Training in progress, epoch 29, checkpoint

checkpoint-2349/model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:200579c2246be7785505d145ec4e354424fcdd19ece082c42b65067ae4724ff8
+oid sha256:d1564c228da2c8e33d8210a81db4adba57ba2b47798fd2d79496c8604dbad25f
 size 592318676

checkpoint-2349/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:258b44e4829b438e45e91011b75bce205afebee1b7a6b793a697096cb305db1a
+oid sha256:53908afcdf21312edfccde83bf331624cf52ceafe6418fdc9d73e59e7d0e495a
 size 1014645498

checkpoint-2349/rng_state.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a61d9b0a50203aa7e50c1fec93ffa66a6f7302e114ab037b17b0a96fea857321
+oid sha256:2ee34e8787f45db18ff26894b40e12d289b82c56f36ac756701b94bfd8c7bd62
 size 14244

checkpoint-2349/trainer_state.json

@@ -11,1159 +11,1159 @@
     {
       "epoch": 1.0,
       "eval_B": {
-        "f1-score": 0.8087666523420713,
-        "precision": 0.8189730200174065,
-        "recall": 0.7988115449915111,
-        "support": 1178.0
+        "f1-score": 0.8266764382557715,
+        "precision": 0.7866108786610879,
+        "recall": 0.871042471042471,
+        "support": 1295.0
       },
       "eval_I": {
-        "f1-score": 0.9374597034171502,
-        "precision": 0.9521418826739427,
-        "recall": 0.9232234509762421,
-        "support": 18899.0
+        "f1-score": 0.9410658617447671,
+        "precision": 0.9344077040239768,
+        "recall": 0.9478195863443808,
+        "support": 20065.0
       },
       "eval_O": {
-        "f1-score": 0.8881362400419788,
-        "precision": 0.863303347862376,
-        "recall": 0.9144400785854617,
-        "support": 10180.0
+        "f1-score": 0.8562443302086484,
+        "precision": 0.8789421405512788,
+        "recall": 0.8346893055064262,
+        "support": 8481.0
       },
-      "eval_accuracy": 0.9154245298608586,
-      "eval_loss": 0.2228638380765915,
+      "eval_accuracy": 0.912335377500754,
+      "eval_loss": 0.2450982630252838,
       "eval_macro avg": {
-        "f1-score": 0.8781208652670669,
-        "precision": 0.8781394168512416,
-        "recall": 0.8788250248510717,
-        "support": 30257.0
-      },
-      "eval_runtime": 4.7167,
-      "eval_samples_per_second": 17.173,
-      "eval_steps_per_second": 2.332,
+        "f1-score": 0.8746622100697291,
+        "precision": 0.8666535744121145,
+        "recall": 0.8845171209644261,
+        "support": 29841.0
+      },
+      "eval_runtime": 4.6269,
+      "eval_samples_per_second": 17.29,
+      "eval_steps_per_second": 2.161,
       "eval_weighted avg": {
-        "f1-score": 0.9158543469268937,
-        "precision": 0.9170673807540184,
-        "recall": 0.9154245298608586,
-        "support": 30257.0
+        "f1-score": 0.9119949287205363,
+        "precision": 0.9122301519092053,
+        "recall": 0.912335377500754,
+        "support": 29841.0
       },
       "step": 81
     },
     {
       "epoch": 2.0,
       "eval_B": {
-        "f1-score": 0.8401515151515152,
-        "precision": 0.7585499316005472,
-        "recall": 0.9414261460101867,
-        "support": 1178.0
+        "f1-score": 0.8833395452851286,
+        "precision": 0.8537463976945245,
+        "recall": 0.915057915057915,
+        "support": 1295.0
       },
       "eval_I": {
-        "f1-score": 0.9568235603470944,
-        "precision": 0.9510219016256338,
-        "recall": 0.9626964389650247,
-        "support": 18899.0
+        "f1-score": 0.9549518227873249,
+        "precision": 0.9516903430183636,
+        "recall": 0.958235733864939,
+        "support": 20065.0
       },
       "eval_O": {
-        "f1-score": 0.9185648054827655,
-        "precision": 0.9430877483443708,
-        "recall": 0.8952848722986247,
-        "support": 10180.0
+        "f1-score": 0.88924750463212,
