Initial revision.

config.json

@@ -0,0 +1,24 @@
+{
+  "_name_or_path": "bert-large-uncased",
+  "architectures": [
+    "BertForQuestionAnswering"
+  ],
+  "attention_probs_dropout_prob": 0.1,
+  "gradient_checkpointing": false,
+  "hidden_act": "gelu",
+  "hidden_dropout_prob": 0.1,
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 4096,
+  "layer_norm_eps": 1e-12,
+  "max_position_embeddings": 512,
+  "model_type": "bert",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 24,
+  "pad_token_id": 0,
+  "position_embedding_type": "absolute",
+  "transformers_version": "4.5.0.dev0",
+  "type_vocab_size": 2,
+  "use_cache": true,
+  "vocab_size": 30522
+}

eval_results.json

@@ -0,0 +1 @@
+{'exact': 78.6321906847469, 'f1': 81.5816656803201, 'total': 11873, 'HasAns_exact': 73.73481781376518, 'HasAns_f1': 79.64222615088413, 'HasAns_total': 5928, 'NoAns_exact': 83.51555929352396, 'NoAns_f1': 83.51555929352396, 'NoAns_total': 5945, 'best_exact': 78.6321906847469, 'best_exact_thresh': 0.0, 'best_f1': 81.58166568032006, 'best_f1_thresh': 0.0, 'epoch': 1.59}

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2403aa6b558c922393a8a0d0f58920bfdd9036cb631d992e9d3c2191200aa117
+size 1336547639

run.sh

@@ -0,0 +1,15 @@
+python run_qa.py \
+  --model_name_or_path bert-large-uncased \
+  --dataset_name squad_v2 \
+  --do_train \
+  --do_eval \
+  --save_steps 2500 \
+  --eval_steps 2500 \
+  --evaluation_strategy steps \
+  --per_device_train_batch_size 12 \
+  --learning_rate 3e-5 \
+  --num_train_epochs 2 \
+  --max_seq_length 384 \
+  --doc_stride 128 \
+  --output_dir bert-large-uncased-squadv2 \
+  --version_2_with_negative 1

scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5848985f52ec5922c5fc3e341345b65668fd7e6496fab9e090deab4a4bc50567
+size 623

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, "special_tokens_map_file": null, "name_or_path": "bert-large-uncased"}

