Skip to content

Commit

Permalink
refactor: all classifier models to pytorch (#1565)
Browse files Browse the repository at this point in the history
  • Loading branch information
IgnatovFedor committed May 20, 2022
1 parent cfdde73 commit d6774bb
Show file tree
Hide file tree
Showing 29 changed files with 116 additions and 1,604 deletions.
28 changes: 10 additions & 18 deletions deeppavlov/configs/classifiers/boolqa_rubert.json
Original file line number Diff line number Diff line change
Expand Up @@ -13,24 +13,21 @@
"in_y": ["y"],
"pipe": [
{
"class_name": "bert_preprocessor",
"vocab_file": "{DOWNLOADS_PATH}/bert_models/rubert_cased_L-12_H-768_A-12_v1/vocab.txt",
"class_name": "torch_transformers_preprocessor",
"vocab_file": "{TRANSFORMER}",
"do_lower_case": false,
"max_seq_length": 128,
"in": ["text_a", "text_b"],
"out": ["bert_features"]
},
{
"class_name": "bert_classifier",
"class_name": "torch_transformers_classifier",
"n_classes": 2,
"one_hot_labels": false,
"bert_config_file": "{DOWNLOADS_PATH}/bert_models/rubert_cased_L-12_H-768_A-12_v1/bert_config.json",
"pretrained_bert": "{DOWNLOADS_PATH}/bert_models/rubert_cased_L-12_H-768_A-12_v1/bert_model.ckpt",
"pretrained_bert": "{TRANSFORMER}",
"save_path": "{MODELS_PATH}/boolqa_rubert/model_rubert",
"load_path": "{MODELS_PATH}/boolqa_rubert/model_rubert",
"keep_prob": 0.5,
"optimizer": "tf.train:AdamOptimizer",
"learning_rate": 2e-05,
"optimizer": "AdamW",
"optimizer_parameters": {"lr": 2e-05},
"learning_rate_drop_patience": 3,
"learning_rate_drop_div": 2.0,
"in": ["bert_features"],
Expand All @@ -50,19 +47,14 @@
"log_every_n_epochs": 1,
"evaluation_targets": ["valid", "train"],
"show_examples": false,
"tensorboard_log_dir": "{MODELS_PATH}/boolqa_rubert/logs"
"class_name": "torch_trainer"
},
"metadata": {
"variables": {
"ROOT_PATH": "~/.deeppavlov",
"DOWNLOADS_PATH": "{ROOT_PATH}/downloads",
"MODELS_PATH": "{ROOT_PATH}/models"
},
"download": [
{
"url": "http://files.deeppavlov.ai/deeppavlov_data/bert/rubert_cased_L-12_H-768_A-12_v1.tar.gz",
"subdir": "{DOWNLOADS_PATH}/bert_models"
}
]
"MODELS_PATH": "{ROOT_PATH}/models",
"TRANSFORMER": "DeepPavlov/rubert-base-cased"
}
}
}
104 changes: 0 additions & 104 deletions deeppavlov/configs/classifiers/paraphraser_bert.json

This file was deleted.