+        "precision": 0.9016969696969697,
+        "recall": 0.877137130055418,
+        "support": 8481.0
       },
-      "eval_accuracy": 0.9391876260038999,
-      "eval_loss": 0.16692574322223663,
+      "eval_accuracy": 0.9333132267685399,
+      "eval_loss": 0.20076116919517517,
       "eval_macro avg": {
-        "f1-score": 0.905179960327125,
-        "precision": 0.8842198605235172,
-        "recall": 0.9331358190912787,
-        "support": 30257.0
-      },
-      "eval_runtime": 4.7116,
-      "eval_samples_per_second": 17.192,
-      "eval_steps_per_second": 2.335,
+        "f1-score": 0.9091796242348579,
+        "precision": 0.9023779034699526,
+        "recall": 0.916810259659424,
+        "support": 29841.0
+      },
+      "eval_runtime": 4.6524,
+      "eval_samples_per_second": 17.195,
+      "eval_steps_per_second": 2.149,
       "eval_weighted avg": {
-        "f1-score": 0.9394089523635119,
-        "precision": 0.9408589092241132,
-        "recall": 0.9391876260038999,
-        "support": 30257.0
+        "f1-score": 0.9331705077630417,
+        "precision": 0.9332314707173981,
+        "recall": 0.9333132267685399,
+        "support": 29841.0
       },
       "step": 162
     },
     {
       "epoch": 3.0,
       "eval_B": {
-        "f1-score": 0.8612244897959184,
-        "precision": 0.8294025157232704,
-        "recall": 0.8955857385398981,
-        "support": 1178.0
+        "f1-score": 0.8927335640138409,
+        "precision": 0.8889739663093414,
+        "recall": 0.8965250965250965,
+        "support": 1295.0
       },
       "eval_I": {
-        "f1-score": 0.951106371634231,
-        "precision": 0.9584654236741712,
-        "recall": 0.9438594634636753,
-        "support": 18899.0
+        "f1-score": 0.9590235396687009,
+        "precision": 0.9586653386454184,
+        "recall": 0.9593820084724645,
+        "support": 20065.0
       },
       "eval_O": {
-        "f1-score": 0.9106743212999903,
-        "precision": 0.9021592442645074,
-        "recall": 0.9193516699410609,
-        "support": 10180.0
+        "f1-score": 0.9006849315068494,
+        "precision": 0.9020697811945594,
+        "recall": 0.8993043273198915,
+        "support": 8481.0
       },
-      "eval_accuracy": 0.9337343424662061,
-      "eval_loss": 0.1742253601551056,
+      "eval_accuracy": 0.9395797727958178,
+      "eval_loss": 0.19863653182983398,
       "eval_macro avg": {
-        "f1-score": 0.9076683942433799,
-        "precision": 0.8966757278873163,
-        "recall": 0.9195989573148781,
-        "support": 30257.0
-      },
-      "eval_runtime": 4.7094,
-      "eval_samples_per_second": 17.2,
-      "eval_steps_per_second": 2.336,
+        "f1-score": 0.9174806783964637,
+        "precision": 0.9165696953831063,
+        "recall": 0.9184038107724842,
+        "support": 29841.0
+      },
+      "eval_runtime": 4.655,
+      "eval_samples_per_second": 17.186,
+      "eval_steps_per_second": 2.148,
       "eval_weighted avg": {
-        "f1-score": 0.9340035812317422,
-        "precision": 0.9344963252190521,
-        "recall": 0.9337343424662061,
-        "support": 30257.0
+        "f1-score": 0.9395665759512081,
+        "precision": 0.9395561516236713,
+        "recall": 0.9395797727958178,
+        "support": 29841.0
       },
       "step": 243
     },
     {
       "epoch": 4.0,
       "eval_B": {
-        "f1-score": 0.8717948717948719,
-        "precision": 0.8254931714719271,
-        "recall": 0.9235993208828522,
-        "support": 1178.0
+        "f1-score": 0.8999616711383672,
+        "precision": 0.893455098934551,
+        "recall": 0.9065637065637066,
+        "support": 1295.0
       },
       "eval_I": {
-        "f1-score": 0.9561271586371415,
-        "precision": 0.9374586871408959,
