update

all_results.json

@@ -1,17 +1,17 @@
 {
     "epoch": 15.0,
-    "eval_accuracy": 0.9779270380613723,
-    "eval_f1": 0.966844646225661,
-    "eval_loss": 0.15447764098644257,
-    "eval_precision": 0.9669064748201439,
-    "eval_recall": 0.9667828255379008,
-    "eval_runtime": 6.5133,
+    "eval_accuracy": 0.978401158962202,
+    "eval_f1": 0.96785434549785,
+    "eval_loss": 0.15796583890914917,
+    "eval_precision": 0.9678388746803069,
+    "eval_recall": 0.9678698168100003,
+    "eval_runtime": 3.7378,
     "eval_samples": 1451,
-    "eval_samples_per_second": 222.775,
-    "eval_steps_per_second": 13.971,
-    "train_loss": 0.03180336131915158,
-    "train_runtime": 835.9078,
+    "eval_samples_per_second": 388.2,
+    "eval_steps_per_second": 24.346,
+    "train_loss": 0.03262665262993644,
+    "train_runtime": 785.1159,
     "train_samples": 13054,
-    "train_samples_per_second": 234.248,
-    "train_steps_per_second": 14.643
+    "train_samples_per_second": 249.403,
+    "train_steps_per_second": 15.59
 }

config.json

@@ -1,5 +1,5 @@
 {
-  "_name_or_path": "/lus/theta-fs0/projects/SolarWindowsADSP/shu/models/revision/bert/batterybert-cased/",
+  "_name_or_path": "/lus/theta-fs0/projects/SolarWindowsADSP/shu/models/best_models/bert-base-cased/",
   "architectures": [
     "BertForTokenClassification"
   ],

eval_results.json

@@ -1,12 +1,12 @@
 {
     "epoch": 15.0,
-    "eval_accuracy": 0.9779270380613723,
-    "eval_f1": 0.966844646225661,
-    "eval_loss": 0.15447764098644257,
-    "eval_precision": 0.9669064748201439,
-    "eval_recall": 0.9667828255379008,
-    "eval_runtime": 6.5133,
+    "eval_accuracy": 0.978401158962202,
+    "eval_f1": 0.96785434549785,
+    "eval_loss": 0.15796583890914917,
+    "eval_precision": 0.9678388746803069,
+    "eval_recall": 0.9678698168100003,
+    "eval_runtime": 3.7378,
     "eval_samples": 1451,
-    "eval_samples_per_second": 222.775,
-    "eval_steps_per_second": 13.971
+    "eval_samples_per_second": 388.2,
+    "eval_steps_per_second": 24.346
 }

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:221cfd68e45249bad38f49fffe8e8bf67e45d77610eab67421b946a4afc4a105
+oid sha256:a2501b95940948e0e7914b12893a4db2e51e91522e86608321b56de7825d840c
 size 431110641

tokenizer_config.json

@@ -1 +1 @@
-{"do_lower_case": false, "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": "/lus/theta-fs0/projects/SolarWindowsADSP/shu/models/revision/bert/batterybert-cased/", "tokenizer_class": "BertTokenizer"}
+{"do_lower_case": false, "unk_token": "[UNK]", "sep_token": "[SEP]", "pad_token": "[PAD]", "cls_token": "[CLS]", "mask_token": "[MASK]", "tokenize_chinese_chars": true, "strip_accents": null, "special_tokens_map_file": null, "name_or_path": "/lus/theta-fs0/projects/SolarWindowsADSP/shu/models/best_models/bert-base-cased/", "tokenizer_class": "BertTokenizer"}

train_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 15.0,
-    "train_loss": 0.03180336131915158,
-    "train_runtime": 835.9078,
+    "train_loss": 0.03262665262993644,
+    "train_runtime": 785.1159,
     "train_samples": 13054,
-    "train_samples_per_second": 234.248,
-    "train_steps_per_second": 14.643
+    "train_samples_per_second": 249.403,
+    "train_steps_per_second": 15.59
 }

