Training in progress, epoch 26, checkpoint

checkpoint-2106/model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:19ac5aeb8b617ba18ec1bb50c72c88798748f0dcdafd06c7b4d07f00d3364f88
+oid sha256:7490d0b0c38cd15ecc6206da45f89e403e4852329a43fcfc31273523fefadcce
 size 592318676

checkpoint-2106/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a688728a8a1d0c2891ff8497ba966bc89c7612354625d010909738446ac209de
+oid sha256:3c0df30e17aa589baa0cee08737872827d404e8be93fa3504f2812c67dd3ffde
 size 1014645498

checkpoint-2106/rng_state.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:164e69b2c55b65211871f7f6a24d516925a9888b1d257795ff838d2a6eb1c882
+oid sha256:26e1215644010fc2b2d486647bdaafbb5b5d68adcf85fb4c517a050afeb97e7f
 size 14244

checkpoint-2106/trainer_state.json

@@ -11,1042 +11,1042 @@
     {
       "epoch": 1.0,
       "eval_B": {
-        "f1-score": 0.7905866302864938,
-        "precision": 0.7768096514745308,
-        "recall": 0.8048611111111111,
-        "support": 1440.0
+        "f1-score": 0.8331822302810515,
+        "precision": 0.7901977644024075,
+        "recall": 0.8811121764141898,
+        "support": 1043.0
       },
       "eval_I": {
-        "f1-score": 0.9005331227169513,
-        "precision": 0.9637593111099372,
-        "recall": 0.8450919534905267,
-        "support": 21587.0
+        "f1-score": 0.9471294753175105,
+        "precision": 0.9443616777446711,
+        "recall": 0.9499135446685879,
+        "support": 17350.0
       },
       "eval_O": {
-        "f1-score": 0.8288892663043479,
-        "precision": 0.7463108800367001,
-        "recall": 0.9320156593144275,
-        "support": 10473.0
+        "f1-score": 0.8992868897421833,
+        "precision": 0.9103731674811195,
+        "recall": 0.8884673748103187,
+        "support": 9226.0
       },
-      "eval_accuracy": 0.8705373134328358,
-      "eval_loss": 0.32930415868759155,
+      "eval_accuracy": 0.9267895289474637,
+      "eval_loss": 0.20478281378746033,
       "eval_macro avg": {
-        "f1-score": 0.8400030064359308,
-        "precision": 0.8289599475403894,
-        "recall": 0.8606562413053552,
-        "support": 33500.0
-      },
-      "eval_runtime": 4.7349,
-      "eval_samples_per_second": 17.107,
-      "eval_steps_per_second": 2.323,
+        "f1-score": 0.8931995317802484,
+        "precision": 0.8816442032093993,
+        "recall": 0.9064976986310321,
+        "support": 27619.0
+      },
+      "eval_runtime": 4.6128,
+      "eval_samples_per_second": 17.343,
+      "eval_steps_per_second": 2.168,
       "eval_weighted avg": {
-        "f1-score": 0.8734092702599646,
-        "precision": 0.8877430445874,
-        "recall": 0.8705373134328358,
-        "support": 33500.0
+        "f1-score": 0.9268447919078652,
+        "precision": 0.9271861479533134,
+        "recall": 0.9267895289474637,
+        "support": 27619.0
       },
       "step": 81
     },
     {
       "epoch": 2.0,
       "eval_B": {
-        "f1-score": 0.8656218739579861,
-        "precision": 0.8325849903784477,
-        "recall": 0.9013888888888889,
-        "support": 1440.0
+        "f1-score": 0.8682547399124939,
+        "precision": 0.8806706114398422,
+        "recall": 0.8561840843720039,
+        "support": 1043.0
       },
       "eval_I": {
-        "f1-score": 0.9547168948727912,
-        "precision": 0.9423736462093862,
-        "recall": 0.9673877796822161,
-        "support": 21587.0
+        "f1-score": 0.9590983771942466,
+        "precision": 0.9592919333448654,
+        "recall": 0.9589048991354466,
+        "support": 17350.0
       },
       "eval_O": {
-        "f1-score": 0.9013528191962081,
-        "precision": 0.9332379102341274,
-        "recall": 0.8715745249689678,
-        "support": 10473.0
+        "f1-score": 0.9226525313717006,
+        "precision": 0.9208594256100194,
+        "recall": 0.9244526338608281,
+        "support": 9226.0
       },
-      "eval_accuracy": 0.9345970149253732,
-      "eval_loss": 0.2162298709154129,
+      "eval_accuracy": 0.9435171439950758,
+      "eval_loss": 0.16574698686599731,
       "eval_macro avg": {
-        "f1-score": 0.9072305293423284,
-        "precision": 0.902732182273987,
-        "recall": 0.9134503978466909,
-        "support": 33500.0
-      },
-      "eval_runtime": 4.753,
-      "eval_samples_per_second": 17.042,
-      "eval_steps_per_second": 2.314,
+        "f1-score": 0.9166685494928136,
+        "precision": 0.9202739901315756,
+        "recall": 0.9131805391227594,
+        "support": 27619.0
+      },
+      "eval_runtime": 4.6238,
+      "eval_samples_per_second": 17.302,
+      "eval_steps_per_second": 2.163,
       "eval_weighted avg": {
-        "f1-score": 0.9342040950316517,
-        "precision": 0.9347982961417612,
-        "recall": 0.9345970149253732,
-        "support": 33500.0
+        "f1-score": 0.9434932036816766,
+        "precision": 0.9434846863370583,
+        "recall": 0.9435171439950758,
+        "support": 27619.0
       },
       "step": 162
     },
     {
       "epoch": 3.0,
       "eval_B": {
-        "f1-score": 0.8603587759409076,
-        "precision": 0.8717034925160371,
-        "recall": 0.8493055555555555,
-        "support": 1440.0
+        "f1-score": 0.8791725434884815,
+        "precision": 0.8625461254612546,
+        "recall": 0.8964525407478428,
+        "support": 1043.0
       },
       "eval_I": {
-        "f1-score": 0.9434195218082959,
-        "precision": 0.9677057963955188,
-        "recall": 0.9203224162690509,
-        "support": 21587.0
+        "f1-score": 0.9588372159825352,
+        "precision": 0.9557349825345015,
+        "recall": 0.9619596541786744,
+        "support": 17350.0
       },
       "eval_O": {
-        "f1-score": 0.8879310344827586,
-        "precision": 0.8459410391631365,
-        "recall": 0.9343072663038289,
-        "support": 10473.0
+        "f1-score": 0.9204284621270084,
+        "precision": 0.9282407407407407,
+        "recall": 0.9127465857359636,
+        "support": 9226.0
       },
-      "eval_accuracy": 0.9216417910447762,
-      "eval_loss": 0.24082203209400177,
+      "eval_accuracy": 0.9430464535283681,
+      "eval_loss": 0.15394224226474762,
       "eval_macro avg": {
-        "f1-score": 0.8972364440773206,
-        "precision": 0.8951167760248975,
-        "recall": 0.9013117460428117,
-        "support": 33500.0
-      },
-      "eval_runtime": 4.7554,
-      "eval_samples_per_second": 17.033,
-      "eval_steps_per_second": 2.313,
+        "f1-score": 0.9194794071993417,
+        "precision": 0.9155072829121655,
+        "recall": 0.9237195935541602,
+        "support": 27619.0
+      },
+      "eval_runtime": 4.6233,
+      "eval_samples_per_second": 17.303,
+      "eval_steps_per_second": 2.163,
       "eval_weighted avg": {
-        "f1-score": 0.9225019575751797,
-        "precision": 0.9255121957960801,
-        "recall": 0.9216417910447762,
-        "support": 33500.0
+        "f1-score": 0.9429985029052194,
+        "precision": 0.9430314866542512,
+        "recall": 0.9430464535283681,
+        "support": 27619.0
       },