-        "recall": 0.9755542621302714,
-        "support": 18899.0
+        "f1-score": 0.9612657113512985,
+        "precision": 0.9436362763454799,
+        "recall": 0.9795664091701969,
+        "support": 20065.0
       },
       "eval_O": {
-        "f1-score": 0.9117828500925355,
-        "precision": 0.956427955133736,
-        "recall": 0.8711198428290766,
-        "support": 10180.0
+        "f1-score": 0.8996847765622102,
+        "precision": 0.9454403741231489,
+        "recall": 0.8581535196321188,
+        "support": 8481.0
       },
-      "eval_accuracy": 0.9383944211256899,
-      "eval_loss": 0.18792153894901276,
+      "eval_accuracy": 0.9418920277470594,
+      "eval_loss": 0.19743724167346954,
       "eval_macro avg": {
-        "f1-score": 0.9132349601748496,
-        "precision": 0.9064599379155197,
-        "recall": 0.9234244752807333,
-        "support": 30257.0
-      },
-      "eval_runtime": 4.7342,
-      "eval_samples_per_second": 17.11,
-      "eval_steps_per_second": 2.324,
+        "f1-score": 0.920304053017292,
+        "precision": 0.9275105831343932,
+        "recall": 0.9147612117886741,
+        "support": 29841.0
+      },
+      "eval_runtime": 4.6615,
+      "eval_samples_per_second": 17.162,
+      "eval_steps_per_second": 2.145,
       "eval_weighted avg": {
-        "f1-score": 0.9379241479327001,
-        "precision": 0.9394817486046585,
-        "recall": 0.9383944211256899,
-        "support": 30257.0
+        "f1-score": 0.9411036309913238,
+        "precision": 0.9419713163409645,
+        "recall": 0.9418920277470594,
+        "support": 29841.0
       },
       "step": 324
     },
     {
       "epoch": 5.0,
       "eval_B": {
-        "f1-score": 0.8742949234488315,
-        "precision": 0.8320552147239264,
-        "recall": 0.9210526315789473,
-        "support": 1178.0
+        "f1-score": 0.8953846153846153,
+        "precision": 0.8919540229885058,
+        "recall": 0.8988416988416988,
+        "support": 1295.0
       },
       "eval_I": {
-        "f1-score": 0.9583957528450158,
-        "precision": 0.9521148825065274,
-        "recall": 0.9647600402137679,
-        "support": 18899.0
+        "f1-score": 0.9602426713742261,
+        "precision": 0.9581224570804803,
+        "recall": 0.9623722900573137,
+        "support": 20065.0
       },
       "eval_O": {
-        "f1-score": 0.9198818996146724,
-        "precision": 0.9375701315923697,
-        "recall": 0.9028487229862475,
-        "support": 10180.0
+        "f1-score": 0.9022119433078337,
+        "precision": 0.9075399665950847,
+        "recall": 0.8969461148449476,
+        "support": 8481.0
       },
-      "eval_accuracy": 0.9422282447037049,
-      "eval_loss": 0.19990424811840057,
+      "eval_accuracy": 0.9410207432726785,
+      "eval_loss": 0.2340461015701294,
       "eval_macro avg": {
-        "f1-score": 0.9175241919695066,
-        "precision": 0.9072467429409411,
-        "recall": 0.9295537982596542,
-        "support": 30257.0
-      },
-      "eval_runtime": 4.7101,
-      "eval_samples_per_second": 17.197,
-      "eval_steps_per_second": 2.335,
+        "f1-score": 0.9192797433555584,
+        "precision": 0.9192054822213569,
+        "recall": 0.9193867012479867,
+        "support": 29841.0
+      },
+      "eval_runtime": 4.6719,
+      "eval_samples_per_second": 17.124,
+      "eval_steps_per_second": 2.14,
       "eval_weighted avg": {
-        "f1-score": 0.9421634164298524,
-        "precision": 0.9425469857238316,
-        "recall": 0.9422282447037049,
-        "support": 30257.0
+        "f1-score": 0.9409353496612266,
+        "precision": 0.9408751053176122,
+        "recall": 0.9410207432726785,
+        "support": 29841.0
       },
       "step": 405