trainer_state.json

@@ -10,335 +10,335 @@
     {
       "epoch": 0.61,
       "learning_rate": 1.9183006535947716e-05,
-      "loss": 0.3785,
+      "loss": 0.4329,
       "step": 500
     },
     {
       "epoch": 1.0,
-      "eval_accuracy": 0.9737126300539971,
-      "eval_f1": 0.9626903878151799,
-      "eval_loss": 0.09840826690196991,
-      "eval_precision": 0.9635228182546037,
-      "eval_recall": 0.9618593944819208,
-      "eval_runtime": 3.9949,
-      "eval_samples_per_second": 363.209,
-      "eval_steps_per_second": 22.779,
+      "eval_accuracy": 0.9725273278019229,
+      "eval_f1": 0.9629807384654764,
+      "eval_loss": 0.10430943965911865,
+      "eval_precision": 0.9637516411028211,
+      "eval_recall": 0.9622110681287765,
+      "eval_runtime": 2.936,
+      "eval_samples_per_second": 494.214,
+      "eval_steps_per_second": 30.995,
       "step": 816
     },
     {
       "epoch": 1.23,
       "learning_rate": 1.8366013071895427e-05,
-      "loss": 0.0836,
+      "loss": 0.0811,
       "step": 1000
     },
     {
       "epoch": 1.84,
       "learning_rate": 1.7549019607843138e-05,
-      "loss": 0.0641,
+      "loss": 0.0607,
       "step": 1500
     },
     {
       "epoch": 2.0,
-      "eval_accuracy": 0.975925194257869,
-      "eval_f1": 0.9640177271490968,
-      "eval_loss": 0.09886540472507477,
-      "eval_precision": 0.9648667691519344,
-      "eval_recall": 0.9631701780747466,
-      "eval_runtime": 3.9485,
-      "eval_samples_per_second": 367.479,
-      "eval_steps_per_second": 23.047,
+      "eval_accuracy": 0.9751349927564862,
+      "eval_f1": 0.9635150914122107,
+      "eval_loss": 0.09400150179862976,
+      "eval_precision": 0.9641165172855314,
+      "eval_recall": 0.9629144154224879,
+      "eval_runtime": 2.7493,
+      "eval_samples_per_second": 527.777,
+      "eval_steps_per_second": 33.1,
       "step": 1632
     },
     {
       "epoch": 2.45,
       "learning_rate": 1.6732026143790852e-05,
-      "loss": 0.0458,
+      "loss": 0.0428,
       "step": 2000
     },
     {
       "epoch": 3.0,
-      "eval_accuracy": 0.976741735809298,
-      "eval_f1": 0.9651566272766523,
-      "eval_loss": 0.09389109164476395,
-      "eval_precision": 0.9654808369057521,
-      "eval_recall": 0.9648326353144282,
-      "eval_runtime": 3.7903,
-      "eval_samples_per_second": 382.817,
-      "eval_steps_per_second": 24.008,
+      "eval_accuracy": 0.9755037534571316,
+      "eval_f1": 0.9632957125043962,
+      "eval_loss": 0.09555820375680923,
+      "eval_precision": 0.9633573141486811,
+      "eval_recall": 0.9632341187378113,
+      "eval_runtime": 2.7415,
+      "eval_samples_per_second": 529.269,
+      "eval_steps_per_second": 33.193,
       "step": 2448
     },
     {
       "epoch": 3.06,
       "learning_rate": 1.5915032679738563e-05,
-      "loss": 0.0394,
+      "loss": 0.0363,
       "step": 2500
     },
     {
       "epoch": 3.68,
       "learning_rate": 1.5098039215686276e-05,
-      "loss": 0.0279,
+      "loss": 0.0249,
       "step": 3000
     },
     {
       "epoch": 4.0,
-      "eval_accuracy": 0.9758725141577769,
-      "eval_f1": 0.963916701321135,
-      "eval_loss": 0.09818906337022781,
-      "eval_precision": 0.9644720417373491,
-      "eval_recall": 0.9633620000639407,
-      "eval_runtime": 3.805,
-      "eval_samples_per_second": 381.341,
-      "eval_steps_per_second": 23.916,
+      "eval_accuracy": 0.9765046753588832,
+      "eval_f1": 0.9649540551338394,
+      "eval_loss": 0.0989251509308815,
+      "eval_precision": 0.9646919734151329,
+      "eval_recall": 0.9652162792928163,
+      "eval_runtime": 2.7759,
+      "eval_samples_per_second": 522.708,
+      "eval_steps_per_second": 32.782,
