Feat/glue superglue update

deeppavlov/_meta.py

@@ -1,4 +1,4 @@
-__version__ = '0.17.1'
+__version__ = '0.17.2'
 __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']

deeppavlov/configs/classifiers/glue/glue_mnli_roberta.json

@@ -121,7 +121,6 @@
     "log_every_n_batches": 250,
     "show_examples": false,
     "evaluation_targets": [
-      "train",
       "valid"
     ],
     "class_name": "torch_trainer",

deeppavlov/configs/classifiers/glue/glue_rte_roberta_mnli.json

@@ -121,7 +121,6 @@
     "log_every_n_epochs": 1,
     "show_examples": false,
     "evaluation_targets": [
-      "train",
       "valid"
     ],
     "class_name": "torch_trainer",

deeppavlov/configs/classifiers/glue/glue_wnli_roberta.json

@@ -0,0 +1,147 @@
+{
+  "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_wnli/{BASE_MODEL}"
+    },
+    "download": [
+      {
+        "url": "http://files.deeppavlov.ai/0.16/classifiers/glue_wnli_roberta.tar.gz",
+        "subdir": "{MODELS_PATH}"
+      }
+    ]
+  },
+  "dataset_reader": {
+    "class_name": "huggingface_dataset_reader",
+    "path": "glue",
+    "name": "wnli",
+    "train": "train",
+    "valid": "validation"
+  },
+  "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": 192,
+        "truncation": "longest_first",
+        "padding": "longest",
+        "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-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": 24,
+    "metrics": [
+      "accuracy"
+    ],
+    "epochs": 1,
+    "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
+  }
+}