     },
     {
       "epoch": 6.0,
       "eval_B": {
-        "f1-score": 0.8808247422680412,
-        "precision": 0.8564554931836408,
-        "recall": 0.9066213921901528,
-        "support": 1178.0
+        "f1-score": 0.8929831438651511,
+        "precision": 0.9068471337579618,
+        "recall": 0.8795366795366796,
+        "support": 1295.0
       },
       "eval_I": {
-        "f1-score": 0.957406827435014,
-        "precision": 0.9446361435855178,
-        "recall": 0.9705275411397428,
-        "support": 18899.0
+        "f1-score": 0.9522293798155866,
+        "precision": 0.9655225409836066,
+        "recall": 0.9392972838275604,
+        "support": 20065.0
       },
       "eval_O": {
-        "f1-score": 0.9163000050574014,
-        "precision": 0.9443344105076619,
-        "recall": 0.8898821218074656,
-        "support": 10180.0
+        "f1-score": 0.8885216003647556,
+        "precision": 0.8599007170435742,
+        "recall": 0.919113312109421,
+        "support": 8481.0
       },
-      "eval_accuracy": 0.9409062365733549,
-      "eval_loss": 0.20615626871585846,
+      "eval_accuracy": 0.930967460875976,
+      "eval_loss": 0.21774999797344208,
       "eval_macro avg": {
-        "f1-score": 0.9181771915868189,
-        "precision": 0.9151420157589403,
-        "recall": 0.922343685045787,
-        "support": 30257.0
-      },
-      "eval_runtime": 4.6949,
-      "eval_samples_per_second": 17.253,
-      "eval_steps_per_second": 2.343,
+        "f1-score": 0.9112447080151643,
+        "precision": 0.9107567972617142,
+        "recall": 0.9126490918245537,
+        "support": 29841.0
+      },
+      "eval_runtime": 4.647,
+      "eval_samples_per_second": 17.215,
+      "eval_steps_per_second": 2.152,
       "eval_weighted avg": {
-        "f1-score": 0.9405948120953971,
-        "precision": 0.941101475610967,
-        "recall": 0.9409062365733549,
-        "support": 30257.0
+        "f1-score": 0.9315521386682287,
+        "precision": 0.9329578366776978,
+        "recall": 0.930967460875976,
+        "support": 29841.0
       },
       "step": 486
     },
     {
       "epoch": 6.17,
-      "grad_norm": 2.4104158878326416,
+      "grad_norm": 9.310327529907227,
       "learning_rate": 1.7530864197530865e-05,
-      "loss": 0.1662,
+      "loss": 0.1627,
       "step": 500
     },
     {
       "epoch": 7.0,
       "eval_B": {
-        "f1-score": 0.8836820083682009,
-        "precision": 0.8712871287128713,
-        "recall": 0.8964346349745331,
-        "support": 1178.0
+        "f1-score": 0.8936490041337843,
+        "precision": 0.8704245973645681,
+        "recall": 0.9181467181467181,
+        "support": 1295.0
       },
       "eval_I": {
-        "f1-score": 0.9614174269469787,
-        "precision": 0.9609854091774159,
-        "recall": 0.9618498333245146,
-        "support": 18899.0
+        "f1-score": 0.9599271850630982,
+        "precision": 0.9477800446905664,
+        "recall": 0.9723897333665587,
+        "support": 20065.0
       },
       "eval_O": {
-        "f1-score": 0.9271751440248166,
-        "precision": 0.9295093296475466,
-        "recall": 0.924852652259332,
-        "support": 10180.0
+        "f1-score": 0.8979841172877214,
+        "precision": 0.9316770186335404,
+        "recall": 0.8666430845419172,
+        "support": 8481.0
       },
-      "eval_accuracy": 0.9468552731599299,
-      "eval_loss": 0.22925782203674316,
+      "eval_accuracy": 0.9399819040916859,
+      "eval_loss": 0.27376389503479004,
       "eval_macro avg": {
-        "f1-score": 0.9240915264466655,
-        "precision": 0.9205939558459445,
-        "recall": 0.9277123735194599,
-        "support": 30257.0
-      },
-      "eval_runtime": 4.7288,
-      "eval_samples_per_second": 17.129,
