64:
2021-11-27 15:50:14.356844: W tensorflow/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library 'libcublasLt.so.11'; dlerror: libcublasLt.so.11: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: /usr/local/cuda/lib64:/usr/local/cuda-10.1/lib64:
2021-11-27 15:50:14.390531: W tensorflow/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library 'libcusolver.so.11'; dlerror: libcusolver.so.11: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: /usr/local/cuda/lib64:/usr/local/cuda-10.1/lib64:
2021-11-27 15:50:14.390891: W tensorflow/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library 'libcusparse.so.11'; dlerror: libcusparse.so.11: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: /usr/local/cuda/lib64:/usr/local/cuda-10.1/lib64:
2021-11-27 15:50:14.391725: W tensorflow/core/common_runtime/gpu/gpu_device.cc:1850] Cannot dlopen some GPU libraries. Please make sure the missing libraries mentioned above are installed properly if you would like to use GPU. Follow the guide at https://www.tensorflow.org/install/gpu for how to download and setup the required libraries for your platform.
Skipping registering GPU devices...
2021-11-27 15:50:14.392514: I tensorflow/core/platform/cpu_feature_guard.cc:151] This TensorFlow binary is optimized with oneAPI Deep Neural Network Library (oneDNN) to use the following CPU instructions in performance-critical operations: AVX2 FMA
To enable them in other operations, rebuild TensorFlow with the appropriate compiler flags.
WARNING:tensorflow:From /EfficientNet-PyTorch/tf_to_pytorch/convert_tf_to_pt/original_tf/eval_ckpt_main.py:120: DatasetV1.make_one_shot_iterator (from tensorflow.python.data.ops.dataset_ops) is deprecated and will be removed in a future version.
Instructions for updating:
This is a deprecated API that should only be used in TF 1 graph mode and legacy TF 2 graph mode available through `tf.compat.v1`. In all other situations -- namely, eager mode and inside `tf.function` -- you can consume dataset elements using `for elem in dataset: ...` or by explicitly creating iterator via `iterator = iter(dataset)` and fetching its elements via `values = next(iterator)`. Furthermore, this API is not available in TF 2. During the transition from TF 1 to TF 2 you can use `tf.compat.v1.data.make_one_shot_iterator(dataset)` to create a TF 1 graph mode style iterator for a dataset created through TF 2 APIs. Note that this should be a transient state of your code base as there are in general no guarantees about the interoperability of TF 1 and TF 2 code.
../pretrained_tensorflow/ckp2_27/EffNet-b0.ckpt/variables/
Traceback (most recent call last):
File "/EfficientNet-PyTorch/tf_to_pytorch/convert_tf_to_pt/load_tf_weights.py", line 166, in <module>
load_and_save_temporary_tensorflow_model(args.model_name, args.tf_checkpoint)
File "/EfficientNet-PyTorch/tf_to_pytorch/convert_tf_to_pt/load_tf_weights.py", line 137, in load_and_save_temporary_tensorflow_model
eval_ckpt_driver.restore_model(sess, model_ckpt)
File "/EfficientNet-PyTorch/tf_to_pytorch/convert_tf_to_pt/original_tf/eval_ckpt_main.py", line 91, in restore_model
saver.restore(sess, checkpoint)
File "/home/viplabgpu/anaconda3/envs/alan_keras/lib/python3.9/site-packages/tensorflow/python/training/saver.py", line 1392, in restore
raise ValueError("Can't load save_path when it is None.")
ValueError: Can't load save_path when it is None.
(alan_keras) viplabgpu@viplabgpu-desktop:~/Documentos/alan/EfficientNet-PyTorch/tf_to_pytorch/convert_tf_to_pt$ python load_tf_weights.py --model_name efficientnet-b0 --tf_checkpoint ../pretrained_tensorflow/ckp2_27/EffNet-b0.ckpt/variables/ --output_file ../pretrained_pytorch/efficientnet-b0.pth
2021-11-27 15:51:20.668813: W tensorflow/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library 'libcudart.so.11.0'; dlerror: libcudart.so.11.0: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: /usr/local/cuda/lib64:/usr/local/cuda-10.1/lib64:
2021-11-27 15:51:20.668834: I tensorflow/stream_executor/cuda/cudart_stub.cc:29] Ignore above cudart dlerror if you do not have a GPU set up on your machine.
WARNING:tensorflow:From /home/viplabgpu/anaconda3/envs/alan_keras/lib/python3.9/site-packages/tensorflow/python/compat/v2_compat.py:111: disable_resource_variables (from tensorflow.python.ops.variable_scope) is deprecated and will be removed in a future version.
Instructions for updating:
non-resource variables are not supported in the long term
../pretrained_tensorflow/ckp2_27/EffNet-b0.ckpt/variables/
2021-11-27 15:51:21.970751: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:939] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2021-11-27 15:51:21.971187: W tensorflow/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library 'libcudart.so.11.0'; dlerror: libcudart.so.11.0: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: /usr/local/cuda/lib64:/usr/local/cuda-10.1/lib64:
2021-11-27 15:51:21.971239: W tensorflow/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library 'libcublas.so.11'; dlerror: libcublas.so.11: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: /usr/local/cuda/lib64:/usr/local/cuda-10.1/lib64:
2021-11-27 15:51:21.971285: W tensorflow/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library 'libcublasLt.so.11'; dlerror: libcublasLt.so.11: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: /usr/local/cuda/lib64:/usr/local/cuda-10.1/lib64:
2021-11-27 15:51:21.972549: W tensorflow/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library 'libcusolver.so.11'; dlerror: libcusolver.so.11: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: /usr/local/cuda/lib64:/usr/local/cuda-10.1/lib64:
2021-11-27 15:51:21.972600: W tensorflow/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library 'libcusparse.so.11'; dlerror: libcusparse.so.11: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: /usr/local/cuda/lib64:/usr/local/cuda-10.1/lib64:
2021-11-27 15:51:21.972668: W tensorflow/core/common_runtime/gpu/gpu_device.cc:1850] Cannot dlopen some GPU libraries. Please make sure the missing libraries mentioned above are installed properly if you would like to use GPU. Follow the guide at https://www.tensorflow.org/install/gpu for how to download and setup the required libraries for your platform.
Skipping registering GPU devices...
2021-11-27 15:51:21.972899: I tensorflow/core/platform/cpu_feature_guard.cc:151] This TensorFlow binary is optimized with oneAPI Deep Neural Network Library (oneDNN) to use the following CPU instructions in performance-critical operations: AVX2 FMA
To enable them in other operations, rebuild TensorFlow with the appropriate compiler flags.
WARNING:tensorflow:From /EfficientNet-PyTorch/tf_to_pytorch/convert_tf_to_pt/original_tf/eval_ckpt_main.py:120: DatasetV1.make_one_shot_iterator (from tensorflow.python.data.ops.dataset_ops) is deprecated and will be removed in a future version.
Instructions for updating:
This is a deprecated API that should only be used in TF 1 graph mode and legacy TF 2 graph mode available through `tf.compat.v1`. In all other situations -- namely, eager mode and inside `tf.function` -- you can consume dataset elements using `for elem in dataset: ...` or by explicitly creating iterator via `iterator = iter(dataset)` and fetching its elements via `values = next(iterator)`. Furthermore, this API is not available in TF 2. During the transition from TF 1 to TF 2 you can use `tf.compat.v1.data.make_one_shot_iterator(dataset)` to create a TF 1 graph mode style iterator for a dataset created through TF 2 APIs. Note that this should be a transient state of your code base as there are in general no guarantees about the interoperability of TF 1 and TF 2 code.