       "step": 243
     },
     {
       "epoch": 4.0,
       "eval_B": {
-        "f1-score": 0.8804020100502512,
-        "precision": 0.8504854368932039,
-        "recall": 0.9125,
-        "support": 1440.0
+        "f1-score": 0.8800746616892207,
+        "precision": 0.8572727272727273,
+        "recall": 0.9041227229146692,
+        "support": 1043.0
       },
       "eval_I": {
-        "f1-score": 0.9589184992350025,
-        "precision": 0.945595388218339,
-        "recall": 0.9726224116366332,
-        "support": 21587.0
+        "f1-score": 0.9537844036697248,
+        "precision": 0.9488876212207644,
+        "recall": 0.9587319884726225,
+        "support": 17350.0
       },
       "eval_O": {
-        "f1-score": 0.9091178797468354,
-        "precision": 0.9427750999897446,
-        "recall": 0.877780960565263,
-        "support": 10473.0
+        "f1-score": 0.9100192149327477,
+        "precision": 0.9220157970853265,
+        "recall": 0.8983308042488619,
+        "support": 9226.0
       },
-      "eval_accuracy": 0.9403880597014925,
-      "eval_loss": 0.22083505988121033,
+      "eval_accuracy": 0.9364929939534379,
+      "eval_loss": 0.19801080226898193,
       "eval_macro avg": {
-        "f1-score": 0.9161461296773631,
-        "precision": 0.9129519750337626,
-        "recall": 0.9209677907339654,
-        "support": 33500.0
-      },
-      "eval_runtime": 4.7449,
-      "eval_samples_per_second": 17.071,
-      "eval_steps_per_second": 2.318,
+        "f1-score": 0.9146260934305644,
+        "precision": 0.9093920485262728,
+        "recall": 0.9203951718787179,
+        "support": 27619.0
+      },
+      "eval_runtime": 4.6132,
+      "eval_samples_per_second": 17.342,
+      "eval_steps_per_second": 2.168,
       "eval_weighted avg": {
-        "f1-score": 0.9399744505088647,
-        "precision": 0.9406253819936745,
-        "recall": 0.9403880597014925,
-        "support": 33500.0
+        "f1-score": 0.9363812792925563,
+        "precision": 0.9364514800186444,
+        "recall": 0.9364929939534379,
+        "support": 27619.0
       },
       "step": 324
     },
     {
       "epoch": 5.0,
       "eval_B": {
-        "f1-score": 0.8879781420765027,
-        "precision": 0.8736559139784946,
-        "recall": 0.9027777777777778,
-        "support": 1440.0
+        "f1-score": 0.883877159309021,
+        "precision": 0.8847262247838616,
+        "recall": 0.8830297219558965,
+        "support": 1043.0
       },
       "eval_I": {
-        "f1-score": 0.9613201259451634,
-        "precision": 0.9539317642765919,
-        "recall": 0.9688238291564367,
-        "support": 21587.0
+        "f1-score": 0.9591500482927107,
+        "precision": 0.9456643513331839,
+        "recall": 0.9730259365994236,
+        "support": 17350.0
       },
       "eval_O": {
-        "f1-score": 0.916006030835076,
-        "precision": 0.9334853291038858,
-        "recall": 0.899169292466342,
-        "support": 10473.0
+        "f1-score": 0.9191176470588236,
+        "precision": 0.9454503781801513,
+        "recall": 0.8942120095382614,
+        "support": 9226.0
       },
-      "eval_accuracy": 0.9442089552238806,
-      "eval_loss": 0.2257838249206543,
+      "eval_accuracy": 0.9432999022412107,
+      "eval_loss": 0.20074255764484406,
       "eval_macro avg": {
-        "f1-score": 0.921768099618914,
-        "precision": 0.9203576691196576,
-        "recall": 0.9235902998001855,
-        "support": 33500.0
-      },
-      "eval_runtime": 4.7845,
-      "eval_samples_per_second": 16.93,
-      "eval_steps_per_second": 2.299,
+        "f1-score": 0.9207149515535183,
+        "precision": 0.9252803180990656,
+        "recall": 0.9167558893645271,
+        "support": 27619.0
+      },
+      "eval_runtime": 4.6092,
+      "eval_samples_per_second": 17.356,
+      "eval_steps_per_second": 2.17,
       "eval_weighted avg": {
-        "f1-score": 0.9440011117702735,
-        "precision": 0.9440889959185916,
-        "recall": 0.9442089552238806,
-        "support": 33500.0
+        "f1-score": 0.9429348139614955,
+        "precision": 0.9432916158141273,
+        "recall": 0.9432999022412107,
+        "support": 27619.0
       },
       "step": 405
     },
     {
       "epoch": 6.0,
       "eval_B": {
-        "f1-score": 0.8873579056148813,
-        "precision": 0.8803827751196173,
-        "recall": 0.8944444444444445,
-        "support": 1440.0
+        "f1-score": 0.8897338403041826,
+        "precision": 0.882186616399623,
+        "recall": 0.8974113135186961,
+        "support": 1043.0
       },
       "eval_I": {
-        "f1-score": 0.9622060984570168,
-        "precision": 0.9539267015706806,
-        "recall": 0.9706304720433594,
-        "support": 21587.0
+        "f1-score": 0.9624160683661308,
+        "precision": 0.9707969272268809,
+        "recall": 0.954178674351585,
+        "support": 17350.0
       },
       "eval_O": {
-        "f1-score": 0.9182769530299342,
-        "precision": 0.9365567911040509,
-        "recall": 0.9006970304592762,
-        "support": 10473.0
+        "f1-score": 0.9297955261331482,
+        "precision": 0.9161493950552341,
+        "recall": 0.9438543247344461,
+        "support": 9226.0
       },
-      "eval_accuracy": 0.9454925373134329,
-      "eval_loss": 0.2242102175951004,
+      "eval_accuracy": 0.9485861182519281,
+      "eval_loss": 0.16565027832984924,
       "eval_macro avg": {
-        "f1-score": 0.9226136523672773,
-        "precision": 0.923622089264783,
-        "recall": 0.9219239823156933,
-        "support": 33500.0
-      },
-      "eval_runtime": 4.7715,
-      "eval_samples_per_second": 16.976,
-      "eval_steps_per_second": 2.305,
+        "f1-score": 0.9273151449344872,
+        "precision": 0.9230443128939126,
+        "recall": 0.9318147708682424,
+        "support": 27619.0
+      },
+      "eval_runtime": 4.6219,
+      "eval_samples_per_second": 17.309,
+      "eval_steps_per_second": 2.164,
       "eval_weighted avg": {
-        "f1-score": 0.945255312255509,
-        "precision": 0.9453351097376493,
-        "recall": 0.9454925373134329,
-        "support": 33500.0
+        "f1-score": 0.9487745648174827,
+        "precision": 0.9491959030765336,
+        "recall": 0.9485861182519281,
+        "support": 27619.0
       },
       "step": 486
     },
     {
       "epoch": 6.17,
-      "grad_norm": 1.8705081939697266,
+      "grad_norm": 4.767655849456787,
       "learning_rate": 1.7530864197530865e-05,
-      "loss": 0.1601,
+      "loss": 0.1585,
       "step": 500
     },
     {
       "epoch": 7.0,
       "eval_B": {
-        "f1-score": 0.8876749743939911,
-        "precision": 0.8730691739422431,
-        "recall": 0.9027777777777778,
-        "support": 1440.0
+        "f1-score": 0.8802588996763754,
+        "precision": 0.85,
+        "recall": 0.912751677852349,
+        "support": 1043.0
       },
       "eval_I": {
-        "f1-score": 0.9611415113943341,
-        "precision": 0.9581108451482232,
-        "recall": 0.9641914114976606,
-        "support": 21587.0
+        "f1-score": 0.958423523458803,
