Release 0.16.0

deeppavlov · Aug 2, 2021 · d60649a · d60649a
2 parents 6c8f892 + 1ca495a
commit d60649a
Show file tree

Hide file tree

Showing 87 changed files with 3,241 additions and 3,447 deletions.
diff --git a/deeppavlov/_meta.py b/deeppavlov/_meta.py
@@ -1,4 +1,4 @@
-__version__ = '0.15.0'
+__version__ = '0.16.0'
 __author__ = 'Neural Networks and Deep Learning lab, MIPT'
 __description__ = 'An open source library for building end-to-end dialog systems and training chatbots.'
 __keywords__ = ['NLP', 'NER', 'SQUAD', 'Intents', 'Chatbot']

diff --git a/deeppavlov/configs/classifiers/glue/glue_mnli_roberta.json b/deeppavlov/configs/classifiers/glue/glue_mnli_roberta.json
@@ -0,0 +1,96 @@
+{
+  "dataset_reader": {
+    "class_name": "huggingface_dataset_reader",
+    "path": "glue",
+    "name": "mnli",
+    "train": "train",
+    "valid": "validation_matched",
+    "test": "test_matched"
+  },
+  "dataset_iterator": {
+    "class_name": "huggingface_dataset_iterator",
+    "features": ["hypothesis", "premise"],
+    "label": "label",
+    "seed": 42
+  },
+  "chainer": {
+    "in": ["hypothesis", "premise"],
+    "in_y": ["y"],
+    "pipe": [
+      {
+        "class_name": "torch_transformers_preprocessor",
+        "vocab_file": "{BASE_MODEL}",
+        "do_lower_case": false,
+        "max_seq_length": 128,
+        "in": ["hypothesis", "premise"],
+        "out": ["bert_features"]
+      },
+      {
+        "id": "classes_vocab",
+        "class_name": "simple_vocab",
+        "fit_on": ["y"],
+        "save_path": "{MODEL_PATH}/classes.dict",
+        "load_path": "{MODEL_PATH}/classes.dict",
+        "in": ["y"],
+        "out": ["y_ids"]
+      },
+      {
+        "in": ["y_ids"],
+        "out": ["y_onehot"],
+        "class_name": "one_hotter",
+        "depth": "#classes_vocab.len",
+        "single_vector": true
+      },
+      {
+        "class_name": "torch_transformers_classifier",
+        "n_classes": "#classes_vocab.len",
+        "return_probas": true,
+        "pretrained_bert": "{BASE_MODEL}",
+        "save_path": "{MODEL_PATH}/model",
+        "load_path": "{MODEL_PATH}/model",
+        "optimizer": "AdamW",
+        "optimizer_parameters": {
+          "lr": 1e-05
+        },
+        "learning_rate_drop_patience": 3,
+        "learning_rate_drop_div": 2.0,
+        "in": ["bert_features"],
+        "in_y": ["y_ids"],
+        "out": ["y_pred_probas"]
+      },
+      {
+        "in": ["y_pred_probas"],
+        "out": ["y_pred_ids"],
+        "class_name": "proba2labels",
+        "max_proba": true
+      },
+      {
+        "in": ["y_pred_ids"],
+        "out": ["y_pred_labels"],
+        "ref": "classes_vocab"
+      }
+    ],
+    "out": ["y_pred_labels"]
+  },
+  "train": {
+    "batch_size": 4,
+    "metrics": ["accuracy"],
+    "validation_patience": 10,
+    "val_every_n_batches": 250,
+    "log_every_n_batches": 250,
+    "show_examples": false,
+    "evaluation_targets": ["train", "valid"],
+    "class_name": "torch_trainer",
+    "tensorboard_log_dir": "{MODEL_PATH}/",
+    "pytest_max_batches": 2
+  },
+  "metadata": {
+    "variables": {
+      "ROOT_PATH": "~/.deeppavlov",
+      "BASE_MODEL": "roberta-large",
+      "DOWNLOADS_PATH": "{ROOT_PATH}/downloads",
+      "MODELS_PATH": "{ROOT_PATH}/models",