-      "eval_steps_per_second": 2.326,
+        "f1-score": 0.9171867688282013,
+        "precision": 0.9166272202295582,
+        "recall": 0.9190598453517312,
+        "support": 29841.0
+      },
+      "eval_runtime": 4.6671,
+      "eval_samples_per_second": 17.141,
+      "eval_steps_per_second": 2.143,
       "eval_weighted avg": {
-        "f1-score": 0.9468700903559945,
-        "precision": 0.9469030129054358,
-        "recall": 0.9468552731599299,
-        "support": 30257.0
+        "f1-score": 0.9394463230910989,
+        "precision": 0.9398464945991887,
+        "recall": 0.9399819040916859,
+        "support": 29841.0
       },
       "step": 567
     },
     {
       "epoch": 8.0,
       "eval_B": {
-        "f1-score": 0.8803630363036303,
-        "precision": 0.8563402889245586,
-        "recall": 0.9057724957555179,
-        "support": 1178.0
+        "f1-score": 0.8948545861297539,
+        "precision": 0.8651766402307137,
+        "recall": 0.9266409266409267,
+        "support": 1295.0
       },
       "eval_I": {
-        "f1-score": 0.9580651967507121,
-        "precision": 0.9550928116947994,
-        "recall": 0.9610561405365363,
-        "support": 18899.0
+        "f1-score": 0.9590139880661254,
+        "precision": 0.9414702069429106,
+        "recall": 0.9772240219287316,
+        "support": 20065.0
       },
       "eval_O": {
-        "f1-score": 0.9208882720333101,
-        "precision": 0.9294576746047628,
-        "recall": 0.912475442043222,
-        "support": 10180.0
+        "f1-score": 0.8927241122423641,
+        "precision": 0.9427035531663825,
+        "recall": 0.8477773847423653,
+        "support": 8481.0
       },
-      "eval_accuracy": 0.9425587467362925,
-      "eval_loss": 0.22973769903182983,
+      "eval_accuracy": 0.9382393351429241,
+      "eval_loss": 0.3103167414665222,
       "eval_macro avg": {
-        "f1-score": 0.9197721683625509,
-        "precision": 0.9136302584080402,
-        "recall": 0.9264346927784253,
-        "support": 30257.0
-      },
-      "eval_runtime": 4.7191,
-      "eval_samples_per_second": 17.164,
-      "eval_steps_per_second": 2.331,
+        "f1-score": 0.9155308954794145,
+        "precision": 0.916450133446669,
+        "recall": 0.9172141111040079,
+        "support": 29841.0
+      },
+      "eval_runtime": 4.6339,
+      "eval_samples_per_second": 17.264,
+      "eval_steps_per_second": 2.158,
       "eval_weighted avg": {
-        "f1-score": 0.9425317916335554,
-        "precision": 0.942623096673485,
-        "recall": 0.9425587467362925,
-        "support": 30257.0
+        "f1-score": 0.9373896838414373,
+        "precision": 0.9385098450391195,
+        "recall": 0.9382393351429241,
+        "support": 29841.0
       },
       "step": 648
     },
     {
       "epoch": 9.0,
       "eval_B": {
-        "f1-score": 0.8885191347753745,
-        "precision": 0.8711256117455138,
-        "recall": 0.9066213921901528,
-        "support": 1178.0
+        "f1-score": 0.9011235955056179,
+        "precision": 0.8749090909090909,
+        "recall": 0.9289575289575289,
+        "support": 1295.0
       },
       "eval_I": {
-        "f1-score": 0.9620326487625066,
-        "precision": 0.9574445783763954,
-        "recall": 0.9666649029049156,
-        "support": 18899.0
+        "f1-score": 0.9615119937082186,
+        "precision": 0.9485040973670175,
+        "recall": 0.9748816346872664,
+        "support": 20065.0
       },
       "eval_O": {
-        "f1-score": 0.9275707898658718,
-        "precision": 0.9382914572864322,
-        "recall": 0.9170923379174852,
-        "support": 10180.0
+        "f1-score": 0.9013722126929674,
+        "precision": 0.9380339155935229,
+        "recall": 0.8674684589081476,
+        "support": 8481.0
