v-1.0.1

config.json

@@ -1,5 +1,5 @@
 {
-  "_name_or_path": "marefa-ner",
+  "_name_or_path": "xlm-roberta-large",
   "architectures": [
     "XLMRobertaForTokenClassification"
   ],
@@ -11,12 +11,48 @@
   "hidden_dropout_prob": 0.1,
   "hidden_size": 1024,
   "id2label": {
-    "0": "O", "1": "B-job", "2": "I-job", "3": "B-nationality", "4": "B-person", "5": "I-person", "6": "B-location", "7": "B-time", "8": "I-time", "9": "B-event", "10": "I-event", "11": "B-organization", "12": "I-organization", "13": "I-location", "14": "I-nationality", "15": "B-product", "16": "I-product", "17": "B-artwork", "18": "I-artwork"
+    "0": "LABEL_0",
+    "1": "LABEL_1",
+    "2": "LABEL_2",
+    "3": "LABEL_3",
+    "4": "LABEL_4",
+    "5": "LABEL_5",
+    "6": "LABEL_6",
+    "7": "LABEL_7",
+    "8": "LABEL_8",
+    "9": "LABEL_9",
+    "10": "LABEL_10",
+    "11": "LABEL_11",
+    "12": "LABEL_12",
+    "13": "LABEL_13",
+    "14": "LABEL_14",
+    "15": "LABEL_15",
+    "16": "LABEL_16",
+    "17": "LABEL_17",
+    "18": "LABEL_18"
   },
   "initializer_range": 0.02,
   "intermediate_size": 4096,
   "label2id": {
-      "O": 0, "B-job": 1, "I-job": 2, "B-nationality": 3, "B-person": 4, "I-person": 5, "B-location": 6, "B-time": 7, "I-time": 8, "B-event": 9, "I-event": 10, "B-organization": 11, "I-organization": 12, "I-location": 13, "I-nationality": 14, "B-product": 15, "I-product": 16, "B-artwork": 17, "I-artwork": 18
+    "LABEL_0": 0,
+    "LABEL_1": 1,
+    "LABEL_10": 10,
+    "LABEL_11": 11,
+    "LABEL_12": 12,
+    "LABEL_13": 13,
+    "LABEL_14": 14,
+    "LABEL_15": 15,
+    "LABEL_16": 16,
+    "LABEL_17": 17,
+    "LABEL_18": 18,
+    "LABEL_2": 2,
+    "LABEL_3": 3,
+    "LABEL_4": 4,
+    "LABEL_5": 5,
+    "LABEL_6": 6,
+    "LABEL_7": 7,
+    "LABEL_8": 8,
+    "LABEL_9": 9
   },
   "layer_norm_eps": 1e-05,
   "max_position_embeddings": 514,
@@ -26,7 +62,7 @@
   "output_past": true,
   "pad_token_id": 1,
   "position_embedding_type": "absolute",
-  "transformers_version": "4.3.2",
+  "transformers_version": "4.6.0",
   "type_vocab_size": 1,
   "use_cache": true,
   "vocab_size": 250002

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:fcb9993a3bf8dc52862ea7ee98642de4675be90a7efccce629abc35749268364
-size 2235610244
+oid sha256:cf4a4d17471d6908750beeb0998d2cb4a744dbe3ca4aaabcd352a28d231ae3b9
+size 2235613943

scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b8da0636dfe95dd35ec9c3de89e2da9babbe4b94bbb061482b1a4915cdb79c23
+oid sha256:c022ff4c770b8f48647cabb531eed6c0b6843207e90c408c88bfa0ee2fcb08dd
 size 623

special_tokens_map.json

@@ -1 +1 @@
-{"bos_token": "<s>", "eos_token": "</s>", "unk_token": "<unk>", "sep_token": "</s>", "pad_token": "<pad>", "cls_token": "<s>", "mask_token": {"content": "<mask>", "single_word": false, "lstrip": true, "rstrip": false, "normalized": true}}
+{"bos_token": "<s>", "eos_token": "</s>", "unk_token": "<unk>", "sep_token": "</s>", "pad_token": "<pad>", "cls_token": "<s>", "mask_token": {"content": "<mask>", "single_word": false, "lstrip": true, "rstrip": false, "normalized": false}}

