pos

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a2501b95940948e0e7914b12893a4db2e51e91522e86608321b56de7825d840c
+oid sha256:fe4fb75cd82c67cea63b9bb08734e851e4e0e2c023be07ce4b9534b582b90862
 size 431110641

trainer_state.json

@@ -1,349 +1,406 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 15.0,
-  "global_step": 12240,
+  "epoch": 18.0,
+  "global_step": 14688,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
       "epoch": 0.61,
-      "learning_rate": 1.9183006535947716e-05,
-      "loss": 0.4329,
+      "learning_rate": 1.9387254901960785e-05,
+      "loss": 0.4753,
       "step": 500
     },
     {
       "epoch": 1.0,
-      "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,
+      "eval_accuracy": 0.9720334966029388,
+      "eval_f1": 0.9606774786290644,
+      "eval_loss": 0.10676991939544678,
+      "eval_precision": 0.9602632083306715,
+      "eval_recall": 0.9610921065251447,
+      "eval_runtime": 4.3302,
+      "eval_samples_per_second": 335.092,
+      "eval_steps_per_second": 21.015,
       "step": 816
     },
     {
       "epoch": 1.23,
-      "learning_rate": 1.8366013071895427e-05,
-      "loss": 0.0811,
+      "learning_rate": 1.877450980392157e-05,
+      "loss": 0.0981,
       "step": 1000
     },
     {
       "epoch": 1.84,
-      "learning_rate": 1.7549019607843138e-05,
-      "loss": 0.0607,
+      "learning_rate": 1.8161764705882355e-05,
+      "loss": 0.0753,
       "step": 1500
     },
     {
       "epoch": 2.0,
-      "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,
+      "eval_accuracy": 0.9730605150892716,
+      "eval_f1": 0.9629144363665415,
+      "eval_loss": 0.10854744911193848,
+      "eval_precision": 0.9624992014310356,
+      "eval_recall": 0.9633300297324083,
+      "eval_runtime": 4.0978,
+      "eval_samples_per_second": 354.094,
+      "eval_steps_per_second": 22.207,
       "step": 1632
     },
     {
       "epoch": 2.45,
-      "learning_rate": 1.6732026143790852e-05,
-      "loss": 0.0428,
+      "learning_rate": 1.7549019607843138e-05,
+      "loss": 0.0535,
       "step": 2000
     },
     {
       "epoch": 3.0,
-      "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,
+      "eval_accuracy": 0.9732580186443356,
+      "eval_f1": 0.9634966531942426,
+      "eval_loss": 0.10905417799949646,
+      "eval_precision": 0.9628967367009388,
+      "eval_recall": 0.9640973176891844,
+      "eval_runtime": 4.1017,
+      "eval_samples_per_second": 353.751,
+      "eval_steps_per_second": 22.186,
       "step": 2448
     },
     {
       "epoch": 3.06,
-      "learning_rate": 1.5915032679738563e-05,
-      "loss": 0.0363,
+      "learning_rate": 1.693627450980392e-05,
+      "loss": 0.046,
       "step": 2500
     },
     {
       "epoch": 3.68,
-      "learning_rate": 1.5098039215686276e-05,
-      "loss": 0.0249,
+      "learning_rate": 1.6323529411764708e-05,
+      "loss": 0.0306,
       "step": 3000
     },
     {
       "epoch": 4.0,
-      "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,
+      "eval_accuracy": 0.9720927476694581,
+      "eval_f1": 0.9631510853418946,
+      "eval_loss": 0.1200980469584465,
+      "eval_precision": 0.9623978549540347,
+      "eval_recall": 0.9639054956999904,
+      "eval_runtime": 4.0941,
+      "eval_samples_per_second": 354.409,
+      "eval_steps_per_second": 22.227,
       "step": 3264
     },
     {
       "epoch": 4.29,
-      "learning_rate": 1.4281045751633989e-05,
-      "loss": 0.0221,
+      "learning_rate": 1.571078431372549e-05,
+      "loss": 0.0271,
       "step": 3500