       },
-      "eval_accuracy": 0.9476484780381399,
-      "eval_loss": 0.24401217699050903,
+      "eval_accuracy": 0.9423611809255722,
+      "eval_loss": 0.32883673906326294,
       "eval_macro avg": {
-        "f1-score": 0.9260408578012509,
-        "precision": 0.9222872158027805,
-        "recall": 0.9301262110041845,
-        "support": 30257.0
-      },
-      "eval_runtime": 4.701,
-      "eval_samples_per_second": 17.23,
-      "eval_steps_per_second": 2.34,
+        "f1-score": 0.9213359339689347,
+        "precision": 0.9204823679565438,
+        "recall": 0.9237692075176476,
+        "support": 29841.0
+      },
+      "eval_runtime": 4.6714,
+      "eval_samples_per_second": 17.126,
+      "eval_steps_per_second": 2.141,
       "eval_weighted avg": {
-        "f1-score": 0.94757580760031,
-        "precision": 0.9476398219436029,
-        "recall": 0.9476484780381399,
-        "support": 30257.0
+        "f1-score": 0.9417992341337168,
+        "precision": 0.9423346276781992,
+        "recall": 0.9423611809255722,
+        "support": 29841.0
       },
       "step": 729
     },
     {
       "epoch": 10.0,
       "eval_B": {
-        "f1-score": 0.875959595959596,
-        "precision": 0.8357748650732459,
-        "recall": 0.9202037351443124,
-        "support": 1178.0
+        "f1-score": 0.8935361216730039,
+        "precision": 0.8801498127340824,
+        "recall": 0.9073359073359073,
+        "support": 1295.0
       },
       "eval_I": {
-        "f1-score": 0.955652945166171,
-        "precision": 0.9424749163879599,
-        "recall": 0.9692047198264458,
-        "support": 18899.0
+        "f1-score": 0.9632332861646931,
+        "precision": 0.9596359319351009,
+        "recall": 0.9668577124345876,
+        "support": 20065.0
       },
       "eval_O": {
-        "f1-score": 0.9127632580563309,
-        "precision": 0.9441469816272966,
-        "recall": 0.8833988212180747,
-        "support": 10180.0
+        "f1-score": 0.9095462405342556,
+        "precision": 0.9200241254523522,
+        "recall": 0.8993043273198915,
+        "support": 8481.0
       },
-      "eval_accuracy": 0.9384274713289487,
-      "eval_loss": 0.29471203684806824,
+      "eval_accuracy": 0.9450755671726819,
+      "eval_loss": 0.3041447699069977,
       "eval_macro avg": {
-        "f1-score": 0.9147919330606994,
-        "precision": 0.9074655876961675,
-        "recall": 0.9242690920629443,
-        "support": 30257.0
-      },
-      "eval_runtime": 4.6985,
-      "eval_samples_per_second": 17.24,
-      "eval_steps_per_second": 2.341,
+        "f1-score": 0.9221052161239841,
+        "precision": 0.9199366233738452,
+        "recall": 0.9244993156967954,
+        "support": 29841.0
+      },
+      "eval_runtime": 4.6532,
+      "eval_samples_per_second": 17.193,
+      "eval_steps_per_second": 2.149,
       "eval_weighted avg": {
-        "f1-score": 0.9381199518045186,
-        "precision": 0.9388833165494999,
-        "recall": 0.9384274713289487,
-        "support": 30257.0
+        "f1-score": 0.9449504651463465,
+        "precision": 0.9449285744355027,
+        "recall": 0.9450755671726819,
+        "support": 29841.0
       },
       "step": 810
     },
     {
       "epoch": 11.0,
       "eval_B": {
-        "f1-score": 0.875515251442704,
-        "precision": 0.8509615384615384,
-        "recall": 0.9015280135823429,
-        "support": 1178.0
+        "f1-score": 0.8924122310305775,
+        "precision": 0.8729689807976366,
+        "recall": 0.9127413127413128,
+        "support": 1295.0
       },
       "eval_I": {
-        "f1-score": 0.9575287538250502,
-        "precision": 0.9547582723972855,
-        "recall": 0.96031536060109,
-        "support": 18899.0
+        "f1-score": 0.9588723786993913,