../pretrained_tensorflow/ckp2_27/EffNet-b0.ckpt/variables/
2021-11-27 15:51:25.229401: W tensorflow/core/framework/op_kernel.cc:1745] OP_REQUIRES failed at save_restore_v2_ops.cc:207 : NOT_FOUND: Key efficientnet-b0/blocks_0/conv2d/kernel/ExponentialMovingAverage not found in checkpoint
WARNING:tensorflow:Restoring an object-based checkpoint using a name-based saver. This may be somewhat fragile, and will re-build the Saver. Instead, consider loading object-based checkpoints using tf.train.Checkpoint().
WARNING:tensorflow:Restoring an object-based checkpoint using a name-based saver. This may be somewhat fragile, and will re-build the Saver. Instead, consider loading object-based checkpoints using tf.train.Checkpoint().
Traceback (most recent call last):
File "/home/viplabgpu/anaconda3/envs/alan_keras/lib/python3.9/site-packages/tensorflow/python/client/session.py", line 1380, in _do_call
return fn(*args)
File "/home/viplabgpu/anaconda3/envs/alan_keras/lib/python3.9/site-packages/tensorflow/python/client/session.py", line 1363, in _run_fn
return self._call_tf_sessionrun(options, feed_dict, fetch_list,
File "/home/viplabgpu/anaconda3/envs/alan_keras/lib/python3.9/site-packages/tensorflow/python/client/session.py", line 1456, in _call_tf_sessionrun
return tf_session.TF_SessionRun_wrapper(self._session, options, feed_dict,
tensorflow.python.framework.errors_impl.NotFoundError: Key efficientnet-b0/blocks_0/conv2d/kernel/ExponentialMovingAverage not found in checkpoint
[[{{node save/RestoreV2}}]]
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "/home/viplabgpu/anaconda3/envs/alan_keras/lib/python3.9/site-packages/tensorflow/python/training/saver.py", line 1404, in restore
sess.run(self.saver_def.restore_op_name,
File "/home/viplabgpu/anaconda3/envs/alan_keras/lib/python3.9/site-packages/tensorflow/python/client/session.py", line 970, in run
result = self._run(None, fetches, feed_dict, options_ptr,
File "/home/viplabgpu/anaconda3/envs/alan_keras/lib/python3.9/site-packages/tensorflow/python/client/session.py", line 1193, in _run
results = self._do_run(handle, final_targets, final_fetches,
File "/home/viplabgpu/anaconda3/envs/alan_keras/lib/python3.9/site-packages/tensorflow/python/client/session.py", line 1373, in _do_run
return self._do_call(_run_fn, feeds, fetches, targets, options,
File "/home/viplabgpu/anaconda3/envs/alan_keras/lib/python3.9/site-packages/tensorflow/python/client/session.py", line 1399, in _do_call
raise type(e)(node_def, op, message) # pylint: disable=no-value-for-parameter
tensorflow.python.framework.errors_impl.NotFoundError: Key efficientnet-b0/blocks_0/conv2d/kernel/ExponentialMovingAverage not found in checkpoint
[[node save/RestoreV2
(defined at /home/viplabgpu/anaconda3/envs/alan_keras/lib/python3.9/site-packages/tensorflow/python/util/tf_stack.py:193)
]]
Errors may have originated from an input operation.
Input Source operations connected to node save/RestoreV2:
In[0] save/Const:
In[1] save/RestoreV2/tensor_names:
In[2] save/RestoreV2/shape_and_slices:
Operation defined at: (most recent call last)
>>> File "/EfficientNet-PyTorch/tf_to_pytorch/convert_tf_to_pt/load_tf_weights.py", line 166, in <module>
>>> load_and_save_temporary_tensorflow_model(args.model_name, args.tf_checkpoint)
>>>
>>> File "/EfficientNet-PyTorch/tf_to_pytorch/convert_tf_to_pt/load_tf_weights.py", line 137, in load_and_save_temporary_tensorflow_model
>>> eval_ckpt_driver.restore_model(sess, model_ckpt)
>>>
>>> File "/EfficientNet-PyTorch/tf_to_pytorch/convert_tf_to_pt/original_tf/eval_ckpt_main.py", line 90, in restore_model
>>> saver = tf.compat.v1.train.Saver(var_dict, max_to_keep=1)
>>>
>>> File "/home/viplabgpu/anaconda3/envs/alan_keras/lib/python3.9/site-packages/tensorflow/python/training/saver.py", line 923, in __init__
>>> self.build()
>>>
>>> File "/home/viplabgpu/anaconda3/envs/alan_keras/lib/python3.9/site-packages/tensorflow/python/training/saver.py", line 935, in build
>>> self._build(self._filename, build_save=True, build_restore=True)
>>>
>>> File "/home/viplabgpu/anaconda3/envs/alan_keras/lib/python3.9/site-packages/tensorflow/python/training/saver.py", line 963, in _build
>>> self.saver_def = self._builder._build_internal( # pylint: disable=protected-access
>>>
>>> File "/home/viplabgpu/anaconda3/envs/alan_keras/lib/python3.9/site-packages/tensorflow/python/training/saver.py", line 533, in _build_internal
>>> restore_op = self._AddRestoreOps(filename_tensor, saveables,
>>>
>>> File "/home/viplabgpu/anaconda3/envs/alan_keras/lib/python3.9/site-packages/tensorflow/python/training/saver.py", line 353, in _AddRestoreOps
>>> all_tensors = self.bulk_restore(filename_tensor, saveables, preferred_shard,
>>>
>>> File "/home/viplabgpu/anaconda3/envs/alan_keras/lib/python3.9/site-packages/tensorflow/python/training/saver.py", line 601, in bulk_restore
>>> return io_ops.restore_v2(filename_tensor, names, slices, dtypes)
>>>
>>> File "/home/viplabgpu/anaconda3/envs/alan_keras/lib/python3.9/site-packages/tensorflow/python/ops/gen_io_ops.py", line 1501, in restore_v2
>>> _, _, _op, _outputs = _op_def_library._apply_op_helper(
>>>
>>> File "/home/viplabgpu/anaconda3/envs/alan_keras/lib/python3.9/site-packages/tensorflow/python/framework/op_def_library.py", line 744, in _apply_op_helper
>>> op = g._create_op_internal(op_type_name, inputs, dtypes=None,
>>>
>>> File "/home/viplabgpu/anaconda3/envs/alan_keras/lib/python3.9/site-packages/tensorflow/python/framework/ops.py", line 3697, in _create_op_internal
>>> ret = Operation(
>>>
>>> File "/home/viplabgpu/anaconda3/envs/alan_keras/lib/python3.9/site-packages/tensorflow/python/framework/ops.py", line 2101, in __init__
>>> self._traceback = tf_stack.extract_stack_for_node(self._c_op)
>>>
>>> File "/home/viplabgpu/anaconda3/envs/alan_keras/lib/python3.9/site-packages/tensorflow/python/util/tf_stack.py", line 193, in extract_stack_for_node
>>> return _tf_stack.extract_stack_for_node(
>>>
Original stack trace for 'save/RestoreV2':
File "/EfficientNet-PyTorch/tf_to_pytorch/convert_tf_to_pt/load_tf_weights.py", line 166, in <module>
load_and_save_temporary_tensorflow_model(args.model_name, args.tf_checkpoint)