tokenizer_config.json

@@ -1 +1 @@
-{"bos_token": "<s>", "eos_token": "</s>", "unk_token": "<unk>", "sep_token": "</s>", "cls_token": "<s>", "pad_token": "<pad>", "mask_token": {"content": "<mask>", "single_word": false, "lstrip": true, "rstrip": false, "normalized": true, "__type": "AddedToken"}, "do_lower_case": false, "model_max_length": 512, "name_or_path": "xlm-roberta-large"}
+{"bos_token": "<s>", "eos_token": "</s>", "sep_token": "</s>", "cls_token": "<s>", "unk_token": "<unk>", "pad_token": "<pad>", "mask_token": {"content": "<mask>", "single_word": false, "lstrip": true, "rstrip": false, "normalized": true, "__type": "AddedToken"}, "model_max_length": 512, "special_tokens_map_file": null, "name_or_path": "xlm-roberta-large"}

trainer_state.json

@@ -0,0 +1,571 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.704618424302297,
+  "global_step": 30000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9794187865316542e-05,
+      "loss": 0.7638,
+      "step": 500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958837573063308e-05,
+      "loss": 0.5165,
+      "step": 1000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.938256359594962e-05,
+      "loss": 0.457,
+      "step": 1500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.917675146126616e-05,
+      "loss": 0.4516,
+      "step": 2000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8970939326582696e-05,
+      "loss": 0.4095,
+      "step": 2500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8765127191899236e-05,
+      "loss": 0.4188,
+      "step": 3000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8559315057215776e-05,
+      "loss": 0.4028,
+      "step": 3500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8353502922532313e-05,
+      "loss": 0.3846,
+      "step": 4000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8147690787848853e-05,
+      "loss": 0.3813,
+      "step": 4500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7941878653165393e-05,
+      "loss": 0.3842,
+      "step": 5000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.773606651848193e-05,
+      "loss": 0.3977,
+      "step": 5500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.753025438379847e-05,
+      "loss": 0.3749,
+      "step": 6000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.732444224911501e-05,
+      "loss": 0.3578,
+      "step": 6500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.711863011443155e-05,
+      "loss": 0.339,
+      "step": 7000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6912817979748087e-05,
+      "loss": 0.3704,
+      "step": 7500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6707005845064627e-05,
+      "loss": 0.3423,
+      "step": 8000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.6501193710381164e-05,
+      "loss": 0.2788,
+      "step": 8500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.6295381575697704e-05,
+      "loss": 0.2832,
+      "step": 9000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.6089569441014244e-05,
+      "loss": 0.3032,
+      "step": 9500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.588375730633078e-05,
+      "loss": 0.2774,
+      "step": 10000
+    },
+    {
+      "epoch": 1.23,
+      "eval_artwork": {
+        "f1": 0.3447251114413076,
+        "number": 295,
+        "precision": 0.30687830687830686,
+        "recall": 0.39322033898305087
+      },
+      "eval_event": {
+        "f1": 0.43632567849686854,
+        "number": 519,
+        "precision": 0.4760820045558087,
+        "recall": 0.4026974951830443
+      },
+      "eval_job": {
+        "f1": 0.6786657067434606,
+        "number": 2257,
+        "precision": 0.7403141361256544,
+        "recall": 0.6264953478068233
+      },
+      "eval_location": {
+        "f1": 0.6351052692799815,
+        "number": 4375,
+        "precision": 0.6466129796305069,
+        "recall": 0.624
+      },
+      "eval_loss": 0.9446586966514587,
+      "eval_nationality": {
+        "f1": 0.6808716707021791,
+        "number": 2137,
+        "precision": 0.7054691419969895,
+        "recall": 0.6579316799251287
+      },
+      "eval_organization": {
+        "f1": 0.5520833333333333,
+        "number": 1982,
+        "precision": 0.6044417767106842,
+        "recall": 0.5080726538849647
+      },
+      "eval_overall_accuracy": 0.8332146471271692,
+      "eval_overall_f1": 0.6393731397679646,
+      "eval_overall_precision": 0.6567667061833157,
+      "eval_overall_recall": 0.6228770933191313,
+      "eval_person": {
+        "f1": 0.7589572006134114,
+        "number": 3322,
+        "precision": 0.7068293949623474,
+        "recall": 0.8193859121011439
+      },
+      "eval_product": {
