Initial commit

README.md

@@ -0,0 +1,29 @@
+---
+language: en
+license: apache-2.0
+datasets:
+- conll2003
+---
+
+[DistilBERT base uncased](https://huggingface.co/distilbert-base-uncased), fine-tuned for NER using the [conll03 english dataset](https://huggingface.co/datasets/conll2003).
+
+## Versions
+
+- Transformers version: 4.3.1
+- Datasets version: 1.3.0
+
+## Training
+
+```
+$ run_ner.py \
+  --model_name_or_path distilbert-base-uncased \
+  --label_all_tokens True \
+  --return_entity_level_metrics True \
+  --dataset_name conll2003 \
+  --output_dir /tmp/distilbert-base-uncased-finetuned-conll03-english \
+  --do_train \
+  --do_eval
+```
+
+After training, we update the labels to match the NER specific labels from the
+dataset [conll2003](https://raw.githubusercontent.com/huggingface/datasets/1.3.0/datasets/conll2003/dataset_infos.json)

config.json

@@ -0,0 +1,46 @@
+{
+  "_name_or_path": "distilbert-base-uncased",
+  "activation": "gelu",
+  "architectures": [
+    "DistilBertForTokenClassification"
+  ],
+  "attention_dropout": 0.1,
+  "dim": 768,
+  "dropout": 0.1,
+  "finetuning_task": "ner",
+  "hidden_dim": 3072,
+  "id2label": {
+    "0": "O",
+    "1": "B-PER",
+    "2": "I-PER",
+    "3": "B-ORG",
+    "4": "I-ORG",
+    "5": "B-LOC",
+    "6": "I-LOC",
+    "7": "B-MISC",
+    "8": "I-MISC"
+  },
+  "initializer_range": 0.02,
+  "label2id": {
+    "B-LOC": 5,
+    "B-MISC": 7,
+    "B-ORG": 3,
+    "B-PER": 1,
+    "I-LOC": 6,
+    "I-MISC": 8,
+    "I-ORG": 4,
+    "I-PER": 2,
+    "O": 0
+  },
+  "max_position_embeddings": 512,
+  "model_type": "distilbert",
+  "n_heads": 12,
+  "n_layers": 6,
+  "pad_token_id": 0,
+  "qa_dropout": 0.1,
+  "seq_classif_dropout": 0.2,
+  "sinusoidal_pos_embds": false,
+  "tie_weights_": true,
+  "transformers_version": "4.3.1",
+  "vocab_size": 30522
+}

eval_results_ner.txt

@@ -0,0 +1,24 @@
+eval_loss = 0.06388235837221146
+eval_LOC_precision = 0.9601215343714394
+eval_LOC_recall = 0.9656226126814362
+eval_LOC_f1 = 0.962864216339745
+eval_LOC_number = 2618
+eval_MISC_precision = 0.848582995951417
+eval_MISC_recall = 0.851340373679935
+eval_MISC_f1 = 0.8499594484995945
+eval_MISC_number = 1231
+eval_ORG_precision = 0.8968140751307656
+eval_ORG_recall = 0.9173151750972762
+eval_ORG_f1 = 0.9069487857658092
+eval_ORG_number = 2056
+eval_PER_precision = 0.9727242852448242
+eval_PER_recall = 0.975609756097561
+eval_PER_f1 = 0.9741648839888103
+eval_PER_number = 3034
+eval_overall_precision = 0.9343243842911028
+eval_overall_recall = 0.9421635529701309
+eval_overall_f1 = 0.9382275942739376
+eval_overall_accuracy = 0.9854321889843837
+eval_runtime = 128.0613
+eval_samples_per_second = 25.378
+epoch = 3.0

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:806a4e6025a37a7a9646904397489f213e68c8822c421b34b4d687f8f3fd5a8f
+size 265520427

special_tokens_map.json

@@ -0,0 +1 @@
+{"unk_token": "[UNK]", "sep_token": "[SEP]", "pad_token": "[PAD]", "cls_token": "[CLS]", "mask_token": "[MASK]"}

tokenizer_config.json

@@ -0,0 +1 @@
+{"do_lower_case": true, "unk_token": "[UNK]", "sep_token": "[SEP]", "pad_token": "[PAD]", "cls_token": "[CLS]", "mask_token": "[MASK]", "tokenize_chinese_chars": true, "strip_accents": null, "model_max_length": 512, "name_or_path": "distilbert-base-uncased"}

train_results.txt

@@ -0,0 +1,3 @@
+epoch = 3.0
+train_runtime = 7133.5904
+train_samples_per_second = 0.738

trainer_state.json

@@ -0,0 +1,83 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "global_step": 5268,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.525436598329537e-05,
+      "loss": 0.2141,
+      "step": 500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.050873196659074e-05,
+      "loss": 0.094,
+      "step": 1000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5763097949886106e-05,
+      "loss": 0.0717,
+      "step": 1500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1017463933181475e-05,
+      "loss": 0.0561,
+      "step": 2000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6271829916476843e-05,
+      "loss": 0.0359,
+      "step": 2500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.152619589977221e-05,
+      "loss": 0.0357,
+      "step": 3000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.678056188306758e-05,
+      "loss": 0.0347,
+      "step": 3500
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2034927866362947e-05,
+      "loss": 0.0151,
+      "step": 4000
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.289293849658315e-06,
+      "loss": 0.0156,
+      "step": 4500
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5436598329536827e-06,
+      "loss": 0.014,
+      "step": 5000
+    },
+    {
+      "epoch": 3.0,
+      "step": 5268,
+      "total_flos": 696221070844428,
+      "train_runtime": 7133.5904,
+      "train_samples_per_second": 0.738
+    }
+  ],
+  "max_steps": 5268,
+  "num_train_epochs": 3,
+  "total_flos": 696221070844428,
+  "trial_name": null,
+  "trial_params": null
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d8829495baa1a2259f501a4f578cbadd01af4203b57ffe21e6c39fe76825838e
+size 2159