     },
     {
       "epoch": 4.9,
-      "learning_rate": 1.3464052287581701e-05,
-      "loss": 0.0177,
+      "learning_rate": 1.5098039215686276e-05,
+      "loss": 0.0222,
       "step": 4000
     },
     {
       "epoch": 5.0,
-      "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,
+      "eval_accuracy": 0.9717372412703429,
+      "eval_f1": 0.9625902498242924,
+      "eval_loss": 0.1297336369752884,
+      "eval_precision": 0.9618834796488428,
+      "eval_recall": 0.963298059400876,
+      "eval_runtime": 3.1221,
+      "eval_samples_per_second": 464.754,
+      "eval_steps_per_second": 29.147,
       "step": 4080
     },
     {
       "epoch": 5.51,
-      "learning_rate": 1.2647058823529412e-05,
-      "loss": 0.0129,
+      "learning_rate": 1.448529411764706e-05,
+      "loss": 0.0165,
       "step": 4500
     },
     {
       "epoch": 6.0,
-      "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,
+      "eval_accuracy": 0.9724285037130669,
+      "eval_f1": 0.9634232047022744,
+      "eval_loss": 0.1293230503797531,
+      "eval_precision": 0.9626544096523988,
+      "eval_recall": 0.9641932286837814,
+      "eval_runtime": 4.1611,
+      "eval_samples_per_second": 348.706,
+      "eval_steps_per_second": 21.869,
       "step": 4896
     },
     {
       "epoch": 6.13,
-      "learning_rate": 1.1830065359477125e-05,
-      "loss": 0.0119,
+      "learning_rate": 1.3872549019607844e-05,
+      "loss": 0.0157,
       "step": 5000
     },
     {
       "epoch": 6.74,
-      "learning_rate": 1.1013071895424838e-05,
-      "loss": 0.0084,
+      "learning_rate": 1.3259803921568627e-05,
+      "loss": 0.0106,
       "step": 5500
     },
     {
       "epoch": 7.0,
-      "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,
+      "eval_accuracy": 0.9736727761099699,
+      "eval_f1": 0.9648310228071295,
+      "eval_loss": 0.14171898365020752,
+      "eval_precision": 0.9639996170172023,
+      "eval_recall": 0.965663863934269,
+      "eval_runtime": 3.07,
+      "eval_samples_per_second": 472.638,
+      "eval_steps_per_second": 29.642,
       "step": 5712
     },
     {
       "epoch": 7.35,
-      "learning_rate": 1.0196078431372549e-05,
-      "loss": 0.0076,
+      "learning_rate": 1.2647058823529412e-05,
+      "loss": 0.0099,
       "step": 6000
     },
     {
       "epoch": 7.97,
-      "learning_rate": 9.379084967320261e-06,
-      "loss": 0.0061,
+      "learning_rate": 1.2034313725490197e-05,
+      "loss": 0.0084,
       "step": 6500
     },
     {
       "epoch": 8.0,
-      "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,
+      "eval_accuracy": 0.9737912782430084,
+      "eval_f1": 0.9657231348883248,
+      "eval_loss": 0.15885183215141296,
+      "eval_precision": 0.9645066649658779,
+      "eval_recall": 0.9669426771955625,
+      "eval_runtime": 3.0514,
+      "eval_samples_per_second": 475.516,
+      "eval_steps_per_second": 29.822,
       "step": 6528
     },
     {
       "epoch": 8.58,
-      "learning_rate": 8.562091503267974e-06,
-      "loss": 0.0048,
+      "learning_rate": 1.142156862745098e-05,
+      "loss": 0.007,
       "step": 7000
     },
     {
       "epoch": 9.0,
-      "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,
+      "eval_accuracy": 0.9737320271764892,
+      "eval_f1": 0.9660194174757282,
+      "eval_loss": 0.16605724394321442,
+      "eval_precision": 0.9650023927261127,
+      "eval_recall": 0.9670385881901595,
+      "eval_runtime": 4.1246,
+      "eval_samples_per_second": 351.79,
+      "eval_steps_per_second": 22.063,