+        "precision": 0.9354661691269275,
+        "recall": 0.9825360230547551,
+        "support": 17350.0
       },
       "eval_O": {
-        "f1-score": 0.9164739884393064,
-        "precision": 0.9247594050743657,
-        "recall": 0.9083357204239473,
-        "support": 10473.0
+        "f1-score": 0.9139526911210146,
+        "precision": 0.9664088931851136,
+        "recall": 0.8668978972469109,
+        "support": 9226.0
       },
-      "eval_accuracy": 0.944089552238806,
-      "eval_loss": 0.22517083585262299,
+      "eval_accuracy": 0.9412723125384699,
+      "eval_loss": 0.2591131329536438,
       "eval_macro avg": {
-        "f1-score": 0.9217634914092105,
-        "precision": 0.9186464747216107,
-        "recall": 0.9251016365664619,
-        "support": 33500.0
-      },
-      "eval_runtime": 4.7926,
-      "eval_samples_per_second": 16.901,
-      "eval_steps_per_second": 2.295,
+        "f1-score": 0.9175450380853977,
+        "precision": 0.917291687437347,
+        "recall": 0.9207285327180049,
+        "support": 27619.0
+      },
+      "eval_runtime": 4.617,
+      "eval_samples_per_second": 17.327,
+      "eval_steps_per_second": 2.166,
       "eval_weighted avg": {
-        "f1-score": 0.9440192791200506,
-        "precision": 0.9440287663891153,
-        "recall": 0.944089552238806,
-        "support": 33500.0
+        "f1-score": 0.9406164485555297,
+        "precision": 0.9425749115781907,
+        "recall": 0.9412723125384699,
+        "support": 27619.0
       },
       "step": 567
     },
     {
       "epoch": 8.0,
       "eval_B": {
-        "f1-score": 0.894540522210919,
-        "precision": 0.8740888005301524,
-        "recall": 0.9159722222222222,
-        "support": 1440.0
+        "f1-score": 0.8687910028116214,
+        "precision": 0.84967919340055,
+        "recall": 0.8887823585810163,
+        "support": 1043.0
       },
       "eval_I": {
-        "f1-score": 0.9638018385639943,
-        "precision": 0.9540255968049378,
-        "recall": 0.9737805160513272,
-        "support": 21587.0
+        "f1-score": 0.9595010203788336,
+        "precision": 0.956997878562009,
+        "recall": 0.9620172910662824,
+        "support": 17350.0
       },
       "eval_O": {
-        "f1-score": 0.9203132648066569,
-        "precision": 0.9441598875163202,
-        "recall": 0.8976415544734078,
-        "support": 10473.0
+        "f1-score": 0.9212035166275324,
+        "precision": 0.9282491471332673,
+        "recall": 0.9142640364188164,
+        "support": 9226.0
       },
-      "eval_accuracy": 0.9474925373134329,
-      "eval_loss": 0.24869142472743988,
+      "eval_accuracy": 0.9432999022412107,
+      "eval_loss": 0.2693409323692322,
       "eval_macro avg": {
-        "f1-score": 0.9262185418605234,
-        "precision": 0.9240914282838034,
-        "recall": 0.9291314309156524,
-        "support": 33500.0
-      },
-      "eval_runtime": 4.7743,
-      "eval_samples_per_second": 16.966,
-      "eval_steps_per_second": 2.304,
+        "f1-score": 0.9164985132726624,
+        "precision": 0.9116420730319422,
+        "recall": 0.9216878953553717,
+        "support": 27619.0
+      },
+      "eval_runtime": 4.6153,
+      "eval_samples_per_second": 17.333,
+      "eval_steps_per_second": 2.167,
       "eval_weighted avg": {
-        "f1-score": 0.947228939205516,
-        "precision": 0.9475052218791054,
-        "recall": 0.9474925373134329,
-        "support": 33500.0
+        "f1-score": 0.9432823550422136,
+        "precision": 0.9433417293609165,
+        "recall": 0.9432999022412107,
+        "support": 27619.0
       },
       "step": 648
     },
     {
       "epoch": 9.0,
       "eval_B": {
-        "f1-score": 0.8851236868858015,
-        "precision": 0.8643282594308405,
-        "recall": 0.9069444444444444,
-        "support": 1440.0
+        "f1-score": 0.8645343367826904,
+        "precision": 0.8485687903970452,
+        "recall": 0.8811121764141898,
+        "support": 1043.0
       },
       "eval_I": {
-        "f1-score": 0.9617001723147618,
-        "precision": 0.9540067462849849,
-        "recall": 0.9695186918052532,
-        "support": 21587.0
+        "f1-score": 0.9567003348057282,
+        "precision": 0.934535260814599,
+        "recall": 0.9799423631123919,
+        "support": 17350.0
       },
       "eval_O": {
-        "f1-score": 0.9165854609237966,
-        "precision": 0.9358272808675754,
-        "recall": 0.8981189725961998,
-        "support": 10473.0
+        "f1-score": 0.9104673003585861,
+        "precision": 0.9586479683567062,
+        "recall": 0.8668978972469109,
+        "support": 9226.0
       },
-      "eval_accuracy": 0.9445074626865672,
-      "eval_loss": 0.3045330345630646,
+      "eval_accuracy": 0.9384481697382236,
+      "eval_loss": 0.3087053894996643,
       "eval_macro avg": {
-        "f1-score": 0.9211364400414532,
-        "precision": 0.9180540955278002,
-        "recall": 0.9248607029486324,
-        "support": 33500.0
-      },
-      "eval_runtime": 4.7599,
-      "eval_samples_per_second": 17.017,
-      "eval_steps_per_second": 2.311,
+        "f1-score": 0.910567323982335,
+        "precision": 0.9139173398561168,
+        "recall": 0.9093174789244975,
+        "support": 27619.0
+      },
+      "eval_runtime": 4.6156,
+      "eval_samples_per_second": 17.333,
+      "eval_steps_per_second": 2.167,
       "eval_weighted avg": {
-        "f1-score": 0.9443044555560967,
-        "precision": 0.9444685205421044,
-        "recall": 0.9445074626865672,
-        "support": 33500.0
+        "f1-score": 0.9377758584761231,
+        "precision": 0.9393435743356523,
+        "recall": 0.9384481697382236,
+        "support": 27619.0
       },
       "step": 729
     },
     {
       "epoch": 10.0,
       "eval_B": {
-        "f1-score": 0.8815068493150685,
-        "precision": 0.8695945945945946,
-        "recall": 0.89375,
-        "support": 1440.0
+        "f1-score": 0.8803126526624329,
+        "precision": 0.897410358565737,
+        "recall": 0.8638542665388304,
+        "support": 1043.0
       },
       "eval_I": {
-        "f1-score": 0.956297959947595,
-        "precision": 0.9660159757999717,
-        "recall": 0.9467735211006625,
-        "support": 21587.0
+        "f1-score": 0.9590983771942466,
+        "precision": 0.9592919333448654,
+        "recall": 0.9589048991354466,
+        "support": 17350.0
       },
       "eval_O": {
-        "f1-score": 0.9101049868766404,
-        "precision": 0.8937678357728068,
-        "recall": 0.9270505108373914,
-        "support": 10473.0
+        "f1-score": 0.9220456265542221,
+        "precision": 0.9197584124245038,
+        "recall": 0.9243442445263386,
+        "support": 9226.0
       },
-      "eval_accuracy": 0.9383283582089552,
-      "eval_loss": 0.32627996802330017,
+      "eval_accuracy": 0.9437705927079184,
+      "eval_loss": 0.2796257436275482,
       "eval_macro avg": {
-        "f1-score": 0.9159699320464346,
-        "precision": 0.909792802055791,
-        "recall": 0.9225246773126846,
-        "support": 33500.0
-      },
-      "eval_runtime": 4.7534,
-      "eval_samples_per_second": 17.04,
-      "eval_steps_per_second": 2.314,
+        "f1-score": 0.9204855521369671,
+        "precision": 0.9254869014450354,
+        "recall": 0.9157011367335386,
+        "support": 27619.0
+      },
+      "eval_runtime": 4.641,
+      "eval_samples_per_second": 17.238,
+      "eval_steps_per_second": 2.155,
       "eval_weighted avg": {
-        "f1-score": 0.9386418940884026,
-        "precision": 0.9392845859659347,
-        "recall": 0.9383283582089552,
-        "support": 33500.0
+        "f1-score": 0.9437458232244597,
+        "precision": 0.9437490553802075,
+        "recall": 0.9437705927079184,
+        "support": 27619.0
       },
       "step": 810
     },
     {
       "epoch": 11.0,
       "eval_B": {
-        "f1-score": 0.8901136755080951,
-        "precision": 0.8831168831168831,
-        "recall": 0.8972222222222223,
-        "support": 1440.0
+        "f1-score": 0.8830917874396135,
+        "precision": 0.8899707887049659,
+        "recall": 0.8763183125599233,
+        "support": 1043.0
       },
       "eval_I": {
-        "f1-score": 0.9620073001078947,
-        "precision": 0.9535359970874671,
-        "recall": 0.9706304720433594,
-        "support": 21587.0
+        "f1-score": 0.9574634882843123,
+        "precision": 0.9647182727751448,
+        "recall": 0.9503170028818444,
+        "support": 17350.0
       },
       "eval_O": {
-        "f1-score": 0.917413322945072,
-        "precision": 0.9361025539103647,
-        "recall": 0.8994557433400172,
-        "support": 10473.0
+        "f1-score": 0.9200619426496501,
+        "precision": 0.9067466582465004,
+        "recall": 0.9337741166269239,
+        "support": 9226.0
       },
-      "eval_accuracy": 0.9452238805970149,
-      "eval_loss": 0.3123624622821808,
+      "eval_accuracy": 0.9419964517180202,
+      "eval_loss": 0.3198166787624359,
       "eval_macro avg": {
-        "f1-score": 0.9231780995203539,
-        "precision": 0.9242518113715716,
-        "recall": 0.922436145868533,
-        "support": 33500.0
-      },
-      "eval_runtime": 4.7539,
-      "eval_samples_per_second": 17.039,
-      "eval_steps_per_second": 2.314,
+        "f1-score": 0.9202057394578587,
+        "precision": 0.9204785732422037,
+        "recall": 0.9201364773562305,
+        "support": 27619.0
+      },
+      "eval_runtime": 4.6345,
+      "eval_samples_per_second": 17.262,
+      "eval_steps_per_second": 2.158,
       "eval_weighted avg": {
-        "f1-score": 0.9449756719810304,
-        "precision": 0.9450588635199616,
-        "recall": 0.9452238805970149,
-        "support": 33500.0
+        "f1-score": 0.942161111514465,
+        "precision": 0.9425303680165918,
+        "recall": 0.9419964517180202,
+        "support": 27619.0
       },
       "step": 891
     },
     {
       "epoch": 12.0,
       "eval_B": {
-        "f1-score": 0.8800819952169457,
-        "precision": 0.8661735036987223,
-        "recall": 0.8944444444444445,
-        "support": 1440.0
+        "f1-score": 0.871306005719733,
+        "precision": 0.8663507109004739,
+        "recall": 0.8763183125599233,
+        "support": 1043.0
       },
       "eval_I": {
-        "f1-score": 0.9595463311598861,
-        "precision": 0.9512039692293687,
-        "recall": 0.9680363181544448,
-        "support": 21587.0
+        "f1-score": 0.9576855173984852,
+        "precision": 0.9464205312922107,
+        "recall": 0.9692219020172911,
+        "support": 17350.0
       },
       "eval_O": {
-        "f1-score": 0.9115367743822197,
-        "precision": 0.9310035842293907,
-        "recall": 0.8928673732454884,
-        "support": 10473.0
+        "f1-score": 0.9162135168127844,
+        "precision": 0.9386084583901774,
+        "recall": 0.8948623455451984,
+        "support": 9226.0
       },
-      "eval_accuracy": 0.9413731343283582,
-      "eval_loss": 0.3427491784095764,
+      "eval_accuracy": 0.9408740359897172,
+      "eval_loss": 0.35264310240745544,
       "eval_macro avg": {
-        "f1-score": 0.9170550335863505,
-        "precision": 0.9161270190524938,
-        "recall": 0.9184493786147926,
-        "support": 33500.0
-      },
-      "eval_runtime": 4.7455,
-      "eval_samples_per_second": 17.069,
-      "eval_steps_per_second": 2.318,
+        "f1-score": 0.9150683466436677,
+        "precision": 0.917126566860954,
+        "recall": 0.9134675200408043,
+        "support": 27619.0
+      },
+      "eval_runtime": 4.6326,
+      "eval_samples_per_second": 17.269,
+      "eval_steps_per_second": 2.159,
       "eval_weighted avg": {
-        "f1-score": 0.9411214734915179,
-        "precision": 0.9412337452750731,
-        "recall": 0.9413731343283582,
-        "support": 33500.0
+        "f1-score": 0.9405699625961891,
+        "precision": 0.9407871989028143,
+        "recall": 0.9408740359897172,
+        "support": 27619.0
       },
       "step": 972
     },
     {
       "epoch": 12.35,
-      "grad_norm": 0.47551417350769043,
+      "grad_norm": 0.3602512776851654,
       "learning_rate": 1.506172839506173e-05,
-      "loss": 0.0222,
+      "loss": 0.021,
       "step": 1000
     },
     {
       "epoch": 13.0,
       "eval_B": {
-        "f1-score": 0.8896434634974534,
-        "precision": 0.8704318936877077,
-        "recall": 0.9097222222222222,
-        "support": 1440.0
+        "f1-score": 0.877092300334768,
+        "precision": 0.875,
+        "recall": 0.8791946308724832,
+        "support": 1043.0
       },
       "eval_I": {
-        "f1-score": 0.9620136667970457,
-        "precision": 0.9556591698665204,
-        "recall": 0.9684532357437347,
-        "support": 21587.0
+        "f1-score": 0.9597242793665263,
+        "precision": 0.9486991778353419,
+        "recall": 0.9710086455331413,
+        "support": 17350.0
       },
       "eval_O": {
-        "f1-score": 0.9183177933177933,
-        "precision": 0.9343808676746714,
-        "recall": 0.9027976701995608,
-        "support": 10473.0
+        "f1-score": 0.9203392649259936,
+        "precision": 0.9419040054464994,
+        "recall": 0.8997398655972252,
+        "support": 9226.0
       },
-      "eval_accuracy": 0.9454029850746268,
-      "eval_loss": 0.3816670775413513,
+      "eval_accuracy": 0.943734385748941,
+      "eval_loss": 0.3594338297843933,
       "eval_macro avg": {
-        "f1-score": 0.9233249745374308,
-        "precision": 0.9201573104096331,
-        "recall": 0.9269910427218392,
-        "support": 33500.0
-      },
-      "eval_runtime": 4.7247,
-      "eval_samples_per_second": 17.144,
-      "eval_steps_per_second": 2.328,
+        "f1-score": 0.9190519482090961,
+        "precision": 0.9218677277606138,
+        "recall": 0.9166477140009498,
+        "support": 27619.0
+      },
+      "eval_runtime": 4.6042,
+      "eval_samples_per_second": 17.376,
+      "eval_steps_per_second": 2.172,
       "eval_weighted avg": {
-        "f1-score": 0.945242324238848,
-        "precision": 0.9453435001186778,
-        "recall": 0.9454029850746268,
-        "support": 33500.0