+        "f1": 0.3201754385964912,
+        "number": 323,
+        "precision": 0.5488721804511278,
+        "recall": 0.2260061919504644
+      },
+      "eval_runtime": 43.8587,
+      "eval_samples_per_second": 35.911,
+      "eval_time": {
+        "f1": 0.5435339308578744,
+        "number": 1689,
+        "precision": 0.5916376306620209,
+        "recall": 0.5026642984014209
+      },
+      "step": 10000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.567794517164732e-05,
+      "loss": 0.2914,
+      "step": 10500
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.547213303696386e-05,
+      "loss": 0.2878,
+      "step": 11000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.52663209022804e-05,
+      "loss": 0.2649,
+      "step": 11500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.506050876759694e-05,
+      "loss": 0.2794,
+      "step": 12000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.4854696632913477e-05,
+      "loss": 0.288,
+      "step": 12500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.4648884498230017e-05,
+      "loss": 0.2663,
+      "step": 13000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4443072363546555e-05,
+      "loss": 0.2853,
+      "step": 13500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4237260228863094e-05,
+      "loss": 0.2718,
+      "step": 14000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4031448094179634e-05,
+      "loss": 0.2944,
+      "step": 14500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.3825635959496172e-05,
+      "loss": 0.2804,
+      "step": 15000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.3619823824812712e-05,
+      "loss": 0.2631,
+      "step": 15500
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3414011690129251e-05,
+      "loss": 0.2937,
+      "step": 16000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3208199555445791e-05,
+      "loss": 0.2301,
+      "step": 16500
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.300238742076233e-05,
+      "loss": 0.2136,
+      "step": 17000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2796575286078868e-05,
+      "loss": 0.2054,
+      "step": 17500
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2590763151395406e-05,
+      "loss": 0.2215,
+      "step": 18000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2384951016711945e-05,
+      "loss": 0.2237,
+      "step": 18500
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.2179138882028485e-05,
+      "loss": 0.2124,
+      "step": 19000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1973326747345024e-05,
+      "loss": 0.2042,
+      "step": 19500
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.1767514612661564e-05,
+      "loss": 0.2253,
+      "step": 20000
+    },
+    {
+      "epoch": 2.47,
+      "eval_artwork": {
+        "f1": 0.30398069963811825,
+        "number": 295,
+        "precision": 0.23595505617977527,
+        "recall": 0.4271186440677966
+      },
+      "eval_event": {
+        "f1": 0.38726790450928383,
+        "number": 519,
+        "precision": 0.35784313725490197,
+        "recall": 0.42196531791907516
+      },
+      "eval_job": {
+        "f1": 0.6671604938271605,
+        "number": 2257,
+        "precision": 0.7534857780256553,
+        "recall": 0.5985821887461231
+      },
+      "eval_location": {
+        "f1": 0.6328576483075834,
+        "number": 4375,
+        "precision": 0.6537524366471735,
+        "recall": 0.6132571428571428
+      },
+      "eval_loss": 1.0017756223678589,
+      "eval_nationality": {
+        "f1": 0.6760161635369623,
+        "number": 2137,
+        "precision": 0.6869565217391305,
+        "recall": 0.6654188114178755
+      },
+      "eval_organization": {
+        "f1": 0.5447824850725049,
+        "number": 1982,
+        "precision": 0.6241042345276873,
+        "recall": 0.48335015136226034
+      },
+      "eval_overall_accuracy": 0.8219457041840011,
+      "eval_overall_f1": 0.6276163319784019,
+      "eval_overall_precision": 0.643866309827597,
+      "eval_overall_recall": 0.6121664003787206,
+      "eval_person": {
+        "f1": 0.7596614950634696,
+        "number": 3322,
+        "precision": 0.7147027600849257,
+        "recall": 0.8106562311860325
+      },
+      "eval_product": {
+        "f1": 0.3699421965317919,
+        "number": 323,
+        "precision": 0.4897959183673469,
+        "recall": 0.29721362229102166
+      },
+      "eval_runtime": 44.5273,
+      "eval_samples_per_second": 35.372,
+      "eval_time": {
+        "f1": 0.5069974554707379,
+        "number": 1689,
+        "precision": 0.547766323024055,
+        "recall": 0.4718768502072232
+      },