       "step": 3264
     },
     {
       "epoch": 4.29,
       "learning_rate": 1.4281045751633989e-05,
-      "loss": 0.0254,
+      "loss": 0.0221,
       "step": 3500
     },
     {
       "epoch": 4.9,
       "learning_rate": 1.3464052287581701e-05,
-      "loss": 0.0194,
+      "loss": 0.0177,
       "step": 4000
     },
     {
       "epoch": 5.0,
-      "eval_accuracy": 0.9765310154089293,
-      "eval_f1": 0.9647630619684083,
-      "eval_loss": 0.10570573806762695,
-      "eval_precision": 0.9649173302631999,
-      "eval_recall": 0.9646088429937019,
-      "eval_runtime": 4.1667,
-      "eval_samples_per_second": 348.241,
-      "eval_steps_per_second": 21.84,
+      "eval_accuracy": 0.976451995258791,
+      "eval_f1": 0.9646675296035285,
+      "eval_loss": 0.10913769155740738,
+      "eval_precision": 0.9643747204294204,
+      "eval_recall": 0.9649605166405576,
+      "eval_runtime": 3.7551,
+      "eval_samples_per_second": 386.403,
+      "eval_steps_per_second": 24.233,
       "step": 4080
     },
     {
       "epoch": 5.51,
       "learning_rate": 1.2647058823529412e-05,
-      "loss": 0.0141,
+      "loss": 0.0129,
       "step": 4500
     },
     {
       "epoch": 6.0,
-      "eval_accuracy": 0.9767680758593441,
-      "eval_f1": 0.9650685041006539,
-      "eval_loss": 0.11329083889722824,
-      "eval_precision": 0.9651765157329241,
-      "eval_recall": 0.9649605166405576,
-      "eval_runtime": 3.8277,
-      "eval_samples_per_second": 379.077,
-      "eval_steps_per_second": 23.774,
+      "eval_accuracy": 0.9780060582115107,
+      "eval_f1": 0.9675017575254042,
+      "eval_loss": 0.11124212294816971,
+      "eval_precision": 0.967038231818327,
+      "eval_recall": 0.9679657278045973,
+      "eval_runtime": 3.8249,
+      "eval_samples_per_second": 379.357,
+      "eval_steps_per_second": 23.792,
       "step": 4896
     },
     {
       "epoch": 6.13,
       "learning_rate": 1.1830065359477125e-05,
-      "loss": 0.0142,
+      "loss": 0.0119,
       "step": 5000
     },
     {
       "epoch": 6.74,
       "learning_rate": 1.1013071895424838e-05,
-      "loss": 0.0097,
+      "loss": 0.0084,
       "step": 5500
     },
     {
       "epoch": 7.0,
-      "eval_accuracy": 0.9780060582115107,
-      "eval_f1": 0.9669496810092579,
-      "eval_loss": 0.1174820214509964,
-      "eval_precision": 0.9672125903652997,
-      "eval_recall": 0.9666869145433038,
-      "eval_runtime": 2.7647,
-      "eval_samples_per_second": 524.836,
-      "eval_steps_per_second": 32.915,
+      "eval_accuracy": 0.9783221388120638,
+      "eval_f1": 0.9677996994212259,
+      "eval_loss": 0.12139276415109634,
+      "eval_precision": 0.967985415933732,
+      "eval_recall": 0.9676140541577416,
+      "eval_runtime": 3.747,
+      "eval_samples_per_second": 387.246,
+      "eval_steps_per_second": 24.286,
       "step": 5712
     },
     {
       "epoch": 7.35,
       "learning_rate": 1.0196078431372549e-05,
-      "loss": 0.0085,
+      "loss": 0.0076,
       "step": 6000
     },
     {
       "epoch": 7.97,
       "learning_rate": 9.379084967320261e-06,
-      "loss": 0.0078,
+      "loss": 0.0061,
       "step": 6500
     },
     {
       "epoch": 8.0,
-      "eval_accuracy": 0.9770841564598973,
-      "eval_f1": 0.965837913843794,
-      "eval_loss": 0.12893585860729218,
-      "eval_precision": 0.9654368311771283,
-      "eval_recall": 0.9662393299018511,
-      "eval_runtime": 3.7734,
-      "eval_samples_per_second": 384.529,
-      "eval_steps_per_second": 24.116,
+      "eval_accuracy": 0.9775319373106809,
+      "eval_f1": 0.9664444551020734,
+      "eval_loss": 0.13343702256679535,
+      "eval_precision": 0.9665217113257019,