+        "f1-score": 0.943447393984779,
+        "precision": 0.9436461164304495,
+        "recall": 0.943734385748941,
+        "support": 27619.0
       },
       "step": 1053
     },
     {
       "epoch": 14.0,
       "eval_B": {
-        "f1-score": 0.8865272540281112,
-        "precision": 0.8754231550440081,
-        "recall": 0.8979166666666667,
-        "support": 1440.0
+        "f1-score": 0.8754052802223252,
+        "precision": 0.8467741935483871,
+        "recall": 0.9060402684563759,
+        "support": 1043.0
       },
       "eval_I": {
-        "f1-score": 0.9599206715247671,
-        "precision": 0.9557331129172981,
-        "recall": 0.9641450873210728,
-        "support": 21587.0
+        "f1-score": 0.9574973153224439,
+        "precision": 0.9392880904856953,
+        "recall": 0.9764265129682997,
+        "support": 17350.0
       },
       "eval_O": {
-        "f1-score": 0.9141367826632559,
-        "precision": 0.9242631270739801,
-        "recall": 0.9042299245679366,
-        "support": 10473.0
+        "f1-score": 0.9130164471825015,
+        "precision": 0.9539388213062477,
+        "recall": 0.8754606546715803,
+        "support": 9226.0
       },
-      "eval_accuracy": 0.9425671641791045,
-      "eval_loss": 0.3827010989189148,
+      "eval_accuracy": 0.9400412759332344,
+      "eval_loss": 0.3840242028236389,
       "eval_macro avg": {
-        "f1-score": 0.9201949027387114,
-        "precision": 0.9184731316784287,
-        "recall": 0.9220972261852255,
-        "support": 33500.0
-      },
-      "eval_runtime": 4.7319,
-      "eval_samples_per_second": 17.118,
-      "eval_steps_per_second": 2.325,
+        "f1-score": 0.9153063475757568,
+        "precision": 0.9133337017801101,
+        "recall": 0.9193091453654186,
+        "support": 27619.0
+      },
+      "eval_runtime": 4.6182,
+      "eval_samples_per_second": 17.323,
+      "eval_steps_per_second": 2.165,
       "eval_weighted avg": {
-        "f1-score": 0.9424525763235196,
-        "precision": 0.9424426203479068,
-        "recall": 0.9425671641791045,
-        "support": 33500.0
+        "f1-score": 0.9395385738014427,
+        "precision": 0.9406884180878825,
+        "recall": 0.9400412759332344,
+        "support": 27619.0
       },
       "step": 1134
     },
     {
       "epoch": 15.0,
       "eval_B": {
-        "f1-score": 0.8871919472405415,
-        "precision": 0.8868841082581541,
-        "recall": 0.8875,
-        "support": 1440.0
+        "f1-score": 0.8797709923664121,
+        "precision": 0.8755935422602089,
+        "recall": 0.8839884947267498,
+        "support": 1043.0
       },
       "eval_I": {
-        "f1-score": 0.9616856471022177,
-        "precision": 0.9571841585975862,
-        "recall": 0.9662296752675221,
-        "support": 21587.0
+        "f1-score": 0.9603688219460511,
+        "precision": 0.954302299112224,
+        "recall": 0.9665129682997118,
+        "support": 17350.0
       },
       "eval_O": {
-        "f1-score": 0.9181813798756087,
-        "precision": 0.9273470977795092,
-        "recall": 0.9091950730449728,
-        "support": 10473.0
+        "f1-score": 0.922722283205269,
+        "precision": 0.9346230820547031,
+        "recall": 0.9111207457186213,
+        "support": 9226.0
       },
-      "eval_accuracy": 0.9450149253731344,
-      "eval_loss": 0.39016807079315186,
+      "eval_accuracy": 0.9448930084362215,
+      "eval_loss": 0.3762665390968323,
       "eval_macro avg": {
-        "f1-score": 0.9223529914061226,
-        "precision": 0.9238051215450831,
-        "recall": 0.920974916104165,
-        "support": 33500.0
-      },
-      "eval_runtime": 4.7376,
-      "eval_samples_per_second": 17.097,
-      "eval_steps_per_second": 2.322,
+        "f1-score": 0.9209540325059108,
+        "precision": 0.9215063078090453,
+        "recall": 0.920540736248361,
+        "support": 27619.0
+      },
+      "eval_runtime": 4.586,
+      "eval_samples_per_second": 17.444,
+      "eval_steps_per_second": 2.181,
       "eval_weighted avg": {
-        "f1-score": 0.944882927148036,
-        "precision": 0.9448344388830637,
-        "recall": 0.9450149253731344,
-        "support": 33500.0
+        "f1-score": 0.944749483712443,
+        "precision": 0.9447562007752335,
+        "recall": 0.9448930084362215,
+        "support": 27619.0
       },
       "step": 1215
     },
     {
       "epoch": 16.0,
       "eval_B": {
-        "f1-score": 0.8897338403041825,
-        "precision": 0.8857536132140399,
-        "recall": 0.89375,
-        "support": 1440.0
+        "f1-score": 0.8858669256920835,
+        "precision": 0.8976377952755905,
+        "recall": 0.8744007670182167,
+        "support": 1043.0
       },
       "eval_I": {
-        "f1-score": 0.9608010520729992,
-        "precision": 0.9570693142121713,
-        "recall": 0.9645620049103627,
-        "support": 21587.0
+        "f1-score": 0.9586212929752427,
+        "precision": 0.9701351590627397,
+        "recall": 0.9473775216138328,
+        "support": 17350.0
       },
       "eval_O": {
-        "f1-score": 0.9159121556540166,
-        "precision": 0.9240112719852298,
-        "recall": 0.9079537859257137,
-        "support": 10473.0
+        "f1-score": 0.9235412474849094,
+        "precision": 0.9027950310559006,
+        "recall": 0.9452633860828095,
+        "support": 9226.0
       },
-      "eval_accuracy": 0.943820895522388,
-      "eval_loss": 0.40087994933128357,
+      "eval_accuracy": 0.9439154205438285,
+      "eval_loss": 0.36898088455200195,
       "eval_macro avg": {
-        "f1-score": 0.9221490160103993,
-        "precision": 0.9222780664704805,
-        "recall": 0.9220885969453588,
-        "support": 33500.0
-      },
-      "eval_runtime": 4.8145,
-      "eval_samples_per_second": 16.824,
-      "eval_steps_per_second": 2.285,
+        "f1-score": 0.9226764887174118,
+        "precision": 0.923522661798077,
+        "recall": 0.922347224904953,
+        "support": 27619.0
+      },
+      "eval_runtime": 4.6024,
+      "eval_samples_per_second": 17.382,
+      "eval_steps_per_second": 2.173,
       "eval_weighted avg": {
-        "f1-score": 0.9437127476806678,
-        "precision": 0.9436689713560499,
-        "recall": 0.943820895522388,
-        "support": 33500.0
+        "f1-score": 0.9441554794131966,
+        "precision": 0.9449027186622512,
+        "recall": 0.9439154205438285,
+        "support": 27619.0
       },
       "step": 1296
     },
     {
       "epoch": 17.0,
       "eval_B": {
-        "f1-score": 0.8904347826086957,
-        "precision": 0.89198606271777,
-        "recall": 0.8888888888888888,
-        "support": 1440.0
+        "f1-score": 0.8863309352517986,
+        "precision": 0.8867562380038387,
+        "recall": 0.8859060402684564,
+        "support": 1043.0
       },
       "eval_I": {
-        "f1-score": 0.9615171394252979,
-        "precision": 0.9589034784611841,
-        "recall": 0.9641450873210728,
-        "support": 21587.0
+        "f1-score": 0.9630614486386496,
+        "precision": 0.9659090909090909,
+        "recall": 0.9602305475504322,
