erip
commented
5 years ago Downgrading to Python 3.6 yields:
(pytext) ➜ pytext git:(master) ✗ pytext export --output-path exported_model.c2 < demo/configs/docnn.json
No config file specified, reading from stdin
Exporting /tmp/model.pt to caffe2 file: exported_model.c2 and onnx file: /tmp/model.onnx
Loading model from /tmp/model.pt...
Task parameters:
{'data_handler': {'columns_to_read': ['doc_label', 'text', 'dict_feat'],
'eval_batch_size': 128,
'eval_path': 'tests/data/test_data_tiny.tsv',
'max_seq_len': -1,
'shuffle': True,
'sort_within_batch': True,
'test_batch_size': 128,
'test_path': 'tests/data/test_data_tiny.tsv',
'train_batch_size': 128,
'train_path': 'tests/data/train_data_tiny.tsv'},
'exporter': None,
'features': {'char_feat': None,
'dense_feat': None,
'dict_feat': None,
'freeze': False,
'load_path': None,
'pretrained_model_embedding': None,
'save_path': None,
'shared_module_key': None,
'word_feat': {'embed_dim': 100,
'embedding_init_range': None,
'embedding_init_strategy': 'random',
'export_input_names': ['tokens_vals'],
'freeze': False,
'lowercase_tokens': True,
'min_freq': 1,
'mlp_layer_dims': [],
'pretrained_embeddings_path': '',
'vocab_file': '',
'vocab_from_all_data': False,
'vocab_from_pretrained_embeddings': False,
'vocab_from_train_data': True,
'vocab_size': 0}},
'featurizer': {'SimpleFeaturizer': {'convert_to_bytes': False,
'lowercase_tokens': True,
'sentence_markers': None,
'split_regex': '\\s+'}},
'labels': {'export_output_names': ['doc_scores'],
'label_weights': {},
'target_prob': False},
'metric_reporter': {'output_path': '/tmp/test_out.txt'},
'model': {'decoder': {'freeze': False,
'hidden_dims': [],
'load_path': None,
'save_path': None,
'shared_module_key': None},
'output_layer': {'freeze': False,
'load_path': None,
'loss': {'CrossEntropyLoss': {}},
'save_path': None,
'shared_module_key': None},
'representation': {'BiLSTMDocAttention': {'dropout': 0.4,
'freeze': False,
'load_path': None,
'lstm': {'bidirectional': True,
'dropout': 0.4,
'freeze': False,
'load_path': None,
'lstm_dim': 32,
'num_layers': 1,
'save_path': None,
'shared_module_key': None},
'mlp_decoder': None,
'pooling': {'SelfAttention': {'attn_dimension': 64,
'dropout': 0.4}},
'save_path': None,
'shared_module_key': None}}},
'optimizer': {'lr': 0.001,
'momentum': 0.0,
'type': 'adam',
'weight_decay': 1e-05},
'scheduler': {'T_max': 1000,
'cooldown': 0,
'cut_frac': 0.1,
'eta_min': 0.0,
'gamma': 0.1,
'lm_gradual_unfreezing': True,
'lm_lr_multiplier': 1.0,
'lm_use_per_layer_lr': False,
'non_pretrained_param_groups': 2,
'patience': 5,
'ratio': 32,
'step_size': 30,
'threshold': 0.0001,
'threshold_is_absolute': False,
'type': 'none'},
'trainer': {'early_stop_after': 0,
'epochs': 10,
'max_clip_norm': None,
'random_seed': 0,
'report_train_metrics': True}}
creating field TextFeatureField
creating field DocLabelField
Loading data...
load_path is not a config of embedding, skipping
save_path is not a config of embedding, skipping
freeze is not a config of embedding, skipping
shared_module_key is not a config of embedding, skipping
dict_feat is not a config of embedding, skipping
char_feat is not a config of embedding, skipping
pretrained_model_embedding is not a config of embedding, skipping
dense_feat is not a config of embedding, skipping
Model has 2 param groups (1 from embedding module) for optimizer
Saving caffe2 model to: exported_model.c2
Saving onnx model to: /tmp/model.onnx
/data/erippeth/anaconda3/envs/pytext/lib/python3.6/site-packages/torch/nn/modules/module.py:487: UserWarning: Implicit dimension choice for softmax has been deprecated. Change the call to include dim=X as an argument.
result = self._slow_forward(*input, **kwargs)
Traceback (most recent call last):
File "/data/erippeth/anaconda3/envs/pytext/bin/pytext", line 11, in <module>
sys.exit(main())
File "/data/erippeth/anaconda3/envs/pytext/lib/python3.6/site-packages/click/core.py", line 764, in __call__
return self.main(*args, **kwargs)
File "/data/erippeth/anaconda3/envs/pytext/lib/python3.6/site-packages/click/core.py", line 717, in main
rv = self.invoke(ctx)
File "/data/erippeth/anaconda3/envs/pytext/lib/python3.6/site-packages/click/core.py", line 1137, in invoke
return _process_result(sub_ctx.command.invoke(sub_ctx))
File "/data/erippeth/anaconda3/envs/pytext/lib/python3.6/site-packages/click/core.py", line 956, in invoke
return ctx.invoke(self.callback, **ctx.params)
File "/data/erippeth/anaconda3/envs/pytext/lib/python3.6/site-packages/click/core.py", line 555, in invoke
return callback(*args, **kwargs)
File "/data/erippeth/anaconda3/envs/pytext/lib/python3.6/site-packages/click/decorators.py", line 17, in new_func
return f(get_current_context(), *args, **kwargs)
File "/data/erippeth/anaconda3/envs/pytext/lib/python3.6/site-packages/pytext/main.py", line 303, in export
export_saved_model_to_caffe2(model, output_path, output_onnx_path)
File "/data/erippeth/anaconda3/envs/pytext/lib/python3.6/site-packages/pytext/workflow.py", line 130, in export_saved_model_to_caffe2
task.export(task.model, export_caffe2_path, export_onnx_path=output_onnx_path)
File "/data/erippeth/anaconda3/envs/pytext/lib/python3.6/site-packages/pytext/task/task.py", line 189, in export
self.exporter.export_to_caffe2(model, export_path, export_onnx_path)
File "/data/erippeth/anaconda3/envs/pytext/lib/python3.6/site-packages/pytext/exporters/exporter.py", line 192, in export_to_caffe2
export_onnx_path,
File "/data/erippeth/anaconda3/envs/pytext/lib/python3.6/site-packages/pytext/utils/onnx_utils.py", line 50, in pytorch_to_caffe2
onnx.checker.check_model(onnx_model)
File "/data/erippeth/anaconda3/envs/pytext/lib/python3.6/site-packages/onnx/checker.py", line 86, in check_model
C.check_model(model.SerializeToString())
onnx.onnx_cpp2py_export.checker.ValidationError: No Op or Function registered for ConstantFill with domain_version of 9
==> Context: Bad node spec: input: "75" output: "76" op_type: "ConstantFill" attribute { name: "input_as_shape" i: 1 type: INT }
Titousensei
houseroad
Steps to reproduce
git clone git@github.com:facebookresearch/pytext.gitconda create -n pytext -yconda activate pytextconda intall pippip install pytext-nlpcd pytextpytext train < demo/configs/docnn.jsonpytext test < demo/configs/docnn.jsonpytext export --output-path exported_model.c2 < demo/configs/docnn.jsonObserved Results
Expected Results