File "/EfficientNet-PyTorch/tf_to_pytorch/convert_tf_to_pt/load_tf_weights.py", line 137, in load_and_save_temporary_tensorflow_model
eval_ckpt_driver.restore_model(sess, model_ckpt)
File "/EfficientNet-PyTorch/tf_to_pytorch/convert_tf_to_pt/original_tf/eval_ckpt_main.py", line 90, in restore_model
saver = tf.compat.v1.train.Saver(var_dict, max_to_keep=1)
File "/home/viplabgpu/anaconda3/envs/alan_keras/lib/python3.9/site-packages/tensorflow/python/training/saver.py", line 923, in __init__
self.build()
File "/home/viplabgpu/anaconda3/envs/alan_keras/lib/python3.9/site-packages/tensorflow/python/training/saver.py", line 935, in build
self._build(self._filename, build_save=True, build_restore=True)
File "/home/viplabgpu/anaconda3/envs/alan_keras/lib/python3.9/site-packages/tensorflow/python/training/saver.py", line 963, in _build
self.saver_def = self._builder._build_internal( # pylint: disable=protected-access
File "/home/viplabgpu/anaconda3/envs/alan_keras/lib/python3.9/site-packages/tensorflow/python/training/saver.py", line 533, in _build_internal
restore_op = self._AddRestoreOps(filename_tensor, saveables,
File "/home/viplabgpu/anaconda3/envs/alan_keras/lib/python3.9/site-packages/tensorflow/python/training/saver.py", line 353, in _AddRestoreOps
all_tensors = self.bulk_restore(filename_tensor, saveables, preferred_shard,
File "/home/viplabgpu/anaconda3/envs/alan_keras/lib/python3.9/site-packages/tensorflow/python/training/saver.py", line 601, in bulk_restore
return io_ops.restore_v2(filename_tensor, names, slices, dtypes)
File "/home/viplabgpu/anaconda3/envs/alan_keras/lib/python3.9/site-packages/tensorflow/python/ops/gen_io_ops.py", line 1501, in restore_v2
_, _, _op, _outputs = _op_def_library._apply_op_helper(
File "/home/viplabgpu/anaconda3/envs/alan_keras/lib/python3.9/site-packages/tensorflow/python/framework/op_def_library.py", line 744, in _apply_op_helper
op = g._create_op_internal(op_type_name, inputs, dtypes=None,
File "/home/viplabgpu/anaconda3/envs/alan_keras/lib/python3.9/site-packages/tensorflow/python/framework/ops.py", line 3697, in _create_op_internal
ret = Operation(
File "/home/viplabgpu/anaconda3/envs/alan_keras/lib/python3.9/site-packages/tensorflow/python/framework/ops.py", line 2101, in __init__
self._traceback = tf_stack.extract_stack_for_node(self._c_op)
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "/EfficientNet-PyTorch/tf_to_pytorch/convert_tf_to_pt/load_tf_weights.py", line 166, in <module>
load_and_save_temporary_tensorflow_model(args.model_name, args.tf_checkpoint)
File "/EfficientNet-PyTorch/tf_to_pytorch/convert_tf_to_pt/load_tf_weights.py", line 137, in load_and_save_temporary_tensorflow_model
eval_ckpt_driver.restore_model(sess, model_ckpt)
File "/EfficientNet-PyTorch/tf_to_pytorch/convert_tf_to_pt/original_tf/eval_ckpt_main.py", line 91, in restore_model
saver.restore(sess, checkpoint)
File "/home/viplabgpu/anaconda3/envs/alan_keras/lib/python3.9/site-packages/tensorflow/python/training/saver.py", line 1430, in restore
self._object_restore_saver = saver_from_object_based_checkpoint(
File "/home/viplabgpu/anaconda3/envs/alan_keras/lib/python3.9/site-packages/tensorflow/python/training/saver.py", line 1809, in saver_from_object_based_checkpoint
raise errors.NotFoundError(
tensorflow.python.framework.errors_impl.NotFoundError:
Existing variables not in the checkpoint: efficientnet-b0/blocks_0/conv2d/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_0/depthwise_conv2d/depthwise_kernel/ExponentialMovingAverage, efficientnet-b0/blocks_0/se/conv2d/bias/ExponentialMovingAverage, efficientnet-b0/blocks_0/se/conv2d/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_0/se/conv2d_1/bias/ExponentialMovingAverage, efficientnet-b0/blocks_0/se/conv2d_1/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_0/tpu_batch_normalization/beta/ExponentialMovingAverage, efficientnet-b0/blocks_0/tpu_batch_normalization/gamma/ExponentialMovingAverage, efficientnet-b0/blocks_0/tpu_batch_normalization/moving_mean/ExponentialMovingAverage, efficientnet-b0/blocks_0/tpu_batch_normalization/moving_variance/ExponentialMovingAverage, efficientnet-b0/blocks_0/tpu_batch_normalization_1/beta/ExponentialMovingAverage, efficientnet-b0/blocks_0/tpu_batch_normalization_1/gamma/ExponentialMovingAverage, efficientnet-b0/blocks_0/tpu_batch_normalization_1/moving_mean/ExponentialMovingAverage, efficientnet-b0/blocks_0/tpu_batch_normalization_1/moving_variance/ExponentialMovingAverage, efficientnet-b0/blocks_1/conv2d/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_1/conv2d_1/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_1/depthwise_conv2d/depthwise_kernel/ExponentialMovingAverage, efficientnet-b0/blocks_1/se/conv2d/bias/ExponentialMovingAverage, efficientnet-b0/blocks_1/se/conv2d/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_1/se/conv2d_1/bias/ExponentialMovingAverage, efficientnet-b0/blocks_1/se/conv2d_1/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_1/tpu_batch_normalization/beta/ExponentialMovingAverage, efficientnet-b0/blocks_1/tpu_batch_normalization/gamma/ExponentialMovingAverage, efficientnet-b0/blocks_1/tpu_batch_normalization/moving_mean/ExponentialMovingAverage, efficientnet-b0/blocks_1/tpu_batch_normalization/moving_variance/ExponentialMovingAverage, efficientnet-b0/blocks_1/tpu_batch_normalization_1/beta/ExponentialMovingAverage, efficientnet-b0/blocks_1/tpu_batch_normalization_1/gamma/ExponentialMovingAverage, efficientnet-b0/blocks_1/tpu_batch_normalization_1/moving_mean/ExponentialMovingAverage, efficientnet-b0/blocks_1/tpu_batch_normalization_1/moving_variance/ExponentialMovingAverage, efficientnet-b0/blocks_1/tpu_batch_normalization_2/beta/ExponentialMovingAverage, efficientnet-b0/blocks_1/tpu_batch_normalization_2/gamma/ExponentialMovingAverage, efficientnet-b0/blocks_1/tpu_batch_normalization_2/moving_mean/ExponentialMovingAverage, efficientnet-b0/blocks_1/tpu_batch_normalization_2/moving_variance/ExponentialMovingAverage, efficientnet-b0/blocks_10/conv2d/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_10/conv2d_1/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_10/depthwise_conv2d/depthwise_kernel/ExponentialMovingAverage, efficientnet-b0/blocks_10/se/conv2d/bias/ExponentialMovingAverage, efficientnet-b0/blocks_10/se/conv2d/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_10/se/conv2d_1/bias/ExponentialMovingAverage, efficientnet-b0/blocks_10/se/conv2d_1/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_10/tpu_batch_normalization/beta/ExponentialMovingAverage, efficientnet-b0/blocks_10/tpu_batch_normalization/gamma/ExponentialMovingAverage, efficientnet-b0/blocks_10/tpu_batch_normalization/moving_mean/ExponentialMovingAverage, efficientnet-b0/blocks_10/tpu_batch_normalization/moving_variance/ExponentialMovingAverage, efficientnet-b0/blocks_10/tpu_batch_normalization_1/beta/ExponentialMovingAverage, efficientnet-b0/blocks_10/tpu_batch_normalization_1/gamma/ExponentialMovingAverage, efficientnet-b0/blocks_10/tpu_batch_normalization_1/moving_mean/ExponentialMovingAverage, efficientnet-b0/blocks_10/tpu_batch_normalization_1/moving_variance/ExponentialMovingAverage, efficientnet-b0/blocks_10/tpu_batch_normalization_2/beta/ExponentialMovingAverage, efficientnet-b0/blocks_10/tpu_batch_normalization_2/gamma/ExponentialMovingAverage, efficientnet-b0/blocks_10/tpu_batch_normalization_2/moving_mean/ExponentialMovingAverage, efficientnet-b0/blocks_10/tpu_batch_normalization_2/moving_variance/ExponentialMovingAverage, efficientnet-b0/blocks_11/conv2d/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_11/conv2d_1/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_11/depthwise_conv2d/depthwise_kernel/ExponentialMovingAverage, efficientnet-b0/blocks_11/se/conv2d/bias/ExponentialMovingAverage, efficientnet-b0/blocks_11/se/conv2d/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_11/se/conv2d_1/bias/ExponentialMovingAverage, efficientnet-b0/blocks_11/se/conv2d_1/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_11/tpu_batch_normalization/beta/ExponentialMovingAverage, efficientnet-b0/blocks_11/tpu_batch_normalization/gamma/ExponentialMovingAverage, efficientnet-b0/blocks_11/tpu_batch_normalization/moving_mean/ExponentialMovingAverage, efficientnet-b0/blocks_11/tpu_batch_normalization/moving_variance/ExponentialMovingAverage, efficientnet-b0/blocks_11/tpu_batch_normalization_1/beta/ExponentialMovingAverage, efficientnet-b0/blocks_11/tpu_batch_normalization_1/gamma/ExponentialMovingAverage, efficientnet-b0/blocks_11/tpu_batch_normalization_1/moving_mean/ExponentialMovingAverage, efficientnet-b0/blocks_11/tpu_batch_normalization_1/moving_variance/ExponentialMovingAverage, efficientnet-b0/blocks_11/tpu_batch_normalization_2/beta/ExponentialMovingAverage, efficientnet-b0/blocks_11/tpu_batch_normalization_2/gamma/ExponentialMovingAverage, efficientnet-b0/blocks_11/tpu_batch_normalization_2/moving_mean/ExponentialMovingAverage, efficientnet-b0/blocks_11/tpu_batch_normalization_2/moving_variance/ExponentialMovingAverage, efficientnet-b0/blocks_12/conv2d/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_12/conv2d_1/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_12/depthwise_conv2d/depthwise_kernel/ExponentialMovingAverage, efficientnet-b0/blocks_12/se/conv2d/bias/ExponentialMovingAverage, efficientnet-b0/blocks_12/se/conv2d/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_12/se/conv2d_1/bias/ExponentialMovingAverage, efficientnet-b0/blocks_12/se/conv2d_1/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_12/tpu_batch_normalization/beta/ExponentialMovingAverage, efficientnet-b0/blocks_12/tpu_batch_normalization/gamma/ExponentialMovingAverage, efficientnet-b0/blocks_12/tpu_batch_normalization/moving_mean/ExponentialMovingAverage, efficientnet-b0/blocks_12/tpu_batch_normalization/moving_variance/ExponentialMovingAverage, efficientnet-b0/blocks_12/tpu_batch_normalization_1/beta/ExponentialMovingAverage, efficientnet-b0/blocks_12/tpu_batch_normalization_1/gamma/ExponentialMovingAverage, efficientnet-b0/blocks_12/tpu_batch_normalization_1/moving_mean/ExponentialMovingAverage, efficientnet-b0/blocks_12/tpu_batch_normalization_1/moving_variance/ExponentialMovingAverage, efficientnet-b0/blocks_12/tpu_batch_normalization_2/beta/ExponentialMovingAverage, efficientnet-b0/blocks_12/tpu_batch_normalization_2/gamma/ExponentialMovingAverage, efficientnet-b0/blocks_12/tpu_batch_normalization_2/moving_mean/ExponentialMovingAverage, efficientnet-b0/blocks_12/tpu_batch_normalization_2/moving_variance/ExponentialMovingAverage, efficientnet-b0/blocks_13/conv2d/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_13/conv2d_1/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_13/depthwise_conv2d/depthwise_kernel/ExponentialMovingAverage, efficientnet-b0/blocks_13/se/conv2d/bias/ExponentialMovingAverage, efficientnet-b0/blocks_13/se/conv2d/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_13/se/conv2d_1/bias/ExponentialMovingAverage, efficientnet-b0/blocks_13/se/conv2d_1/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_13/tpu_batch_normalization/beta/ExponentialMovingAverage, efficientnet-b0/blocks_13/tpu_batch_normalization/gamma/ExponentialMovingAverage, efficientnet-b0/blocks_13/tpu_batch_normalization/moving_mean/ExponentialMovingAverage, efficientnet-b0/blocks_13/tpu_batch_normalization/moving_variance/ExponentialMovingAverage, efficientnet-b0/blocks_13/tpu_batch_normalization_1/beta/ExponentialMovingAverage, efficientnet-b0/blocks_13/tpu_batch_normalization_1/gamma/ExponentialMovingAverage, efficientnet-b0/blocks_13/tpu_batch_normalization_1/moving_mean/ExponentialMovingAverage, efficientnet-b0/blocks_13/tpu_batch_normalization_1/moving_variance/ExponentialMovingAverage, efficientnet-b0/blocks_13/tpu_batch_normalization_2/beta/ExponentialMovingAverage, efficientnet-b0/blocks_13/tpu_batch_normalization_2/gamma/ExponentialMovingAverage, efficientnet-b0/blocks_13/tpu_batch_normalization_2/moving_mean/ExponentialMovingAverage, efficientnet-b0/blocks_13/tpu_batch_normalization_2/moving_variance/ExponentialMovingAverage, efficientnet-b0/blocks_14/conv2d/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_14/conv2d_1/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_14/depthwise_conv2d/depthwise_kernel/ExponentialMovingAverage, efficientnet-b0/