+        "support": 17350.0
       },
       "eval_O": {
-        "f1-score": 0.9182546920750733,
-        "precision": 0.9232625482625483,
-        "recall": 0.9133008689009835,
-        "support": 10473.0
+        "f1-score": 0.9296685529506872,
+        "precision": 0.9245363918962375,
+        "recall": 0.9348580099718188,
+        "support": 9226.0
       },
-      "eval_accuracy": 0.9450149253731344,
-      "eval_loss": 0.42787015438079834,
+      "eval_accuracy": 0.9489481878417032,
+      "eval_loss": 0.3569599986076355,
       "eval_macro avg": {
-        "f1-score": 0.9234022047030224,
-        "precision": 0.9247173631471673,
-        "recall": 0.9221116150369818,
-        "support": 33500.0
-      },
-      "eval_runtime": 4.7453,
-      "eval_samples_per_second": 17.07,
-      "eval_steps_per_second": 2.318,
+        "f1-score": 0.9263536456137119,
+        "precision": 0.9257339069363891,
+        "recall": 0.9269981992635691,
+        "support": 27619.0
+      },
+      "eval_runtime": 4.6143,
+      "eval_samples_per_second": 17.337,
+      "eval_steps_per_second": 2.167,
       "eval_weighted avg": {
-        "f1-score": 0.9449366556964975,
-        "precision": 0.9448847160539952,
-        "recall": 0.9450149253731344,
-        "support": 33500.0
+        "f1-score": 0.9490090650954502,
+        "precision": 0.9490996138580476,
+        "recall": 0.9489481878417032,
+        "support": 27619.0
       },
       "step": 1377
     },
     {
       "epoch": 18.0,
       "eval_B": {
-        "f1-score": 0.8895027624309393,
-        "precision": 0.8846153846153846,
-        "recall": 0.8944444444444445,
-        "support": 1440.0
+        "f1-score": 0.8778409090909091,
+        "precision": 0.8671655753040225,
+        "recall": 0.8887823585810163,
+        "support": 1043.0
       },
       "eval_I": {
-        "f1-score": 0.9607206210979795,
-        "precision": 0.9555912007332722,
-        "recall": 0.9659054060314078,
-        "support": 21587.0
+        "f1-score": 0.9557028055829692,
+        "precision": 0.9355230471984544,
+        "recall": 0.9767723342939482,
+        "support": 17350.0
       },
       "eval_O": {
-        "f1-score": 0.9159781610861478,
-        "precision": 0.9271322378716745,
-        "recall": 0.905089277188962,
-        "support": 10473.0
+        "f1-score": 0.9098012570069645,
+        "precision": 0.9524599881446354,
+        "recall": 0.8707999132885325,
+        "support": 9226.0
       },
-      "eval_accuracy": 0.943820895522388,
-      "eval_loss": 0.422620564699173,
+      "eval_accuracy": 0.938049893189471,
+      "eval_loss": 0.4151785373687744,
       "eval_macro avg": {
-        "f1-score": 0.9220671815383555,
-        "precision": 0.9224462744067772,
-        "recall": 0.9218130425549381,
-        "support": 33500.0
-      },
-      "eval_runtime": 4.7703,
-      "eval_samples_per_second": 16.98,
-      "eval_steps_per_second": 2.306,
+        "f1-score": 0.9144483238936143,
+        "precision": 0.9183828702157042,
+        "recall": 0.9121182020544989,
+        "support": 27619.0
+      },
+      "eval_runtime": 4.6304,
+      "eval_samples_per_second": 17.277,
+      "eval_steps_per_second": 2.16,
       "eval_weighted avg": {
-        "f1-score": 0.9436716210924735,
-        "precision": 0.9436432636210552,
-        "recall": 0.943820895522388,
-        "support": 33500.0
+        "f1-score": 0.9374292386470396,
+        "precision": 0.9385993125948691,
+        "recall": 0.938049893189471,
+        "support": 27619.0
       },
       "step": 1458
     },
     {
       "epoch": 18.52,
-      "grad_norm": 0.1027584820985794,
+      "grad_norm": 10.955327033996582,
       "learning_rate": 1.2592592592592593e-05,
-      "loss": 0.0054,
+      "loss": 0.0058,
       "step": 1500
     },
     {
       "epoch": 19.0,
       "eval_B": {
-        "f1-score": 0.8924247665167762,
-        "precision": 0.8890420399724328,
-        "recall": 0.8958333333333334,
-        "support": 1440.0
+        "f1-score": 0.8850241545893719,
+        "precision": 0.8919182083739046,
+        "recall": 0.87823585810163,
+        "support": 1043.0
       },
       "eval_I": {
-        "f1-score": 0.9599648245857633,
-        "precision": 0.959121387283237,
-        "recall": 0.9608097466067541,
-        "support": 21587.0
+        "f1-score": 0.9626489459211732,
+        "precision": 0.956838628857761,
+        "recall": 0.9685302593659942,
+        "support": 17350.0
       },
       "eval_O": {
-        "f1-score": 0.9158252380724505,
-        "precision": 0.9179777436684574,
-        "recall": 0.913682803399217,
-        "support": 10473.0
+        "f1-score": 0.927475898334794,
+        "precision": 0.9375415282392027,
+        "recall": 0.9176241057879905,
+        "support": 9226.0
       },
-      "eval_accuracy": 0.9432835820895522,
-      "eval_loss": 0.4375981390476227,
+      "eval_accuracy": 0.9481154277852203,
+      "eval_loss": 0.3719603419303894,
       "eval_macro avg": {
-        "f1-score": 0.9227382763916633,
-        "precision": 0.9220470569747091,
-        "recall": 0.9234419611131015,
-        "support": 33500.0
-      },
-      "eval_runtime": 4.7487,
-      "eval_samples_per_second": 17.057,
-      "eval_steps_per_second": 2.316,
+        "f1-score": 0.9250496662817796,
+        "precision": 0.9287661218236227,
+        "recall": 0.9214634077518715,
+        "support": 27619.0
+      },
+      "eval_runtime": 4.6041,
+      "eval_samples_per_second": 17.376,
+      "eval_steps_per_second": 2.172,
       "eval_weighted avg": {
-        "f1-score": 0.9432623895656658,
-        "precision": 0.9432464129636805,
-        "recall": 0.9432835820895522,
-        "support": 33500.0
+        "f1-score": 0.9479681394332119,
+        "precision": 0.9479408755404257,
+        "recall": 0.9481154277852203,
+        "support": 27619.0
       },
       "step": 1539
     },
     {
       "epoch": 20.0,
       "eval_B": {
-        "f1-score": 0.8907273354015857,
-        "precision": 0.8843258042436687,
-        "recall": 0.8972222222222223,
-        "support": 1440.0
+        "f1-score": 0.8742857142857142,
+        "precision": 0.8684957426679281,
+        "recall": 0.8801534036433365,
+        "support": 1043.0
       },
       "eval_I": {
-        "f1-score": 0.9620682542229149,
-        "precision": 0.9520958083832335,
-        "recall": 0.972251818223931,
-        "support": 21587.0
+        "f1-score": 0.9602204921293401,
+        "precision": 0.9469289396996189,
+        "recall": 0.9738904899135447,
+        "support": 17350.0
       },
       "eval_O": {
-        "f1-score": 0.9177252296267345,
-        "precision": 0.9396698349174587,
-        "recall": 0.8967822018523823,
-        "support": 10473.0
+        "f1-score": 0.9205305394560855,
+        "precision": 0.9473503097040605,
+        "recall": 0.8951875135486668,
+        "support": 9226.0
       },
-      "eval_accuracy": 0.9454328358208955,
-      "eval_loss": 0.44444534182548523,
+      "eval_accuracy": 0.9440602483797386,
+      "eval_loss": 0.36765894293785095,
       "eval_macro avg": {