       "step": 7344
     },
     {
       "epoch": 9.19,
-      "learning_rate": 7.745098039215687e-06,
-      "loss": 0.0047,
+      "learning_rate": 1.0808823529411765e-05,
+      "loss": 0.0066,
       "step": 7500
     },
     {
       "epoch": 9.8,
-      "learning_rate": 6.928104575163399e-06,
-      "loss": 0.0044,
+      "learning_rate": 1.0196078431372549e-05,
+      "loss": 0.0051,
       "step": 8000
     },
     {
       "epoch": 10.0,
-      "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,
+      "eval_accuracy": 0.9731592668668037,
+      "eval_f1": 0.964738573573094,
+      "eval_loss": 0.17197225987911224,
+      "eval_precision": 0.9638150547241456,
+      "eval_recall": 0.965663863934269,
+      "eval_runtime": 3.0554,
+      "eval_samples_per_second": 474.895,
+      "eval_steps_per_second": 29.783,
       "step": 8160
     },
     {
       "epoch": 10.42,
-      "learning_rate": 6.111111111111112e-06,
-      "loss": 0.0034,
+      "learning_rate": 9.583333333333335e-06,
+      "loss": 0.0046,
       "step": 8500
     },
     {
       "epoch": 11.0,
-      "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,
+      "eval_accuracy": 0.9732382682888292,
+      "eval_f1": 0.9645942795087595,
+      "eval_loss": 0.1706034392118454,
+      "eval_precision": 0.9636862594932669,
+      "eval_recall": 0.9655040122766073,
+      "eval_runtime": 4.2231,
+      "eval_samples_per_second": 343.586,
+      "eval_steps_per_second": 21.548,
       "step": 8976
     },
     {
       "epoch": 11.03,
-      "learning_rate": 5.294117647058824e-06,
-      "loss": 0.0034,
+      "learning_rate": 8.970588235294119e-06,
+      "loss": 0.0051,
       "step": 9000
     },
     {
       "epoch": 11.64,
-      "learning_rate": 4.477124183006537e-06,
-      "loss": 0.0025,
+      "learning_rate": 8.357843137254903e-06,
+      "loss": 0.0032,
       "step": 9500
     },
     {
       "epoch": 12.0,
-      "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,
+      "eval_accuracy": 0.973475272554906,
+      "eval_f1": 0.9657750934713834,
+      "eval_loss": 0.1721695363521576,
+      "eval_precision": 0.9653432139777046,
+      "eval_recall": 0.9662073595703188,
+      "eval_runtime": 4.1749,
+      "eval_samples_per_second": 347.551,
+      "eval_steps_per_second": 21.797,
       "step": 9792
     },
     {
       "epoch": 12.25,
-      "learning_rate": 3.6601307189542484e-06,
-      "loss": 0.0022,
+      "learning_rate": 7.745098039215687e-06,
+      "loss": 0.0037,
       "step": 10000
     },
     {
       "epoch": 12.87,
-      "learning_rate": 2.843137254901961e-06,
-      "loss": 0.0024,
+      "learning_rate": 7.132352941176472e-06,
+      "loss": 0.0031,
       "step": 10500
     },
     {
       "epoch": 13.0,
-      "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,
+      "eval_accuracy": 0.9734357718438932,
+      "eval_f1": 0.9649974450689831,
+      "eval_loss": 0.18985576927661896,
+      "eval_precision": 0.9639814962513957,
+      "eval_recall": 0.9660155375811247,
+      "eval_runtime": 3.0596,
+      "eval_samples_per_second": 474.243,
+      "eval_steps_per_second": 29.742,
       "step": 10608
     },
     {
       "epoch": 13.48,
-      "learning_rate": 2.0261437908496734e-06,
-      "loss": 0.0014,
+      "learning_rate": 6.519607843137256e-06,
+      "loss": 0.0028,
       "step": 11000
     },
     {
       "epoch": 14.0,
-      "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,
+      "eval_accuracy": 0.9733962711328804,
+      "eval_f1": 0.9652715654952077,