blocks_14/se/conv2d/bias/ExponentialMovingAverage, efficientnet-b0/blocks_14/se/conv2d/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_14/se/conv2d_1/bias/ExponentialMovingAverage, efficientnet-b0/blocks_14/se/conv2d_1/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_14/tpu_batch_normalization/beta/ExponentialMovingAverage, efficientnet-b0/blocks_14/tpu_batch_normalization/gamma/ExponentialMovingAverage, efficientnet-b0/blocks_14/tpu_batch_normalization/moving_mean/ExponentialMovingAverage, efficientnet-b0/blocks_14/tpu_batch_normalization/moving_variance/ExponentialMovingAverage, efficientnet-b0/blocks_14/tpu_batch_normalization_1/beta/ExponentialMovingAverage, efficientnet-b0/blocks_14/tpu_batch_normalization_1/gamma/ExponentialMovingAverage, efficientnet-b0/blocks_14/tpu_batch_normalization_1/moving_mean/ExponentialMovingAverage, efficientnet-b0/blocks_14/tpu_batch_normalization_1/moving_variance/ExponentialMovingAverage, efficientnet-b0/blocks_14/tpu_batch_normalization_2/beta/ExponentialMovingAverage, efficientnet-b0/blocks_14/tpu_batch_normalization_2/gamma/ExponentialMovingAverage, efficientnet-b0/blocks_14/tpu_batch_normalization_2/moving_mean/ExponentialMovingAverage, efficientnet-b0/blocks_14/tpu_batch_normalization_2/moving_variance/ExponentialMovingAverage, efficientnet-b0/blocks_15/conv2d/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_15/conv2d_1/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_15/depthwise_conv2d/depthwise_kernel/ExponentialMovingAverage, efficientnet-b0/blocks_15/se/conv2d/bias/ExponentialMovingAverage, efficientnet-b0/blocks_15/se/conv2d/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_15/se/conv2d_1/bias/ExponentialMovingAverage, efficientnet-b0/blocks_15/se/conv2d_1/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_15/tpu_batch_normalization/beta/ExponentialMovingAverage, efficientnet-b0/blocks_15/tpu_batch_normalization/gamma/ExponentialMovingAverage, efficientnet-b0/blocks_15/tpu_batch_normalization/moving_mean/ExponentialMovingAverage, efficientnet-b0/blocks_15/tpu_batch_normalization/moving_variance/ExponentialMovingAverage, efficientnet-b0/blocks_15/tpu_batch_normalization_1/beta/ExponentialMovingAverage, efficientnet-b0/blocks_15/tpu_batch_normalization_1/gamma/ExponentialMovingAverage, efficientnet-b0/blocks_15/tpu_batch_normalization_1/moving_mean/ExponentialMovingAverage, efficientnet-b0/blocks_15/tpu_batch_normalization_1/moving_variance/ExponentialMovingAverage, efficientnet-b0/blocks_15/tpu_batch_normalization_2/beta/ExponentialMovingAverage, efficientnet-b0/blocks_15/tpu_batch_normalization_2/gamma/ExponentialMovingAverage, efficientnet-b0/blocks_15/tpu_batch_normalization_2/moving_mean/ExponentialMovingAverage, efficientnet-b0/blocks_15/tpu_batch_normalization_2/moving_variance/ExponentialMovingAverage, efficientnet-b0/blocks_2/conv2d/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_2/conv2d_1/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_2/depthwise_conv2d/depthwise_kernel/ExponentialMovingAverage, efficientnet-b0/blocks_2/se/conv2d/bias/ExponentialMovingAverage, efficientnet-b0/blocks_2/se/conv2d/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_2/se/conv2d_1/bias/ExponentialMovingAverage, efficientnet-b0/blocks_2/se/conv2d_1/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_2/tpu_batch_normalization/beta/ExponentialMovingAverage, efficientnet-b0/blocks_2/tpu_batch_normalization/gamma/ExponentialMovingAverage, efficientnet-b0/blocks_2/tpu_batch_normalization/moving_mean/ExponentialMovingAverage, efficientnet-b0/blocks_2/tpu_batch_normalization/moving_variance/ExponentialMovingAverage, efficientnet-b0/blocks_2/tpu_batch_normalization_1/beta/ExponentialMovingAverage, efficientnet-b0/blocks_2/tpu_batch_normalization_1/gamma/ExponentialMovingAverage, efficientnet-b0/blocks_2/tpu_batch_normalization_1/moving_mean/ExponentialMovingAverage, efficientnet-b0/blocks_2/tpu_batch_normalization_1/moving_variance/ExponentialMovingAverage, efficientnet-b0/blocks_2/tpu_batch_normalization_2/beta/ExponentialMovingAverage, efficientnet-b0/blocks_2/tpu_batch_normalization_2/gamma/ExponentialMovingAverage, efficientnet-b0/blocks_2/tpu_batch_normalization_2/moving_mean/ExponentialMovingAverage, efficientnet-b0/blocks_2/tpu_batch_normalization_2/moving_variance/ExponentialMovingAverage, efficientnet-b0/blocks_3/conv2d/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_3/conv2d_1/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_3/depthwise_conv2d/depthwise_kernel/ExponentialMovingAverage, efficientnet-b0/blocks_3/se/conv2d/bias/ExponentialMovingAverage, efficientnet-b0/blocks_3/se/conv2d/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_3/se/conv2d_1/bias/ExponentialMovingAverage, efficientnet-b0/blocks_3/se/conv2d_1/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_3/tpu_batch_normalization/beta/ExponentialMovingAverage, efficientnet-b0/blocks_3/tpu_batch_normalization/gamma/ExponentialMovingAverage, efficientnet-b0/blocks_3/tpu_batch_normalization/moving_mean/ExponentialMovingAverage, efficientnet-b0/blocks_3/tpu_batch_normalization/moving_variance/ExponentialMovingAverage, efficientnet-b0/blocks_3/tpu_batch_normalization_1/beta/ExponentialMovingAverage, efficientnet-b0/blocks_3/tpu_batch_normalization_1/gamma/ExponentialMovingAverage, efficientnet-b0/blocks_3/tpu_batch_normalization_1/moving_mean/ExponentialMovingAverage, efficientnet-b0/blocks_3/tpu_batch_normalization_1/moving_variance/ExponentialMovingAverage, efficientnet-b0/blocks_3/tpu_batch_normalization_2/beta/ExponentialMovingAverage, efficientnet-b0/blocks_3/tpu_batch_normalization_2/gamma/ExponentialMovingAverage, efficientnet-b0/blocks_3/tpu_batch_normalization_2/moving_mean/ExponentialMovingAverage, efficientnet-b0/blocks_3/tpu_batch_normalization_2/moving_variance/ExponentialMovingAverage, efficientnet-b0/blocks_4/conv2d/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_4/conv2d_1/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_4/depthwise_conv2d/depthwise_kernel/ExponentialMovingAverage, efficientnet-b0/blocks_4/se/conv2d/bias/ExponentialMovingAverage, efficientnet-b0/blocks_4/se/conv