-        "f1-score": 0.9235069397504118,
-        "precision": 0.9253638158481202,
-        "recall": 0.922085414099512,
-        "support": 33500.0
-      },
-      "eval_runtime": 4.7631,
-      "eval_samples_per_second": 17.006,
-      "eval_steps_per_second": 2.309,
+        "f1-score": 0.9183455819570466,
+        "precision": 0.9209249973572025,
+        "recall": 0.9164104690351826,
+        "support": 27619.0
+      },
+      "eval_runtime": 4.6404,
+      "eval_samples_per_second": 17.24,
+      "eval_steps_per_second": 2.155,
       "eval_weighted avg": {
-        "f1-score": 0.9451388387095264,
-        "precision": 0.9452980165603072,
-        "recall": 0.9454328358208955,
-        "support": 33500.0
+        "f1-score": 0.9437170171065533,
+        "precision": 0.9441077562808465,
+        "recall": 0.9440602483797386,
+        "support": 27619.0
       },
       "step": 1620
     },
     {
       "epoch": 21.0,
       "eval_B": {
-        "f1-score": 0.890790825059911,
-        "precision": 0.8784604996623903,
-        "recall": 0.9034722222222222,
-        "support": 1440.0
+        "f1-score": 0.8885767790262172,
+        "precision": 0.868252516010979,
+        "recall": 0.909875359539789,
+        "support": 1043.0
       },
       "eval_I": {
-        "f1-score": 0.9615375783600082,
-        "precision": 0.9533284764593388,
-        "recall": 0.9698892852179553,
-        "support": 21587.0
+        "f1-score": 0.9616143138880155,
+        "precision": 0.9537903271531439,
+        "recall": 0.9695677233429395,
+        "support": 17350.0
       },
       "eval_O": {
-        "f1-score": 0.9162201656113005,
-        "precision": 0.9351695336581486,
-        "recall": 0.8980234889716414,
-        "support": 10473.0
+        "f1-score": 0.9245376759591498,
+        "precision": 0.9420632242096973,
+        "recall": 0.9076522870149577,
+        "support": 9226.0
       },
-      "eval_accuracy": 0.9445671641791045,
-      "eval_loss": 0.4363422691822052,
+      "eval_accuracy": 0.9466309424671422,
+      "eval_loss": 0.374899685382843,
       "eval_macro avg": {
-        "f1-score": 0.9228495230104065,
-        "precision": 0.9223195032599593,
-        "recall": 0.9237949988039397,
-        "support": 33500.0
-      },
-      "eval_runtime": 4.7902,
-      "eval_samples_per_second": 16.91,
-      "eval_steps_per_second": 2.296,
+        "f1-score": 0.9249095896244609,
+        "precision": 0.9213686891246068,
+        "recall": 0.9290317899658954,
+        "support": 27619.0
+      },
+      "eval_runtime": 4.5979,
+      "eval_samples_per_second": 17.399,
+      "eval_steps_per_second": 2.175,
       "eval_weighted avg": {
-        "f1-score": 0.9443290831818184,
-        "precision": 0.9444332975177129,
-        "recall": 0.9445671641791045,
-        "support": 33500.0
+        "f1-score": 0.9464708542988716,
+        "precision": 0.9466427045463329,
+        "recall": 0.9466309424671422,
+        "support": 27619.0
       },
       "step": 1701
     },
     {
       "epoch": 22.0,
       "eval_B": {
-        "f1-score": 0.8907216494845361,
-        "precision": 0.8816326530612245,
-        "recall": 0.9,
-        "support": 1440.0
+        "f1-score": 0.875898418782942,
+        "precision": 0.8754789272030651,
+        "recall": 0.8763183125599233,
+        "support": 1043.0
       },
       "eval_I": {
-        "f1-score": 0.961758728985775,
-        "precision": 0.9588801399825022,
-        "recall": 0.9646546532635383,
-        "support": 21587.0
+        "f1-score": 0.9606979453806588,
+        "precision": 0.9475306911822412,
+        "recall": 0.9742363112391931,
+        "support": 17350.0
       },
       "eval_O": {
-        "f1-score": 0.9186952756663139,
-        "precision": 0.9258217783380199,
-        "recall": 0.9116776472834909,
-        "support": 10473.0
+        "f1-score": 0.9217236387930074,
+        "precision": 0.94757326007326,
+        "recall": 0.897246910903967,
+        "support": 9226.0
       },
-      "eval_accuracy": 0.9453134328358209,
-      "eval_loss": 0.3998025953769684,
+      "eval_accuracy": 0.9448205945182664,
+      "eval_loss": 0.3899074196815491,
       "eval_macro avg": {
-        "f1-score": 0.9237252180455416,
-        "precision": 0.9221115237939155,
-        "recall": 0.925444100182343,
-        "support": 33500.0
-      },
-      "eval_runtime": 4.7313,
-      "eval_samples_per_second": 17.12,
-      "eval_steps_per_second": 2.325,
+        "f1-score": 0.919440000985536,
+        "precision": 0.9235276261528554,
+        "recall": 0.9159338449010278,
+        "support": 27619.0
+      },
+      "eval_runtime": 4.5958,
+      "eval_samples_per_second": 17.407,
+      "eval_steps_per_second": 2.176,
       "eval_weighted avg": {
-        "f1-score": 0.9452424023858794,
-        "precision": 0.9452247190072991,
-        "recall": 0.9453134328358209,
-        "support": 33500.0
+        "f1-score": 0.9444764001104068,
+        "precision": 0.9448239585256736,
+        "recall": 0.9448205945182664,
+        "support": 27619.0
       },
       "step": 1782
     },
     {
       "epoch": 23.0,
       "eval_B": {
-        "f1-score": 0.8938053097345132,
-        "precision": 0.8765020026702269,
-        "recall": 0.9118055555555555,
-        "support": 1440.0
+        "f1-score": 0.8766079085278704,
+        "precision": 0.8712121212121212,
+        "recall": 0.8820709491850431,
+        "support": 1043.0
       },
       "eval_I": {
-        "f1-score": 0.9620159689850926,
-        "precision": 0.9584770313146641,
-        "recall": 0.9655811367952934,
-        "support": 21587.0
+        "f1-score": 0.9571775605897973,
+        "precision": 0.9385663638378019,
+        "recall": 0.9765417867435159,
+        "support": 17350.0
       },
       "eval_O": {
-        "f1-score": 0.918660748745658,
-        "precision": 0.9284251584592882,
-        "recall": 0.9090995894204144,
-        "support": 10473.0
+        "f1-score": 0.9132322264193494,
+        "precision": 0.9515920573375631,
+        "recall": 0.877845220030349,
+        "support": 9226.0
       },
-      "eval_accuracy": 0.9456119402985075,
-      "eval_loss": 0.4420088231563568,
+      "eval_accuracy": 0.9400050689742568,
+      "eval_loss": 0.4506397247314453,
       "eval_macro avg": {
-        "f1-score": 0.9248273424884212,
-        "precision": 0.9211347308147264,
-        "recall": 0.9288287605904211,
-        "support": 33500.0
-      },
-      "eval_runtime": 4.7375,
-      "eval_samples_per_second": 17.098,
-      "eval_steps_per_second": 2.322,
+        "f1-score": 0.9156725651790056,
+        "precision": 0.9204568474624955,
+        "recall": 0.9121526519863027,
+        "support": 27619.0
+      },
+      "eval_runtime": 4.6193,
+      "eval_samples_per_second": 17.319,
+      "eval_steps_per_second": 2.165,
       "eval_weighted avg": {
-        "f1-score": 0.9455299220929005,
-        "precision": 0.9455583057725045,
-        "recall": 0.9456119402985075,
-        "support": 33500.0
+        "f1-score": 0.939455202786939,
+        "precision": 0.9403739808105457,