33 changes: 14 additions & 19 deletions deeppavlov/configs/classifiers/paraphraser_rubert.json
Original file line number Diff line number Diff line change
Expand Up @@ -14,24 +14,21 @@
"in_y": ["y"],
"pipe": [
{
"class_name": "bert_preprocessor",
"vocab_file": "{DOWNLOADS_PATH}/bert_models/rubert_cased_L-12_H-768_A-12_v1/vocab.txt",
"class_name": "torch_transformers_preprocessor",
"vocab_file": "{TRANSFORMER}",
"do_lower_case": false,
"max_seq_length": 64,
"in": ["text_a", "text_b"],
"out": ["bert_features"]
},
{
"class_name": "bert_classifier",
"class_name": "torch_transformers_classifier",
"n_classes": 2,
"one_hot_labels": false,
"bert_config_file": "{DOWNLOADS_PATH}/bert_models/rubert_cased_L-12_H-768_A-12_v1/bert_config.json",
"pretrained_bert": "{DOWNLOADS_PATH}/bert_models/rubert_cased_L-12_H-768_A-12_v1/bert_model.ckpt",
"save_path": "{MODELS_PATH}/paraphraser_rubert/model_rubert",
"load_path": "{MODELS_PATH}/paraphraser_rubert/model_rubert",
"keep_prob": 0.5,
"optimizer": "tf.train:AdamOptimizer",
"learning_rate": 2e-05,
"pretrained_bert": "{TRANSFORMER}",
"save_path": "{MODEL_PATH}/model",
"load_path": "{MODEL_PATH}/model",
"optimizer": "AdamW",
"optimizer_parameters": {"lr": 2e-05},
"learning_rate_drop_patience": 3,
"learning_rate_drop_div": 2.0,
"in": ["bert_features"],
Expand All @@ -51,13 +48,15 @@
"log_every_n_batches": 50,
"validate_best": true,
"test_best": true,
"tensorboard_log_dir": "{MODELS_PATH}/paraphraser_rubert/logs"
"class_name": "torch_trainer"
},
"metadata": {
"variables": {
"ROOT_PATH": "~/.deeppavlov",
"DOWNLOADS_PATH": "{ROOT_PATH}/downloads",
"MODELS_PATH": "{ROOT_PATH}/models"
"MODELS_PATH": "{ROOT_PATH}/models",
"MODEL_PATH": "{MODELS_PATH}/classifiers/paraphraser_rubert_torch",
"TRANSFORMER": "DeepPavlov/rubert-base-cased"
},
"download": [
{
Expand All @@ -69,12 +68,8 @@
"subdir": "{DOWNLOADS_PATH}/paraphraser_data"
},
{
"url": "http://files.deeppavlov.ai/deeppavlov_data/bert/rubert_cased_L-12_H-768_A-12_v1.tar.gz",
"subdir": "{DOWNLOADS_PATH}/bert_models"
},
{
"url": "http://files.deeppavlov.ai/deeppavlov_data/classifiers/paraphraser_rubert_v0.tar.gz",
"subdir": "{ROOT_PATH}/models"
"url": "http://files.deeppavlov.ai/v1/classifiers/paraphraser_rubert/paraphraser_rubert_v1.tar.gz",
"subdir": "{MODEL_PATH}"
}
]
}
Expand Down
25 changes: 10 additions & 15 deletions deeppavlov/configs/classifiers/rusentiment_bert.json
Original file line number Diff line number Diff line change
Expand Up @@ -30,8 +30,8 @@
],
"pipe": [
{
"class_name": "bert_preprocessor",
"vocab_file": "{DOWNLOADS_PATH}/bert_models/multi_cased_L-12_H-768_A-12/vocab.txt",
"class_name": "torch_transformers_preprocessor",
"vocab_file": "{TRANSFORMER}",
"do_lower_case": false,
"max_seq_length": 64,
"in": [
Expand Down Expand Up @@ -60,16 +60,14 @@
"single_vector": true
},
{
"class_name": "bert_classifier",
"class_name": "torch_transformers_classifier",
"n_classes": "#classes_vocab.len",
"return_probas": true,
"one_hot_labels": true,
"bert_config_file": "{DOWNLOADS_PATH}/bert_models/multi_cased_L-12_H-768_A-12/bert_config.json",
"pretrained_bert": "{DOWNLOADS_PATH}/bert_models/multi_cased_L-12_H-768_A-12/bert_model.ckpt",
"pretrained_bert": "{TRANSFORMER}",
"save_path": "{MODEL_PATH}/model",
