A TensorFlow implementation of the paper One-Shot Visual Imitation Learning via Meta-Learning (Finn*, Yu* et al., 2017). Here are the instructions to run our experiments shown in the paper.
First clone the fork of the gym repo found here, and following the instructions there to install gym. Switch to branch mil.
Then go to the mil directory and run ./scripts/get_data.sh to download the data.
After downloading the data, training and testing scripts for MIL are available in scripts/.
Note: The code only includes the simulated experiments.
错误文档3.22（./scripts/run_sim_push.sh） Traceback (most recent call last): File "/home/lip/.pyenv/versions/3.5.4/lib/python3.5/site-packages/tensorflow/python/framework/op_def_library.py", line 510, in _apply_op_helper preferred_dtype=default_dtype) File "/home/lip/.pyenv/versions/3.5.4/lib/python3.5/site-packages/tensorflow/python/framework/ops.py", line 1036, in internal_convert_to_tensor ret = conversion_func(value, dtype=dtype, name=name, as_ref=as_ref) File "/home/lip/.pyenv/versions/3.5.4/lib/python3.5/site-packages/tensorflow/python/framework/constant_op.py", line 235, in _constant_tensor_conversion_function return constant(v, dtype=dtype, name=name) File "/home/lip/.pyenv/versions/3.5.4/lib/python3.5/site-packages/tensorflow/python/framework/constant_op.py", line 214, in constant value, dtype=dtype, shape=shape, verify_shape=verify_shape)) File "/home/lip/.pyenv/versions/3.5.4/lib/python3.5/site-packages/tensorflow/python/framework/tensor_util.py", line 442, in make_tensor_proto _GetDenseDimensions(values))) ValueError: Argument must be a dense tensor: range(0, 20) - got shape [20], but wanted [].
During handling of the above exception, another exception occurred:
Traceback (most recent call last): File "/home/lip/.pyenv/versions/3.5.4/lib/python3.5/site-packages/tensorflow/python/framework/op_def_library.py", line 524, in _apply_op_helper values, as_ref=input_arg.is_ref).dtype.name File "/home/lip/.pyenv/versions/3.5.4/lib/python3.5/site-packages/tensorflow/python/framework/ops.py", line 1036, in internal_convert_to_tensor ret = conversion_func(value, dtype=dtype, name=name, as_ref=as_ref) File "/home/lip/.pyenv/versions/3.5.4/lib/python3.5/site-packages/tensorflow/python/framework/constant_op.py", line 235, in _constant_tensor_conversion_function return constant(v, dtype=dtype, name=name) File "/home/lip/.pyenv/versions/3.5.4/lib/python3.5/site-packages/tensorflow/python/framework/constant_op.py", line 214, in constant value, dtype=dtype, shape=shape, verify_shape=verify_shape)) File "/home/lip/.pyenv/versions/3.5.4/lib/python3.5/site-packages/tensorflow/python/framework/tensor_util.py", line 442, in make_tensor_proto _GetDenseDimensions(values))) ValueError: Argument must be a dense tensor: range(0, 20) - got shape [20], but wanted [].
During handling of the above exception, another exception occurred:
Traceback (most recent call last): File "main.py", line 295, in main() File "main.py", line 261, in main model.init_network(graph, input_tensors=train_input_tensors, restore_iter=FLAGS.restore_iter) File "/home/lip/mil/mil.py", line 38, in init_network network_config=self.network_params) File "/home/lip/mil/mil.py", line 574, in construct_model unused = batch_metalearn((inputa[0], inputb[0], actiona[0], actionb[0])) File "/home/lip/mil/mil.py", line 460, in batch_metalearn local_outputa, final_eept_preda = self.forward(inputa, state_inputa, weights, network_config=network_config) File "/home/lip/mil/mil.py", line 225, in forward context = tf.transpose(tf.gather(tf.transpose(tf.zeros_like(flatten_image)), range(FLAGS.bt_dim))) File "/home/lip/.pyenv/versions/3.5.4/lib/python3.5/site-packages/tensorflow/python/ops/array_ops.py", line 2667, in gather params, indices, validate_indices=validate_indices, name=name) File "/home/lip/.pyenv/versions/3.5.4/lib/python3.5/site-packages/tensorflow/python/ops/gen_array_ops.py", line 1777, in gather validate_indices=validate_indices, name=name) File "/home/lip/.pyenv/versions/3.5.4/lib/python3.5/site-packages/tensorflow/python/framework/op_def_library.py", line 528, in _apply_op_helper (input_name, err)) ValueError: Tried to convert 'indices' to a tensor and failed. Error: Argument must be a dense tensor: range(0, 20) - got shape [20], but wanted [].