+        "recall": 0.9400050689742568,
+        "support": 27619.0
       },
       "step": 1863
     },
     {
       "epoch": 24.0,
       "eval_B": {
-        "f1-score": 0.89482699554642,
-        "precision": 0.8830290736984449,
-        "recall": 0.9069444444444444,
-        "support": 1440.0
+        "f1-score": 0.8786853552440792,
+        "precision": 0.8859649122807017,
+        "recall": 0.8715244487056567,
+        "support": 1043.0
       },
       "eval_I": {
-        "f1-score": 0.962135833103575,
-        "precision": 0.9544605005242285,
-        "recall": 0.969935609394543,
-        "support": 21587.0
+        "f1-score": 0.9558590028580501,
+        "precision": 0.9575452599919023,
+        "recall": 0.954178674351585,
+        "support": 17350.0
       },
       "eval_O": {
-        "f1-score": 0.9185192391885975,
-        "precision": 0.9362356207854026,
-        "recall": 0.9014608994557434,
-        "support": 10473.0
+        "f1-score": 0.9160280626011873,
+        "precision": 0.9121883061049011,
+        "recall": 0.9199002818122697,
+        "support": 9226.0
       },
-      "eval_accuracy": 0.945820895522388,
-      "eval_loss": 0.4822453558444977,
+      "eval_accuracy": 0.9396067924255042,
+      "eval_loss": 0.42392534017562866,
       "eval_macro avg": {
-        "f1-score": 0.9251606892795309,
-        "precision": 0.924575065002692,
-        "recall": 0.9261136510982436,
-        "support": 33500.0
-      },
-      "eval_runtime": 4.7345,
-      "eval_samples_per_second": 17.108,
-      "eval_steps_per_second": 2.323,
+        "f1-score": 0.9168574735677723,
+        "precision": 0.9185661594591683,
+        "recall": 0.9152011349565038,
+        "support": 27619.0
+      },
+      "eval_runtime": 4.6242,
+      "eval_samples_per_second": 17.3,
+      "eval_steps_per_second": 2.163,
       "eval_weighted avg": {
-        "f1-score": 0.9456068386512209,
-        "precision": 0.945692428281427,
-        "recall": 0.945820895522388,
-        "support": 33500.0
+        "f1-score": 0.9396392856607877,
+        "precision": 0.9396908279261412,
+        "recall": 0.9396067924255042,
+        "support": 27619.0
       },
       "step": 1944
     },
     {
       "epoch": 24.69,
-      "grad_norm": 0.10262901335954666,
+      "grad_norm": 0.030169092118740082,
       "learning_rate": 1.0123456790123458e-05,
-      "loss": 0.0037,
+      "loss": 0.0026,
       "step": 2000
     },
     {
       "epoch": 25.0,
       "eval_B": {
-        "f1-score": 0.8864342423196411,
-        "precision": 0.8812628689087165,
-        "recall": 0.8916666666666667,
-        "support": 1440.0
+        "f1-score": 0.8887832699619772,
+        "precision": 0.88124410933082,
+        "recall": 0.8964525407478428,
+        "support": 1043.0
       },
       "eval_I": {
-        "f1-score": 0.9595285234126801,
-        "precision": 0.9537320946409775,
-        "recall": 0.9653958400889424,
-        "support": 21587.0
+        "f1-score": 0.9607269808508199,
+        "precision": 0.9556860955857192,
+        "recall": 0.9658213256484149,
+        "support": 17350.0
       },
       "eval_O": {
-        "f1-score": 0.9129445923058311,
-        "precision": 0.9255298273155416,
-        "recall": 0.9006970304592762,
-        "support": 10473.0
+        "f1-score": 0.9237260273972602,
+        "precision": 0.9340647163120568,
+        "recall": 0.9136137004118795,
+        "support": 9226.0
       },
-      "eval_accuracy": 0.942,
-      "eval_loss": 0.46240004897117615,
+      "eval_accuracy": 0.9457619754516818,
+      "eval_loss": 0.398232638835907,
       "eval_macro avg": {
-        "f1-score": 0.9196357860127174,
-        "precision": 0.9201749302884119,
-        "recall": 0.9192531790716284,
-        "support": 33500.0
-      },
-      "eval_runtime": 4.7408,
-      "eval_samples_per_second": 17.086,
-      "eval_steps_per_second": 2.32,
+        "f1-score": 0.9244120927366858,
+        "precision": 0.9236649737428652,
+        "recall": 0.9252958556027124,
+        "support": 27619.0
+      },
+      "eval_runtime": 4.5953,
+      "eval_samples_per_second": 17.409,
+      "eval_steps_per_second": 2.176,
       "eval_weighted avg": {
-        "f1-score": 0.941823171912501,
-        "precision": 0.9418002131259403,
-        "recall": 0.942,
-        "support": 33500.0
+        "f1-score": 0.9456501103261954,
+        "precision": 0.9456523566073828,
+        "recall": 0.9457619754516818,
+        "support": 27619.0
       },
       "step": 2025
     },
     {
       "epoch": 26.0,
       "eval_B": {
-        "f1-score": 0.892392049348869,
-        "precision": 0.8809201623815968,
-        "recall": 0.9041666666666667,
-        "support": 1440.0
+        "f1-score": 0.8827717133364974,
+        "precision": 0.8740601503759399,
+        "recall": 0.8916586768935763,
+        "support": 1043.0
       },
       "eval_I": {
-        "f1-score": 0.9613188432492171,
-        "precision": 0.9556420233463035,
-        "recall": 0.9670635104461018,
-        "support": 21587.0
+        "f1-score": 0.9590283123770422,
+        "precision": 0.9489364103142809,
+        "recall": 0.9693371757925072,
+        "support": 17350.0
       },
       "eval_O": {
-        "f1-score": 0.9172881355932202,
-        "precision": 0.9306278864105335,
-        "recall": 0.904325408192495,
-        "support": 10473.0
+        "f1-score": 0.9188171447557869,
+        "precision": 0.9393115942028986,
+        "recall": 0.8991979189247779,
+        "support": 9226.0
       },
-      "eval_accuracy": 0.9447462686567164,
-      "eval_loss": 0.4685353934764862,
+      "eval_accuracy": 0.9429740396104132,
+      "eval_loss": 0.4066696763038635,
       "eval_macro avg": {
-        "f1-score": 0.9236663427304355,
-        "precision": 0.9223966907128114,
-        "recall": 0.9251851951017546,
-        "support": 33500.0
-      },
-      "eval_runtime": 4.7522,
-      "eval_samples_per_second": 17.045,
-      "eval_steps_per_second": 2.315,
+        "f1-score": 0.9202057234897755,
+        "precision": 0.9207693849643731,
+        "recall": 0.9200645905369539,
+        "support": 27619.0
+      },
+      "eval_runtime": 4.6361,
+      "eval_samples_per_second": 17.256,
+      "eval_steps_per_second": 2.157,
       "eval_weighted avg": {
-        "f1-score": 0.9445908377418215,
-        "precision": 0.9446100073487664,
-        "recall": 0.9447462686567164,
-        "support": 33500.0
+        "f1-score": 0.9427162132687112,
+        "precision": 0.942893668268613,
+        "recall": 0.9429740396104132,
+        "support": 27619.0
       },
       "step": 2106
     }
@@ -1056,7 +1056,7 @@
   "num_input_tokens_seen": 0,
   "num_train_epochs": 50,
   "save_steps": 500,
-  "total_flos": 3726842868730800.0,
+  "total_flos": 3738452971125600.0,
   "train_batch_size": 4,
   "trial_name": null,
   "trial_params": null

checkpoint-2106/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b8ca363922308e3892528dd031742ffedc372734b464a042d09444af9700901e
+oid sha256:821ca433db9ddb761291ab61fc06d651ad248cd9d560cdb25f7f905e5aaff14e
 size 4856