"load_path": "{MODEL_PATH}/model",
"keep_prob": 0.5,
"learning_rate": 1e-05,
"optimizer_parameters": {"lr": 1e-05},
"learning_rate_drop_patience": 5,
"learning_rate_drop_div": 2.0,
"in": [
Expand Down Expand Up @@ -123,23 +121,20 @@
"valid",
"test"
],
"tensorboard_log_dir": "{MODEL_PATH}/"
"class_name": "torch_trainer"
},
"metadata": {
"variables": {
"ROOT_PATH": "~/.deeppavlov",
"DOWNLOADS_PATH": "{ROOT_PATH}/downloads",
"MODELS_PATH": "{ROOT_PATH}/models",
"MODEL_PATH": "{MODELS_PATH}/classifiers/rusentiment_bert_v0/"
"MODEL_PATH": "{MODELS_PATH}/classifiers/rusentiment_bert_torch",
"TRANSFORMER": "bert-base-multilingual-cased"
},
"download": [
{
"url": "http://files.deeppavlov.ai/deeppavlov_data/bert/multi_cased_L-12_H-768_A-12.zip",
"subdir": "{DOWNLOADS_PATH}/bert_models"
},
{
"url": "http://files.deeppavlov.ai/deeppavlov_data/classifiers/rusentiment_bert_v0.tar.gz",
"subdir": "{MODELS_PATH}/classifiers/"
"url": "http://files.deeppavlov.ai/v1/classifiers/rusentiment_bert/rusentiment_bert_torch.tar.gz",
"subdir": "{MODEL_PATH}"
}
]
}
Expand Down
25 changes: 10 additions & 15 deletions deeppavlov/configs/classifiers/rusentiment_convers_bert.json
Original file line number Diff line number Diff line change
Expand Up @@ -30,8 +30,8 @@
],
"pipe": [
{
"class_name": "bert_preprocessor",
"vocab_file": "{DOWNLOADS_PATH}/bert_models/ru_conversational_cased_L-12_H-768_A-12/vocab.txt",
"class_name": "torch_transformers_preprocessor",
"vocab_file": "{TRANSFORMER}",
"do_lower_case": false,
"max_seq_length": 64,
"in": [
Expand Down Expand Up @@ -60,16 +60,14 @@
"single_vector": true
},
{
"class_name": "bert_classifier",
"class_name": "torch_transformers_classifier",
"n_classes": "#classes_vocab.len",
"return_probas": true,
"one_hot_labels": true,
"bert_config_file": "{DOWNLOADS_PATH}/bert_models/ru_conversational_cased_L-12_H-768_A-12/bert_config.json",
"pretrained_bert": "{DOWNLOADS_PATH}/bert_models/ru_conversational_cased_L-12_H-768_A-12/bert_model.ckpt",
"pretrained_bert": "{TRANSFORMER}",
"save_path": "{MODEL_PATH}/model",
"load_path": "{MODEL_PATH}/model",
"keep_prob": 0.5,
"learning_rate": 1e-05,
"optimizer_parameters": {"lr": 1e-05},
"learning_rate_drop_patience": 5,
"learning_rate_drop_div": 2.0,
"in": [
Expand Down Expand Up @@ -123,23 +121,20 @@
"valid",
"test"
],
"tensorboard_log_dir": "{MODEL_PATH}/"
"class_name": "torch_trainer"
},
"metadata": {
"variables": {
"ROOT_PATH": "~/.deeppavlov",
"DOWNLOADS_PATH": "{ROOT_PATH}/downloads",
"MODELS_PATH": "{ROOT_PATH}/models",
"MODEL_PATH": "{MODELS_PATH}/classifiers/rusentiment_convers_bert_v0/"
"MODEL_PATH": "{MODELS_PATH}/classifiers/rusentiment_convers_bert_torch",
"TRANSFORMER": "DeepPavlov/rubert-base-cased-conversational"
},
"download": [
{
"url": "http://files.deeppavlov.ai/deeppavlov_data/bert/ru_conversational_cased_L-12_H-768_A-12.tar.gz",
"subdir": "{DOWNLOADS_PATH}/bert_models"
},
{
"url": "http://files.deeppavlov.ai/deeppavlov_data/classifiers/rusentiment_convers_bert_v0.tar.gz",
"subdir": "{MODELS_PATH}/classifiers/"
"url": "http://files.deeppavlov.ai/v1/classifiers/rusentiment_convers_bert/rusentiment_convers_bert_torch.tar.gz",
"subdir": "{MODEL_PATH}"
}
]
}
Expand Down

0 comments on commit d6774bb

Please sign in to comment.