2d/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_4/se/conv2d_1/bias/ExponentialMovingAverage, efficientnet-b0/blocks_4/se/conv2d_1/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_4/tpu_batch_normalization/beta/ExponentialMovingAverage, efficientnet-b0/blocks_4/tpu_batch_normalization/gamma/ExponentialMovingAverage, efficientnet-b0/blocks_4/tpu_batch_normalization/moving_mean/ExponentialMovingAverage, efficientnet-b0/blocks_4/tpu_batch_normalization/moving_variance/ExponentialMovingAverage, efficientnet-b0/blocks_4/tpu_batch_normalization_1/beta/ExponentialMovingAverage, efficientnet-b0/blocks_4/tpu_batch_normalization_1/gamma/ExponentialMovingAverage, efficientnet-b0/blocks_4/tpu_batch_normalization_1/moving_mean/ExponentialMovingAverage, efficientnet-b0/blocks_4/tpu_batch_normalization_1/moving_variance/ExponentialMovingAverage, efficientnet-b0/blocks_4/tpu_batch_normalization_2/beta/ExponentialMovingAverage, efficientnet-b0/blocks_4/tpu_batch_normalization_2/gamma/ExponentialMovingAverage, efficientnet-b0/blocks_4/tpu_batch_normalization_2/moving_mean/ExponentialMovingAverage, efficientnet-b0/blocks_4/tpu_batch_normalization_2/moving_variance/ExponentialMovingAverage, efficientnet-b0/blocks_5/conv2d/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_5/conv2d_1/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_5/depthwise_conv2d/depthwise_kernel/ExponentialMovingAverage, efficientnet-b0/blocks_5/se/conv2d/bias/ExponentialMovingAverage, efficientnet-b0/blocks_5/se/conv2d/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_5/se/conv2d_1/bias/ExponentialMovingAverage, efficientnet-b0/blocks_5/se/conv2d_1/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_5/tpu_batch_normalization/beta/ExponentialMovingAverage, efficientnet-b0/blocks_5/tpu_batch_normalization/gamma/ExponentialMovingAverage, efficientnet-b0/blocks_5/tpu_batch_normalization/moving_mean/ExponentialMovingAverage, efficientnet-b0/blocks_5/tpu_batch_normalization/moving_variance/ExponentialMovingAverage, efficientnet-b0/blocks_5/tpu_batch_normalization_1/beta/ExponentialMovingAverage, efficientnet-b0/blocks_5/tpu_batch_normalization_1/gamma/ExponentialMovingAverage, efficientnet-b0/blocks_5/tpu_batch_normalization_1/moving_mean/ExponentialMovingAverage, efficientnet-b0/blocks_5/tpu_batch_normalization_1/moving_variance/ExponentialMovingAverage, efficientnet-b0/blocks_5/tpu_batch_normalization_2/beta/ExponentialMovingAverage, efficientnet-b0/blocks_5/tpu_batch_normalization_2/gamma/ExponentialMovingAverage, efficientnet-b0/blocks_5/tpu_batch_normalization_2/moving_mean/ExponentialMovingAverage, efficientnet-b0/blocks_5/tpu_batch_normalization_2/moving_variance/ExponentialMovingAverage, efficientnet-b0/blocks_6/conv2d/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_6/conv2d_1/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_6/depthwise_conv2d/depthwise_kernel/ExponentialMovingAverage, efficientnet-b0/blocks_6/se/conv2d/bias/ExponentialMovingAverage, efficientnet-b0/blocks_6/se/conv2d/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_6/se/conv2d_1/bias/ExponentialMovingAverage, efficientnet-b0/blocks_6/se/conv2d_1/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_6/tpu_batch_normalization/beta/ExponentialMovingAverage, efficientnet-b0/blocks_6/tpu_batch_normalization/gamma/ExponentialMovingAverage, efficientnet-b0/blocks_6/tpu_batch_normalization/moving_mean/ExponentialMovingAverage, efficientnet-b0/blocks_6/tpu_batch_normalization/moving_variance/ExponentialMovingAverage, efficientnet-b0/blocks_6/tpu_batch_normalization_1/beta/ExponentialMovingAverage, efficientnet-b0/blocks_6/tpu_batch_normalization_1/gamma/ExponentialMovingAverage, efficientnet-b0/blocks_6/tpu_batch_normalization_1/moving_mean/ExponentialMovingAverage, efficientnet-b0/blocks_6/tpu_batch_normalization_1/moving_variance/ExponentialMovingAverage, efficientnet-b0/blocks_6/tpu_batch_normalization_2/beta/ExponentialMovingAverage, efficientnet-b0/blocks_6/tpu_batch_normalization_2/gamma/ExponentialMovingAverage, efficientnet-b0/blocks_6/tpu_batch_normalization_2/moving_mean/ExponentialMovingAverage, efficientnet-b0/blocks_6/tpu_batch_normalization_2/moving_variance/ExponentialMovingAverage, efficientnet-b0/blocks_7/conv2d/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_7/conv2d_1/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_7/depthwise_conv2d/depthwise_kernel/ExponentialMovingAverage, efficientnet-b0/blocks_7/se/conv2d/bias/ExponentialMovingAverage, efficientnet-b0/blocks_7/se/conv2d/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_7/se/conv2d_1/bias/ExponentialMovingAverage, efficientnet-b0/blocks_7/se/conv2d_1/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_7/tpu_batch_normalization/beta/ExponentialMovingAverage, efficientnet-b0/blocks_7/tpu_batch_normalization/gamma/ExponentialMovingAverage, efficientnet-b0/blocks_7/tpu_batch_normalization/moving_mean/ExponentialMovingAverage, efficientnet-b0/blocks_7/tpu_batch_normalization/moving_variance/ExponentialMovingAverage, efficientnet-b0/blocks_7/tpu_batch_normalization_1/beta/ExponentialMovingAverage, efficientnet-b0/blocks_7/tpu_batch_normalization_1/gamma/ExponentialMovingAverage, efficientnet-b0/blocks_7/tpu_batch_normalization_1/moving_mean/ExponentialMovingAverage, efficientnet-b0/blocks_7/tpu_batch_normalization_1/moving_variance/ExponentialMovingAverage, efficientnet-b0/blocks_7/tpu_batch_normalization_2/beta/ExponentialMovingAverage, efficientnet-b0/blocks_7/tpu_batch_normalization_2/gamma/ExponentialMovingAverage, efficientnet-b0/blocks_7/tpu_batch_normalization_2/moving_mean/ExponentialMovingAverage, efficientnet-b0/blocks_7/tpu_batch_normalization_2/moving_variance/ExponentialMovingAverage, efficientnet-b0/blocks_8/conv2d/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_8/conv2d_1/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_8/depthwise_conv2d/depthwise_kernel/ExponentialMovingAverage, efficientnet-b0/blocks_8/se/conv2d/bias/ExponentialMovingAverage, efficientnet-b0/blocks_8/se/conv2d/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_8/se/conv2d_1/bias/ExponentialMovingAverage, efficientnet-b0