+      "eval_recall": 0.9663672112279804,
+      "eval_runtime": 2.7649,
+      "eval_samples_per_second": 524.787,
+      "eval_steps_per_second": 32.912,
       "step": 6528
     },
     {
       "epoch": 8.58,
       "learning_rate": 8.562091503267974e-06,
-      "loss": 0.0061,
+      "loss": 0.0048,
       "step": 7000
     },
     {
       "epoch": 9.0,
-      "eval_accuracy": 0.9767680758593441,
-      "eval_f1": 0.9654114677360088,
-      "eval_loss": 0.1360652893781662,
-      "eval_precision": 0.9648719422622469,
-      "eval_recall": 0.96595159691806,
-      "eval_runtime": 4.9673,
-      "eval_samples_per_second": 292.108,
-      "eval_steps_per_second": 18.32,
+      "eval_accuracy": 0.9775055972606348,
+      "eval_f1": 0.9661577811525858,
+      "eval_loss": 0.13892702758312225,
+      "eval_precision": 0.9662041181736795,
+      "eval_recall": 0.9661114485757217,
+      "eval_runtime": 3.8371,
+      "eval_samples_per_second": 378.146,
+      "eval_steps_per_second": 23.716,
       "step": 7344
     },
     {
       "epoch": 9.19,
       "learning_rate": 7.745098039215687e-06,
-      "loss": 0.0057,
+      "loss": 0.0047,
       "step": 7500
     },
     {
       "epoch": 9.8,
       "learning_rate": 6.928104575163399e-06,
-      "loss": 0.0046,
+      "loss": 0.0044,
       "step": 8000
     },
     {
       "epoch": 10.0,
-      "eval_accuracy": 0.976741735809298,
-      "eval_f1": 0.9652162792928163,
-      "eval_loss": 0.14587946236133575,
-      "eval_precision": 0.9652162792928163,
-      "eval_recall": 0.9652162792928163,
-      "eval_runtime": 2.7677,
-      "eval_samples_per_second": 524.265,
-      "eval_steps_per_second": 32.879,
+      "eval_accuracy": 0.9781640985117872,
+      "eval_f1": 0.967271564473579,
+      "eval_loss": 0.14489226043224335,
+      "eval_precision": 0.9674726540011515,
+      "eval_recall": 0.9670705585216919,
+      "eval_runtime": 3.7406,
+      "eval_samples_per_second": 387.909,
+      "eval_steps_per_second": 24.328,
       "step": 8160
     },
     {
       "epoch": 10.42,
       "learning_rate": 6.111111111111112e-06,
-      "loss": 0.0042,
+      "loss": 0.0034,
       "step": 8500
     },
     {
       "epoch": 11.0,
-      "eval_accuracy": 0.977848017911234,
-      "eval_f1": 0.9667578068786875,
-      "eval_loss": 0.14097169041633606,
-      "eval_precision": 0.9670206640649991,
-      "eval_recall": 0.9664950925541098,
-      "eval_runtime": 2.8235,
-      "eval_samples_per_second": 513.9,
-      "eval_steps_per_second": 32.229,
+      "eval_accuracy": 0.9779797181614646,
+      "eval_f1": 0.9672149490880608,
+      "eval_loss": 0.14611582458019257,
+      "eval_precision": 0.9671994884910486,
+      "eval_recall": 0.9672304101793535,
+      "eval_runtime": 2.8969,
+      "eval_samples_per_second": 500.886,
+      "eval_steps_per_second": 31.413,
       "step": 8976
     },
     {
       "epoch": 11.03,
       "learning_rate": 5.294117647058824e-06,
-      "loss": 0.0041,
+      "loss": 0.0034,
       "step": 9000
     },
     {
       "epoch": 11.64,
       "learning_rate": 4.477124183006537e-06,
-      "loss": 0.0034,
+      "loss": 0.0025,
       "step": 9500
     },
     {
       "epoch": 12.0,
-      "eval_accuracy": 0.9781377584617411,
-      "eval_f1": 0.9671292447400396,
-      "eval_loss": 0.14326775074005127,
-      "eval_precision": 0.9672838913940325,
-      "eval_recall": 0.9669746475270948,
-      "eval_runtime": 3.9931,
-      "eval_samples_per_second": 363.381,
-      "eval_steps_per_second": 22.79,
+      "eval_accuracy": 0.9784274990122481,
+      "eval_f1": 0.9676501159350764,
+      "eval_loss": 0.14966140687465668,
+      "eval_precision": 0.9680061428205784,