trainer_state.json

@@ -0,0 +1,345 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.5938069216757742,
+  "global_step": 17500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9316939890710385e-05,
+      "loss": 2.0652,
+      "step": 500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.8633879781420765e-05,
+      "loss": 1.4682,
+      "step": 1000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.795081967213115e-05,
+      "loss": 1.2878,
+      "step": 1500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.7267759562841533e-05,
+      "loss": 1.2088,
+      "step": 2000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.6584699453551913e-05,
+      "loss": 1.1539,
+      "step": 2500
+    },
+    {
+      "HasAns_exact": 59.78407557354926,
+      "HasAns_f1": 65.49896652234426,
+      "HasAns_total": 5928,
+      "NoAns_exact": 80.87468460891506,
+      "NoAns_f1": 80.87468460891506,
+      "NoAns_total": 5945,
+      "best_exact": 70.34447907015918,
+      "best_exact_thresh": 0.0,
+      "best_f1": 73.19783319670329,
+      "best_f1_thresh": 0.0,
+      "epoch": 0.23,
+      "exact": 70.34447907015918,
+      "f1": 73.19783319670331,
+      "step": 2500,
+      "total": 11873
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.5901639344262294e-05,
+      "loss": 1.1802,
+      "step": 3000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5218579234972678e-05,
+      "loss": 1.0921,
+      "step": 3500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.453551912568306e-05,
+      "loss": 1.047,
+      "step": 4000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.3852459016393442e-05,
+      "loss": 1.0404,
+      "step": 4500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3169398907103826e-05,
+      "loss": 1.0074,
+      "step": 5000
+    },
+    {
+      "HasAns_exact": 65.9412955465587,
+      "HasAns_f1": 71.69124855898053,
+      "HasAns_total": 5928,
+      "NoAns_exact": 83.11185870479395,
+      "NoAns_f1": 83.11185870479395,
+      "NoAns_total": 5945,
+      "best_exact": 74.53886970437127,
+      "best_exact_thresh": 0.0,
+      "best_f1": 77.40972976144505,
+      "best_f1_thresh": 0.0,
+      "epoch": 0.46,
+      "exact": 74.53886970437127,
+      "f1": 77.40972976144509,
+      "step": 5000,
+      "total": 11873
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.248633879781421e-05,
+      "loss": 1.017,
+      "step": 5500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.180327868852459e-05,
+      "loss": 0.9725,
+      "step": 6000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1120218579234974e-05,
+      "loss": 0.9779,
+      "step": 6500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.0437158469945358e-05,
+      "loss": 0.9617,
+      "step": 7000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.975409836065574e-05,
+      "loss": 0.929,
+      "step": 7500
+    },
+    {
+      "HasAns_exact": 72.33468286099865,
+      "HasAns_f1": 78.59956513847068,
+      "HasAns_total": 5928,
+      "NoAns_exact": 79.22624053826745,
+      "NoAns_f1": 79.22624053826745,
+      "NoAns_total": 5945,
+      "best_exact": 75.78539543502063,
+      "best_exact_thresh": 0.0,
+      "best_f1": 78.9133514815846,
+      "best_f1_thresh": 0.0,
+      "epoch": 0.68,
+      "exact": 75.78539543502063,
+      "f1": 78.9133514815847,
+      "step": 7500,
+      "total": 11873
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.907103825136612e-05,
+      "loss": 0.9407,
+      "step": 8000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8387978142076503e-05,
+      "loss": 0.921,
+      "step": 8500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7704918032786887e-05,
+      "loss": 0.9025,
+      "step": 9000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7021857923497267e-05,
+      "loss": 0.8724,
+      "step": 9500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.633879781420765e-05,
+      "loss": 0.8794,
+      "step": 10000
+    },
+    {
+      "HasAns_exact": 75.55668016194332,
+      "HasAns_f1": 81.90355530149,
+      "HasAns_total": 5928,
+      "NoAns_exact": 75.02102607232969,
+      "NoAns_f1": 75.02102607232969,
+      "NoAns_total": 5945,
+      "best_exact": 75.2884696369915,
+      "best_exact_thresh": 0.0,
+      "best_f1": 78.45736341507896,
+      "best_f1_thresh": 0.0,
+      "epoch": 0.91,
+      "exact": 75.2884696369915,
+      "f1": 78.45736341507894,
+      "step": 10000,
+      "total": 11873
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5655737704918035e-05,
+      "loss": 0.8788,
+      "step": 10500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4972677595628415e-05,
+      "loss": 0.8614,
+      "step": 11000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4289617486338798e-05,
+      "loss": 0.5678,
+      "step": 11500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3606557377049181e-05,
+      "loss": 0.5483,
+      "step": 12000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2923497267759564e-05,
+      "loss": 0.5724,
+      "step": 12500
+    },
+    {
+      "HasAns_exact": 73.582995951417,
+      "HasAns_f1": 80.21421656448703,
+      "HasAns_total": 5928,
+      "NoAns_exact": 80.31959629941127,
+      "NoAns_f1": 80.31959629941127,
+      "NoAns_total": 5945,
+      "best_exact": 76.95611892529269,
+      "best_exact_thresh": 0.0,
+      "best_f1": 80.26698187436004,
+      "best_f1_thresh": 0.0,
+      "epoch": 1.14,
+      "exact": 76.95611892529269,
+      "f1": 80.26698187436024,
+      "step": 12500,
+      "total": 11873
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2240437158469946e-05,
+      "loss": 0.5641,
+      "step": 13000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.1557377049180328e-05,
+      "loss": 0.5712,
+      "step": 13500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.087431693989071e-05,
+      "loss": 0.5751,
+      "step": 14000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.0191256830601094e-05,
+      "loss": 0.5325,
+      "step": 14500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.508196721311476e-06,
+      "loss": 0.5416,
+      "step": 15000
+    },
+    {
+      "HasAns_exact": 72.94197031039137,
+      "HasAns_f1": 79.45948359952195,
+      "HasAns_total": 5928,
+      "NoAns_exact": 83.06139613120268,
+      "NoAns_f1": 83.06139613120268,
+      "NoAns_total": 5945,
+      "best_exact": 78.00892781942221,
+      "best_exact_thresh": 0.0,
+      "best_f1": 81.26301851073573,
+      "best_f1_thresh": 0.0,
+      "epoch": 1.37,
+      "exact": 78.00892781942221,
+      "f1": 81.2630185107358,
+      "step": 15000,
+      "total": 11873
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.825136612021857e-06,
+      "loss": 0.5361,
+      "step": 15500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.14207650273224e-06,
+      "loss": 0.5417,
+      "step": 16000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.459016393442623e-06,
+      "loss": 0.5478,
+      "step": 16500
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.775956284153006e-06,
+      "loss": 0.5157,
+      "step": 17000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.092896174863388e-06,
+      "loss": 0.5517,
+      "step": 17500
+    },
+    {
+      "HasAns_exact": 73.73481781376518,
+      "HasAns_f1": 79.64222615088413,
+      "HasAns_total": 5928,
+      "NoAns_exact": 83.51555929352396,
+      "NoAns_f1": 83.51555929352396,
+      "NoAns_total": 5945,
+      "best_exact": 78.6321906847469,
+      "best_exact_thresh": 0.0,
+      "best_f1": 81.58166568032006,
+      "best_f1_thresh": 0.0,
+      "epoch": 1.59,
+      "exact": 78.6321906847469,
+      "f1": 81.5816656803201,
+      "step": 17500,
+      "total": 11873
+    }
+  ],
+  "max_steps": 21960,
+  "num_train_epochs": 2,
+  "total_flos": 1.616435859777915e+17,
+  "trial_name": null,
+  "trial_params": null
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:11c9922bff70152d6882db2c0e1bcf88b1d47aa6153da091a6a0230a18497005
+size 2351