+      "MODEL_PATH": "{MODELS_PATH}/classifiers/glue_mnli_{BASE_MODEL}"
+    }
+  }
+}
diff --git a/deeppavlov/configs/classifiers/glue/glue_rte_roberta_mnli.json b/deeppavlov/configs/classifiers/glue/glue_rte_roberta_mnli.json
@@ -0,0 +1,96 @@
+{
+  "dataset_reader": {
+    "class_name": "huggingface_dataset_reader",
+    "path": "glue",
+    "name": "rte",
+    "train": "train",
+    "valid": "validation",
+    "test": "test"
+  },
+  "dataset_iterator": {
+    "class_name": "huggingface_dataset_iterator",
+    "features": ["sentence1", "sentence2"],
+    "label": "label",
+    "seed": 42
+  },
+  "chainer": {
+    "in": ["sentence1", "sentence2"],
+    "in_y": ["y"],
+    "pipe": [
+      {
+        "class_name": "torch_transformers_preprocessor",
+        "vocab_file": "{BASE_MODEL}",
+        "do_lower_case": false,
+        "max_seq_length": 256,
+        "in": ["sentence1", "sentence2"],
+        "out": ["bert_features"]
+      },
+      {
+        "id": "classes_vocab",
+        "class_name": "simple_vocab",
+        "fit_on": ["y"],
+        "save_path": "{MODEL_PATH}/classes.dict",
+        "load_path": "{MODEL_PATH}/classes.dict",
+        "in": ["y"],
+        "out": ["y_ids"]
+      },
+      {
+        "in": ["y_ids"],
+        "out": ["y_onehot"],
+        "class_name": "one_hotter",
+        "depth": "#classes_vocab.len",
+        "single_vector": true
+      },
+      {
+        "class_name": "torch_transformers_classifier",
+        "n_classes": "#classes_vocab.len",
+        "return_probas": true,
+        "pretrained_bert": "{BASE_MODEL}",
+        "save_path": "{MODEL_PATH}/model",
+        "load_path": "{MODEL_PATH}/model",
+        "optimizer": "AdamW",
+        "optimizer_parameters": {
+          "lr": 1e-06
+        },
+        "learning_rate_drop_patience": 3,
+        "learning_rate_drop_div": 2.0,
+        "in": ["bert_features"],
+        "in_y": ["y_ids"],
+        "out": ["y_pred_probas"]
+      },
+      {
+        "in": ["y_pred_probas"],
+        "out": ["y_pred_ids"],
+        "class_name": "proba2labels",
+        "max_proba": true
+      },
+      {
+        "in": ["y_pred_ids"],
+        "out": ["y_pred_labels"],
+        "ref": "classes_vocab"
+      }
+    ],
+    "out": ["y_pred_labels"]
+  },
+  "train": {
+    "batch_size": 4,
+    "metrics": ["accuracy"],
+    "epochs": 2,
+    "val_every_n_epochs": 1,
+    "log_every_n_epochs": 1,
+    "show_examples": false,
+    "evaluation_targets": ["train", "valid"],
+    "class_name": "torch_trainer",
+    "tensorboard_log_dir": "{MODEL_PATH}/",
+    "pytest_max_batches": 2
+  },
+  "metadata": {
+    "variables": {
+      "BASE_MODEL": "roberta-large-mnli",
+      "ROOT_PATH": "~/.deeppavlov",
+      "DOWNLOADS_PATH": "{ROOT_PATH}/downloads",
+      "MODELS_PATH": "{ROOT_PATH}/models",
+      "MODEL_PATH": "{MODELS_PATH}/classifiers/glue_rte_{BASE_MODEL}"
+    }
+  }
+}
diff --git a/deeppavlov/configs/classifiers/intents_dstc2.json b/deeppavlov/configs/classifiers/intents_dstc2.json
@@ -87,7 +87,7 @@
         "in": "y_pred_probas",
         "out": "y_pred_ids",
         "class_name": "proba2labels",
-        "confident_threshold": 0.5
+        "confidence_threshold": 0.5
       },
       {
         "in": "y_pred_ids",