+      "eval_recall": 0.9672943508424182,
+      "eval_runtime": 2.7972,
+      "eval_samples_per_second": 518.731,
+      "eval_steps_per_second": 32.532,
       "step": 9792
     },
     {
       "epoch": 12.25,
       "learning_rate": 3.6601307189542484e-06,
-      "loss": 0.0027,
+      "loss": 0.0022,
       "step": 10000
     },
     {
       "epoch": 12.87,
       "learning_rate": 2.843137254901961e-06,
-      "loss": 0.0026,
+      "loss": 0.0024,
       "step": 10500
     },
     {
       "epoch": 13.0,
-      "eval_accuracy": 0.9774529171605426,
-      "eval_f1": 0.9662260441474992,
-      "eval_loss": 0.15390940010547638,
-      "eval_precision": 0.9661488300728807,
-      "eval_recall": 0.9663032705649157,
-      "eval_runtime": 3.7949,
-      "eval_samples_per_second": 382.357,
-      "eval_steps_per_second": 23.98,
+      "eval_accuracy": 0.977848017911234,
+      "eval_f1": 0.9670199670679264,
+      "eval_loss": 0.15780866146087646,
+      "eval_precision": 0.9670972692971798,
+      "eval_recall": 0.9669426771955625,
+      "eval_runtime": 3.8111,
+      "eval_samples_per_second": 380.734,
+      "eval_steps_per_second": 23.878,
       "step": 10608
     },
     {
       "epoch": 13.48,
       "learning_rate": 2.0261437908496734e-06,
-      "loss": 0.0022,
+      "loss": 0.0014,
       "step": 11000
     },
     {
       "epoch": 14.0,
-      "eval_accuracy": 0.9776109574608192,
-      "eval_f1": 0.9663320117662106,
-      "eval_loss": 0.15523000061511993,
-      "eval_precision": 0.9664247114124005,
-      "eval_recall": 0.9662393299018511,
-      "eval_runtime": 3.7736,
-      "eval_samples_per_second": 384.509,
-      "eval_steps_per_second": 24.115,
+      "eval_accuracy": 0.978111418411695,
+      "eval_f1": 0.9674521213671389,
+      "eval_loss": 0.15866732597351074,
+      "eval_precision": 0.9675139888089528,
+      "eval_recall": 0.9673902618370153,
+      "eval_runtime": 2.8382,
+      "eval_samples_per_second": 511.243,
+      "eval_steps_per_second": 32.063,
       "step": 11424
     },
     {
       "epoch": 14.09,
       "learning_rate": 1.2091503267973858e-06,
-      "loss": 0.0021,
+      "loss": 0.0019,
       "step": 11500
     },
     {
       "epoch": 14.71,
       "learning_rate": 3.921568627450981e-07,
-      "loss": 0.0018,
+      "loss": 0.0016,
       "step": 12000
     },
     {
       "epoch": 15.0,
-      "eval_accuracy": 0.9779270380613723,
-      "eval_f1": 0.966844646225661,
-      "eval_loss": 0.15447764098644257,
-      "eval_precision": 0.9669064748201439,
-      "eval_recall": 0.9667828255379008,
-      "eval_runtime": 3.7735,
-      "eval_samples_per_second": 384.523,
-      "eval_steps_per_second": 24.116,
+      "eval_accuracy": 0.978401158962202,
+      "eval_f1": 0.96785434549785,
+      "eval_loss": 0.15796583890914917,
+      "eval_precision": 0.9678388746803069,
+      "eval_recall": 0.9678698168100003,
+      "eval_runtime": 3.7576,
+      "eval_samples_per_second": 386.151,
+      "eval_steps_per_second": 24.218,
       "step": 12240
     },
     {
       "epoch": 15.0,
       "step": 12240,
       "total_flos": 5.119363918698906e+16,
-      "train_loss": 0.03180336131915158,
-      "train_runtime": 835.9078,
-      "train_samples_per_second": 234.248,
-      "train_steps_per_second": 14.643
+      "train_loss": 0.03262665262993644,
+      "train_runtime": 785.1159,
+      "train_samples_per_second": 249.403,
+      "train_steps_per_second": 15.59
     }
   ],
   "max_steps": 12240,

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:28805bbddbea78cbc7984bc64d6c51834153424f84a3475453bb119c607a9e77
+oid sha256:f2ad2280933fd2dfde8ee13fc62c6b195937dddc4b2505fce56231e549d3c475
 size 3119