+      "step": 20000
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.1561702477978102e-05,
+      "loss": 0.2197,
+      "step": 20500
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.1355890343294642e-05,
+      "loss": 0.2239,
+      "step": 21000
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.115007820861118e-05,
+      "loss": 0.2231,
+      "step": 21500
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.0944266073927721e-05,
+      "loss": 0.2045,
+      "step": 22000
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.0738453939244258e-05,
+      "loss": 0.1935,
+      "step": 22500
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0532641804560796e-05,
+      "loss": 0.2156,
+      "step": 23000
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0326829669877336e-05,
+      "loss": 0.2088,
+      "step": 23500
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0121017535193875e-05,
+      "loss": 0.2022,
+      "step": 24000
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 9.915205400510415e-06,
+      "loss": 0.1918,
+      "step": 24500
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.709393265826953e-06,
+      "loss": 0.1494,
+      "step": 25000
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.503581131143494e-06,
+      "loss": 0.1502,
+      "step": 25500
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 9.297768996460032e-06,
+      "loss": 0.1741,
+      "step": 26000
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 9.09195686177657e-06,
+      "loss": 0.1609,
+      "step": 26500
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 8.88614472709311e-06,
+      "loss": 0.1584,
+      "step": 27000
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.680332592409649e-06,
+      "loss": 0.1524,
+      "step": 27500
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 8.47452045772619e-06,
+      "loss": 0.1513,
+      "step": 28000
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 8.268708323042726e-06,
+      "loss": 0.1595,
+      "step": 28500
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 8.062896188359266e-06,
+      "loss": 0.1631,
+      "step": 29000
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 7.857084053675805e-06,
+      "loss": 0.1795,
+      "step": 29500
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 7.651271918992345e-06,
+      "loss": 0.1602,
+      "step": 30000
+    },
+    {
+      "epoch": 3.7,
+      "eval_artwork": {
+        "f1": 0.32594936708860756,
+        "number": 295,
+        "precision": 0.3056379821958457,
+        "recall": 0.34915254237288135
+      },
+      "eval_event": {
+        "f1": 0.42843232716650437,
+        "number": 519,
+        "precision": 0.4330708661417323,
+        "recall": 0.4238921001926782
+      },
+      "eval_job": {
+        "f1": 0.6932195353247985,
+        "number": 2257,
+        "precision": 0.7455379908210097,
+        "recall": 0.6477625166149756
+      },
+      "eval_location": {
+        "f1": 0.6270144927536232,
+        "number": 4375,
+        "precision": 0.636235294117647,
+        "recall": 0.6180571428571429
+      },
+      "eval_loss": 1.0499603748321533,
+      "eval_nationality": {
+        "f1": 0.6754837129561596,
+        "number": 2137,
+        "precision": 0.7086330935251799,
+        "recall": 0.6452971455311184
+      },
+      "eval_organization": {
+        "f1": 0.5687919463087249,
+        "number": 1982,
+        "precision": 0.6380175658720201,
+        "recall": 0.5131180625630676
+      },
+      "eval_overall_accuracy": 0.8283547593168672,
+      "eval_overall_f1": 0.6409662071489569,
+      "eval_overall_precision": 0.6613380326011706,
+      "eval_overall_recall": 0.6218119415350021,
+      "eval_person": {
+        "f1": 0.7734217353704237,
+        "number": 3322,
+        "precision": 0.7419800884955752,
+        "recall": 0.8076459963877183
+      },
+      "eval_product": {
+        "f1": 0.4022556390977444,
+        "number": 323,
+        "precision": 0.5119617224880383,
+        "recall": 0.33126934984520123
+      },
+      "eval_runtime": 43.8719,
+      "eval_samples_per_second": 35.9,
+      "eval_time": {
+        "f1": 0.5277161862527716,
+        "number": 1689,
+        "precision": 0.5674386920980926,
+        "recall": 0.4931912374185909
+      },
+      "step": 30000
+    }
+  ],
+  "max_steps": 48588,
+  "num_train_epochs": 6,
+  "total_flos": 3961931727576942.0,
+  "trial_name": null,
+  "trial_params": null
+}

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d8929d8bac9dae4c7101ccfe58a662b7132989582536babf570858e8bf404898
-size 3055
+oid sha256:e527ceb877458f834cb18192456e4ff3a0e2f3d9c8ae07805f23f677c165715f
+size 2479