diff --git a/deeppavlov/configs/classifiers/intents_dstc2_bert.json b/deeppavlov/configs/classifiers/intents_dstc2_bert.json
@@ -59,7 +59,7 @@
         "in": "y_pred_probas",
         "out": "y_pred_ids",
         "class_name": "proba2labels",
-        "confident_threshold": 0.5
+        "confidence_threshold": 0.5
       },
       {
         "in": "y_pred_ids",

diff --git a/deeppavlov/configs/classifiers/intents_dstc2_big.json b/deeppavlov/configs/classifiers/intents_dstc2_big.json
@@ -87,7 +87,7 @@
         "in": "y_pred_probas",
         "out": "y_pred_ids",
         "class_name": "proba2labels",
-        "confident_threshold": 0.5
+        "confidence_threshold": 0.5
       },
       {
         "in": "y_pred_ids",

diff --git a/deeppavlov/configs/classifiers/paraphraser_convers_distilrubert_2L.json b/deeppavlov/configs/classifiers/paraphraser_convers_distilrubert_2L.json
@@ -0,0 +1,93 @@
+ {
+  "dataset_reader": {
+    "class_name": "paraphraser_reader",
+    "data_path": "{DOWNLOADS_PATH}/paraphraser_data",
+    "do_lower_case": false
+  },
+  "dataset_iterator": {
+    "class_name": "siamese_iterator",
+    "seed": 243,
+    "len_valid": 500
+  },
+  "chainer": {
+    "in": ["text_a", "text_b"],
+    "in_y": ["y"],
+    "pipe": [
+      {
+        "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": "torch_transformers_classifier",
+        "n_classes": 2,
+        "return_probas": false,
+        "pretrained_bert": "{TRANSFORMER}",
+        "save_path": "{MODEL_PATH}/model",
+        "load_path": "{MODEL_PATH}/model",
+        "attention_probs_keep_prob": 0.11,
+        "hidden_keep_prob": 1.0, 
+        "optimizer": "AdamW",
+        "optimizer_parameters": {
+          "lr": 1.89e-05
+        },
+        "learning_rate_drop_patience": 3,
+        "learning_rate_drop_div": 1.5,
+        "in": [
+          "bert_features"
+        ],
+        "in_y": [
+          "y"
+        ],
+        "out": [
+          "predictions"
+        ]
+      }
+    ],
+    "out": ["predictions"]
+  },
+  "train": {
+    "epochs": 100,
+    "batch_size": 64,
+    "metrics": [
+        "f1",
+        "accuracy"
+    ],
+    "validation_patience": 7,
+    "val_every_n_batches": 50,
+    "log_every_n_batches": 50,
+    "evaluation_targets": [
+      "train",
+      "valid",
+      "test"
+    ],
+    "tensorboard_log_dir": "{MODEL_PATH}/",
+    "class_name": "torch_trainer"
+  },
+  "metadata": {
+    "variables": {
+      "ROOT_PATH": "~/.deeppavlov",
+      "DOWNLOADS_PATH": "{ROOT_PATH}/downloads",
+      "TRANSFORMER": "DeepPavlov/distilrubert-tiny-cased-conversational",
+      "MODELS_PATH": "{ROOT_PATH}/models",
+      "MODEL_PATH": "{MODELS_PATH}/paraphraser_convers_distilrubert_2L"
+    },
+    "download": [
+      {
+        "url": "http://files.deeppavlov.ai/deeppavlov_data/classifiers/paraphraser_convers_distilrubert_2L.tar.gz",
+        "subdir": "{MODELS_PATH}"
+      },
+      {
+        "url": "http://files.deeppavlov.ai/datasets/paraphraser.zip",
+        "subdir": "{DOWNLOADS_PATH}/paraphraser_data"
+      },
+      {
+        "url": "http://files.deeppavlov.ai/datasets/paraphraser_gold.zip",
+        "subdir": "{DOWNLOADS_PATH}/paraphraser_data"
+      }
+    ]
+  }
+}