+      "eval_loss": 0.1955101639032364,
+      "eval_precision": 0.9646243734235816,
+      "eval_recall": 0.9659196265865277,
+      "eval_runtime": 3.05,
+      "eval_samples_per_second": 475.739,
+      "eval_steps_per_second": 29.836,
       "step": 11424
     },
     {
       "epoch": 14.09,
-      "learning_rate": 1.2091503267973858e-06,
-      "loss": 0.0019,
+      "learning_rate": 5.90686274509804e-06,
+      "loss": 0.0028,
       "step": 11500
     },
     {
       "epoch": 14.71,
-      "learning_rate": 3.921568627450981e-07,
-      "loss": 0.0016,
+      "learning_rate": 5.294117647058824e-06,
+      "loss": 0.0026,
       "step": 12000
     },
     {
       "epoch": 15.0,
-      "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,
+      "eval_accuracy": 0.9738110285985148,
+      "eval_f1": 0.9658246656760773,
+      "eval_loss": 0.19060379266738892,
+      "eval_precision": 0.9653465346534653,
+      "eval_recall": 0.9663032705649157,
+      "eval_runtime": 4.1019,
+      "eval_samples_per_second": 353.738,
+      "eval_steps_per_second": 22.185,
       "step": 12240
     },
     {
-      "epoch": 15.0,
-      "step": 12240,
-      "total_flos": 5.119363918698906e+16,
-      "train_loss": 0.03262665262993644,
-      "train_runtime": 785.1159,
-      "train_samples_per_second": 249.403,
-      "train_steps_per_second": 15.59
+      "epoch": 15.32,
+      "learning_rate": 4.681372549019608e-06,
+      "loss": 0.0019,
+      "step": 12500
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 4.068627450980392e-06,
+      "loss": 0.0021,
+      "step": 13000
+    },
+    {
+      "epoch": 16.0,
+      "eval_accuracy": 0.973969031442566,
+      "eval_f1": 0.9657132819112408,
+      "eval_loss": 0.2030537873506546,
+      "eval_precision": 0.964773452456924,
+      "eval_recall": 0.9666549442117714,
+      "eval_runtime": 4.0654,
+      "eval_samples_per_second": 356.916,
+      "eval_steps_per_second": 22.384,
+      "step": 13056
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 3.4558823529411766e-06,
+      "loss": 0.0017,
+      "step": 13500
+    },
+    {
+      "epoch": 17.0,
+      "eval_accuracy": 0.9744035392637067,
+      "eval_f1": 0.9668322128359497,
+      "eval_loss": 0.1976209431886673,
+      "eval_precision": 0.9658605066683683,
+      "eval_recall": 0.9678058761469357,
+      "eval_runtime": 4.0709,
+      "eval_samples_per_second": 356.434,
+      "eval_steps_per_second": 22.354,
+      "step": 13872
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 2.843137254901961e-06,
+      "loss": 0.0015,
+      "step": 14000
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 2.2303921568627456e-06,
+      "loss": 0.0015,
+      "step": 14500
+    },
+    {
+      "epoch": 18.0,
+      "eval_accuracy": 0.9746997945963027,
+      "eval_f1": 0.9669350201265093,
+      "eval_loss": 0.19767752289772034,
+      "eval_precision": 0.9662250598563448,
+      "eval_recall": 0.967646024489274,
+      "eval_runtime": 4.1567,
+      "eval_samples_per_second": 349.074,
+      "eval_steps_per_second": 21.892,
+      "step": 14688
     }
   ],
-  "max_steps": 12240,
-  "num_train_epochs": 15,
-  "total_flos": 5.119363918698906e+16,
+  "max_steps": 16320,
+  "num_train_epochs": 20,
+  "total_flos": 6.14323669958656e+16,
   "trial_name": null,
   "trial_params": null
 }

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f2ad2280933fd2dfde8ee13fc62c6b195937dddc4b2505fce56231e549d3c475
-size 3119
+oid sha256:c3f5d9fbb7dcb238c7722f2a11d35ccddcf96b37c8ed9a1d879458d25efd5245
+size 3183