+        "precision": 0.9483330083837005,
+        "recall": 0.9696486419137802,
+        "support": 20065.0
       },
       "eval_O": {
-        "f1-score": 0.9195242814667988,
-        "precision": 0.9278,
-        "recall": 0.9113948919449901,
-        "support": 10180.0
+        "f1-score": 0.8956965718453683,
+        "precision": 0.9243507715468574,
+        "recall": 0.8687654757693668,
+        "support": 8481.0
       },
-      "eval_accuracy": 0.9415672406385299,
-      "eval_loss": 0.30941879749298096,
+      "eval_accuracy": 0.9385074226735028,
+      "eval_loss": 0.3531416654586792,
       "eval_macro avg": {
-        "f1-score": 0.9175227622448511,
-        "precision": 0.9111732702862746,
-        "recall": 0.9244127553761411,
-        "support": 30257.0
-      },
-      "eval_runtime": 4.7017,
-      "eval_samples_per_second": 17.228,
-      "eval_steps_per_second": 2.34,
+        "f1-score": 0.9156603938584458,
+        "precision": 0.9152175869093981,
+        "recall": 0.9170518101414866,
+        "support": 29841.0
+      },
+      "eval_runtime": 4.6508,
+      "eval_samples_per_second": 17.201,
+      "eval_steps_per_second": 2.15,
       "eval_weighted avg": {
-        "f1-score": 0.9415490653426031,
-        "precision": 0.9416470001105195,
-        "recall": 0.9415672406385299,
-        "support": 30257.0
+        "f1-score": 0.938033267772811,
+        "precision": 0.9382465579853486,
+        "recall": 0.9385074226735028,
+        "support": 29841.0
       },
       "step": 891
     },
     {
       "epoch": 12.0,
       "eval_B": {
-        "f1-score": 0.8818181818181817,
-        "precision": 0.8590982286634461,
-        "recall": 0.9057724957555179,
-        "support": 1178.0
+        "f1-score": 0.8970199924556771,
+        "precision": 0.8768436578171092,
+        "recall": 0.9181467181467181,
+        "support": 1295.0
       },
       "eval_I": {
-        "f1-score": 0.9542487087386278,
-        "precision": 0.9457436856875585,
-        "recall": 0.9629080903751521,
-        "support": 18899.0
+        "f1-score": 0.9595004897159647,
+        "precision": 0.9431046931407943,
+        "recall": 0.9764764515325193,
+        "support": 20065.0
       },
       "eval_O": {
-        "f1-score": 0.910740239562973,
-        "precision": 0.9297042873222142,
-        "recall": 0.8925343811394892,
-        "support": 10180.0
+        "f1-score": 0.8953122104873078,
+        "precision": 0.9400778210116731,
+        "recall": 0.8546162009197029,
+        "support": 8481.0
       },
-      "eval_accuracy": 0.9370063125888224,
-      "eval_loss": 0.28610122203826904,
+      "eval_accuracy": 0.9393116852652391,
+      "eval_loss": 0.4237186312675476,
       "eval_macro avg": {
-        "f1-score": 0.9156023767065942,
-        "precision": 0.9115154005577396,
-        "recall": 0.9204049890900531,
-        "support": 30257.0
-      },
-      "eval_runtime": 4.6911,
-      "eval_samples_per_second": 17.267,
-      "eval_steps_per_second": 2.345,
+        "f1-score": 0.9172775642196499,
+        "precision": 0.9200087239898588,
+        "recall": 0.9164131235329801,
+        "support": 29841.0
+      },
+      "eval_runtime": 4.6388,
+      "eval_samples_per_second": 17.246,
+      "eval_steps_per_second": 2.156,
       "eval_weighted avg": {
-        "f1-score": 0.9367902899621314,
-        "precision": 0.9369738332985704,
-        "recall": 0.9370063125888224,
-        "support": 30257.0
+        "f1-score": 0.9385463313402296,
+        "precision": 0.9393689288141548,
+        "recall": 0.9393116852652391,
+        "support": 29841.0
       },
       "step": 972
     },
     {
       "epoch": 12.35,
-      "grad_norm": 0.9797911643981934,
+      "grad_norm": 0.5223535895347595,