/blocks_8/se/conv2d_1/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_8/tpu_batch_normalization/beta/ExponentialMovingAverage, efficientnet-b0/blocks_8/tpu_batch_normalization/gamma/ExponentialMovingAverage, efficientnet-b0/blocks_8/tpu_batch_normalization/moving_mean/ExponentialMovingAverage, efficientnet-b0/blocks_8/tpu_batch_normalization/moving_variance/ExponentialMovingAverage, efficientnet-b0/blocks_8/tpu_batch_normalization_1/beta/ExponentialMovingAverage, efficientnet-b0/blocks_8/tpu_batch_normalization_1/gamma/ExponentialMovingAverage, efficientnet-b0/blocks_8/tpu_batch_normalization_1/moving_mean/ExponentialMovingAverage, efficientnet-b0/blocks_8/tpu_batch_normalization_1/moving_variance/ExponentialMovingAverage, efficientnet-b0/blocks_8/tpu_batch_normalization_2/beta/ExponentialMovingAverage, efficientnet-b0/blocks_8/tpu_batch_normalization_2/gamma/ExponentialMovingAverage, efficientnet-b0/blocks_8/tpu_batch_normalization_2/moving_mean/ExponentialMovingAverage, efficientnet-b0/blocks_8/tpu_batch_normalization_2/moving_variance/ExponentialMovingAverage, efficientnet-b0/blocks_9/conv2d/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_9/conv2d_1/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_9/depthwise_conv2d/depthwise_kernel/ExponentialMovingAverage, efficientnet-b0/blocks_9/se/conv2d/bias/ExponentialMovingAverage, efficientnet-b0/blocks_9/se/conv2d/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_9/se/conv2d_1/bias/ExponentialMovingAverage, efficientnet-b0/blocks_9/se/conv2d_1/kernel/ExponentialMovingAverage, efficientnet-b0/blocks_9/tpu_batch_normalization/beta/ExponentialMovingAverage, efficientnet-b0/blocks_9/tpu_batch_normalization/gamma/ExponentialMovingAverage, efficientnet-b0/blocks_9/tpu_batch_normalization/moving_mean/ExponentialMovingAverage, efficientnet-b0/blocks_9/tpu_batch_normalization/moving_variance/ExponentialMovingAverage, efficientnet-b0/blocks_9/tpu_batch_normalization_1/beta/ExponentialMovingAverage, efficientnet-b0/blocks_9/tpu_batch_normalization_1/gamma/ExponentialMovingAverage, efficientnet-b0/blocks_9/tpu_batch_normalization_1/moving_mean/ExponentialMovingAverage, efficientnet-b0/blocks_9/tpu_batch_normalization_1/moving_variance/ExponentialMovingAverage, efficientnet-b0/blocks_9/tpu_batch_normalization_2/beta/ExponentialMovingAverage, efficientnet-b0/blocks_9/tpu_batch_normalization_2/gamma/ExponentialMovingAverage, efficientnet-b0/blocks_9/tpu_batch_normalization_2/moving_mean/ExponentialMovingAverage, efficientnet-b0/blocks_9/tpu_batch_normalization_2/moving_variance/ExponentialMovingAverage, efficientnet-b0/head/conv2d/kernel/ExponentialMovingAverage, efficientnet-b0/head/dense/bias/ExponentialMovingAverage, efficientnet-b0/head/dense/kernel/ExponentialMovingAverage, efficientnet-b0/head/tpu_batch_normalization/beta/ExponentialMovingAverage, efficientnet-b0/head/tpu_batch_normalization/gamma/ExponentialMovingAverage, efficientnet-b0/head/tpu_batch_normalization/moving_mean/ExponentialMovingAverage, efficientnet-b0/head/tpu_batch_normalization/moving_variance/ExponentialMovingAverage, efficientnet-b0/stem/conv2d/kernel/ExponentialMovingAverage, efficientnet-b0/stem/tpu_batch_normalization/beta/ExponentialMovingAverage, efficientnet-b0/stem/tpu_batch_normalization/gamma/ExponentialMovingAverage, efficientnet-b0/stem/tpu_batch_normalization/moving_mean/ExponentialMovingAverage, efficientnet-b0/stem/tpu_batch_normalization/moving_variance/ExponentialMovingAverage
Variables names when this checkpoint was written which don't exist now: Adam/batch_normalization/beta/m, Adam/batch_normalization/beta/v, Adam/batch_normalization/gamma/m, Adam/batch_normalization/gamma/v, Adam/beta_1, Adam/beta_2, Adam/decay, Adam/iter, Adam/learning_rate, Adam/pred/bias/m, Adam/pred/bias/v, Adam/pred/kernel/m, Adam/pred/kernel/v, batch_normalization/beta, batch_normalization/gamma, batch_normalization/moving_mean, batch_normalization/moving_variance, block1a_bn/beta, block1a_bn/gamma, block1a_bn/moving_mean, block1a_bn/moving_variance, block1a_dwconv/depthwise_kernel, block1a_project_bn/beta, block1a_project_bn/gamma, block1a_project_bn/moving_mean, block1a_project_bn/moving_variance, block1a_project_conv/kernel, block1a_se_expand/bias, block1a_se_expand/kernel, block1a_se_reduce/bias, block1a_se_reduce/kernel, block2a_bn/beta, block2a_bn/gamma, block2a_bn/moving_mean, block2a_bn/moving_variance, block2a_dwconv/depthwise_kernel, block2a_expand_bn/beta, block2a_expand_bn/gamma, block2a_expand_bn/moving_mean, block2a_expand_bn/moving_variance, block2a_expand_conv/kernel, block2a_project_bn/beta, block2a_project_bn/gamma, block2a_project_bn/moving_mean, block2a_project_bn/moving_variance, block2a_project_conv/kernel, block2a_se_expand/bias, block2a_se_expand/kernel, block2a_se_reduce/bias, block2a_se_reduce/kernel, block2b_bn/beta, block2b_bn/gamma, block2b_bn/moving_mean, block2b_bn/moving_variance, block2b_dwconv/depthwise_kernel, block2b_expand_bn/beta, block2b_expand_bn/gamma, block2b_expand_bn/moving_mean, block2b_expand_bn/moving_variance, block2b_expand_conv/kernel, block2b_project_bn/beta, block2b_project_bn/gamma, block2b_project_bn/moving_mean, block2b_project_bn/moving_variance, block2b_project_conv/kernel, block2b_se_expand/bias, block2b_se_expand/kernel, block2b_se_reduce/bias, block2b_se_reduce/kernel, block3a_bn/beta, block3a_bn/gamma, block3a_bn/moving_mean, block3a_bn/moving_variance, block3a_dwconv/depthwise_kernel, block3a_expand_bn/beta, block3a_expand_bn/gamma, block3a_expand_bn/moving_mean, block3a_expand_bn/moving_variance, block3a_expand_conv/kernel, block3a_project_bn/beta, block3a_project_bn/gamma, block3a_project_bn/moving_mean, block3a_project_bn/moving_variance, block3a_project_conv/kernel, block3a_se_expand/bias, block3a_se_expand/kernel, block3a_se_reduce/bias, block3a_se_reduce/kernel, block3b_bn/beta, block3b_bn/gamma, block3b_bn/moving_mean, block3b_bn/moving_variance, block3b_dwconv/depthwise_kernel, block3b_expand_bn/beta, block3b_expand_bn/gamma, block3b_expand_bn/moving_mean, block3b_expand_bn/moving_variance, block3b_expand_conv/kernel, block3b_project_bn/beta, block3b_project_bn/gamma, block3b_project_bn/moving_mean, block3b_project_bn/moving_variance, block3b_project_conv/kernel, block3b_se_expand/bias, block3b_se_expand/kernel, block3b_se_reduce/bias, block3b_se_reduce/kernel, block4a_bn/beta, block4a_bn/gamma, block4a_bn/moving_mean, block4a_bn/moving_variance, block4a_dwconv/depthwise_kernel, block4a_expand_bn/beta, block4a_expand_bn/gamma, block4a_expand_bn/moving_mean, block4a_expand_bn/moving_variance, block4a_expand_conv/kernel, block4a_project_bn/beta, block4a_project_bn/gamma, block4a_project_bn/moving_mean, block4a_project_bn/moving_variance, block4a_project_conv/kernel, block4a_se_expand/bias, block4a_se_expand/kernel, block4a_se_reduce/bias, block4a_se_reduce/kernel, block4b_bn/beta, block4b_bn/gamma, block4b_bn/moving_mean, block4b_bn/moving_variance, block4b_dwconv/depthwise_kernel, block4b_expand_bn/beta, block4b_expand_bn/gamma, block4b_expand_bn/moving_mean, block4b_expand_bn/moving_variance, block4b_expand_conv/kernel, block4b_project_bn/beta, block4b_project_bn/gamma, block4b_project_bn/moving_mean, block4b_project_bn/moving_variance, block4b_project_conv/kernel, block4b_se_expand/bias, block4b_se_expand/kernel, block4b_se_reduce/bias, block4b_se_reduce/kernel, block4c_bn/beta, block4c_bn/gamma, block4c_bn/moving_mean, block4c_bn/moving_variance, block4c_dwconv/depthwise_kernel, block4c_expand_bn/beta, block4c_expand_bn/gamma, block4c_expand_bn/moving_mean, block4c_expand_bn/moving_variance, block4c_expand_conv/kernel, block4c_project_bn/beta, block4c_project_bn/gamma, block4c_project_bn/moving_mean, block4c_project_bn/moving_variance, block4c_project_conv/kernel, block4c_se_expand/bias, block4c_se_expand/kernel, block4c_se_reduce/bias, block4c_se_reduce/kernel, block5a_bn/beta, block5a_bn/gamma, block5a_bn/moving_mean, block5a_bn/moving_variance, block5a_dwconv/depthwise_kernel, block5a_expand_bn/beta, block5a_expand_bn/gamma, block5a_expand_bn/moving_mean, block5a_expand_bn/moving_variance, block5a_expand_conv/kernel, block5a_project_bn/beta, block5a_project_bn/gamma, block5a_project_bn/moving_mean, block5a_project_bn/moving_variance, block5a_project_conv/kernel, block5a_se_expand/bias, block5a_se_expand/kernel, block5a_se_reduce/bias, block5a_se_reduce/kernel, block5b_bn/beta, block5b_bn/gamma, block5b_bn/moving_mean, block5b_bn/moving_variance, block5b_dwconv/depthwise_kernel, block5b_expand_bn/beta, block5b_expand_bn/gamma, block5b_expand_bn/moving_mean, block5b_expand_bn/moving_variance, block5b_expand_conv/kernel, block5b_project_bn/beta, block5b_project_bn/gamma, block5b_project_bn/moving_mean, block5b_project_bn/moving_variance, block5b_project_conv/kernel, block5b_se_expand/bias, block5b_se_expand/kernel, block5b_se_reduce/bias, block5b_se_reduce/kernel, block5c_bn/beta, block5c_bn/gamma, block5c_bn/moving_mean, block5c_bn/moving_variance, block5c_dwconv/depthwise_kernel, block5c_expand_bn/beta, block5c_expand_bn/gamma, block5c_expand_bn/moving_mean, block5c_expand_bn/moving_variance, block5c_expand_conv/kernel, block5c_project_bn/beta, block5c_project_bn/gamma, block5c_project_bn/moving_mean, block5c_project_bn/moving_variance, block5c_project_conv/kernel, block5c_se_expand/bias, block5c_se_expand/kernel, block5c_se_reduce/bias, block5c_se_reduce/kernel, block6a_bn/beta, block6a_bn/gamma, block6a_bn/moving_mean, block6a_bn/moving_variance, block6a_dwconv/depthwise_kernel, block6a_expand_bn/beta, block6a_expand_bn/gamma, block6a_expand_bn/moving_mean, block6a_expand_bn/moving_variance, block6a_expand_conv/kernel, block6a_project_bn/beta, block6a_project_bn/gamma, block6a_project_bn/moving_mean, block6a_project_bn/moving_variance, block6a_project_conv/kernel, block6a_se_expand/bias, block6a_se_expand/kernel, block6a_se_reduce/bias, block6a_se_reduce/kernel, block6b_bn/beta, block6b_bn/gamma, block6b_bn/moving_mean, block6b_bn/moving_variance, block6b_dwconv/depthwise_kernel, block6b_expand_bn/beta, block6b_expand_bn/gamma, block6b_expand_bn/moving_mean, block6b_expand_bn/moving_variance, block6b_expand_conv/kernel, block6b_project_bn/beta, block6b_project_bn/gamma, block6b_project_bn/moving_mean, block6b_project_bn/moving_variance, block6b_project_conv/kernel, block6b_se_expand/bias, block6b_se_expand/kernel, block6b_se_reduce/bias, block6b_se_reduce/kernel, block6c_bn/beta, block6c_bn/gamma, block6c_bn/moving_mean, block6c_bn/moving_variance, block6c_dwconv/depthwise_kernel, block6c_expand_bn/beta, block6c_expand_bn/gamma, block6c_expand_bn/moving_mean, block6c_expand_bn/moving_variance, block6c_expand_conv/kernel, block6c_project_bn/beta, block6c_project_bn/gamma, block6c_project_bn/moving_mean, block6c_project_bn/moving_variance, block6c_project_conv/kernel, block6c_se_expand/bias, block6c_se_expand/kernel, block6c_se_reduce/bias, block6c_se_reduce/kernel, block6d_bn/beta, block6d_bn/gamma, block6d_bn/moving_mean, block6d_bn/moving_variance, block6d_dwconv/depthwise_kernel, block6d_expand_bn/beta, block6d_expand_bn/gamma, block6d_expand_bn/moving_mean, block6d_expand_bn/moving_variance, block6d_expand_conv/kernel, block6d_project_bn/beta, block6d_project_bn/gamma, block6d_project_bn/moving_mean, block6d_project_bn/moving_variance, block6d_project_conv/kernel, block6d_se_expand/bias, block6d_se_expand/kernel, block6d_se_reduce/bias, block6d_se_reduce/kernel, block7a_bn/beta, block7a_bn/gamma, block7a_bn/moving_mean, block7a_bn/moving_variance, block7a_dwconv/depthwise_kernel, block7a_expand_bn/beta, block7a_expand_bn/gamma, block7a_expand_bn/moving_mean, block7a_expand_bn/moving_variance, block7a_expand_conv/kernel, block7a_project_bn/beta, block7a_project_bn/gamma, block7a_project_bn/moving_mean, block7a_project_bn/moving_variance, block7a_project_conv/kernel, block7a_se_expand/bias, block7a_se_expand/kernel, block7a_se_reduce/bias, block7a_se_reduce/kernel, count, normalization/count, normalization/mean, normalization/variance, pred/bias, pred/kernel, stem_bn/beta, stem_bn/gamma, stem_bn/moving_mean, stem_bn/moving_variance, stem_conv/kernel, top_bn/beta, top_bn/gamma, top_bn/moving_mean, top_bn/moving_variance, top_conv/kernel, total
(0 variable name(s) did match)
Could someone help me on this issue?