       "learning_rate": 1.506172839506173e-05,
-      "loss": 0.0233,
+      "loss": 0.0196,
       "step": 1000
     },
     {
       "epoch": 13.0,
       "eval_B": {
-        "f1-score": 0.8875154511742892,
-        "precision": 0.8622898318654924,
-        "recall": 0.9142614601018676,
-        "support": 1178.0
+        "f1-score": 0.8891411275066211,
+        "precision": 0.8716617210682492,
+        "recall": 0.9073359073359073,
+        "support": 1295.0
       },
       "eval_I": {
-        "f1-score": 0.9602009576889865,
-        "precision": 0.9497877627083549,
-        "recall": 0.9708450182549341,
-        "support": 18899.0
+        "f1-score": 0.9610692322951795,
+        "precision": 0.9563736860287223,
+        "recall": 0.9658111138798904,
+        "support": 20065.0
       },
       "eval_O": {
-        "f1-score": 0.9229994967287367,
-        "precision": 0.9463364293085655,
-        "recall": 0.9007858546168959,
-        "support": 10180.0
+        "f1-score": 0.9043145233678416,
+        "precision": 0.918104495747266,
+        "recall": 0.8909326730338404,
+        "support": 8481.0
       },
-      "eval_accuracy": 0.9450705621839575,
-      "eval_loss": 0.3311145305633545,
+      "eval_accuracy": 0.9419925605710264,
+      "eval_loss": 0.4308999180793762,
       "eval_macro avg": {
-        "f1-score": 0.9235719685306708,
-        "precision": 0.9194713412941375,
-        "recall": 0.9286307776578991,
-        "support": 30257.0
-      },
-      "eval_runtime": 4.7187,
-      "eval_samples_per_second": 17.166,
-      "eval_steps_per_second": 2.331,
+        "f1-score": 0.9181749610565474,
+        "precision": 0.9153799676147458,
+        "recall": 0.9213598980832126,
+        "support": 29841.0
+      },
+      "eval_runtime": 4.6555,
+      "eval_samples_per_second": 17.184,
+      "eval_steps_per_second": 2.148,
       "eval_weighted avg": {
-        "f1-score": 0.9448546114137558,
-        "precision": 0.9452199887538072,
-        "recall": 0.9450705621839575,
-        "support": 30257.0
+        "f1-score": 0.9418177466843106,
+        "precision": 0.9418211242043584,
+        "recall": 0.9419925605710264,
+        "support": 29841.0
       },
       "step": 1053
     },
     {
       "epoch": 14.0,
       "eval_B": {
-        "f1-score": 0.8758762886597938,
-        "precision": 0.8516439454691259,
-        "recall": 0.9015280135823429,
-        "support": 1178.0
+        "f1-score": 0.9025601834161253,
+        "precision": 0.8933434190620272,
+        "recall": 0.911969111969112,
+        "support": 1295.0
       },
       "eval_I": {
-        "f1-score": 0.9588910891089109,
-        "precision": 0.9569456155143339,
-        "recall": 0.9608444891264089,
-        "support": 18899.0
+        "f1-score": 0.960806215404473,
+        "precision": 0.9570764513895758,
+        "recall": 0.9645651632195366,
+        "support": 20065.0
       },
       "eval_O": {
-        "f1-score": 0.9223310576827941,
-        "precision": 0.9290412597169623,
-        "recall": 0.9157170923379175,
-        "support": 10180.0
+        "f1-score": 0.9030873763261413,
+        "precision": 0.9131011208870676,
+        "recall": 0.8932908855087843,
+        "support": 8481.0
       },
-      "eval_accuracy": 0.9433519516145025,
-      "eval_loss": 0.32673054933547974,
+      "eval_accuracy": 0.9420260715123487,
+      "eval_loss": 0.4198700487613678,
       "eval_macro avg": {
-        "f1-score": 0.9190328118171663,
-        "precision": 0.9125436069001407,
-        "recall": 0.9260298650155564,
-        "support": 30257.0
-      },
-      "eval_runtime": 4.7171,
-      "eval_samples_per_second": 17.171,
-      "eval_steps_per_second": 2.332,
+        "f1-score": 0.9221512583822467,
+        "precision": 0.9211736637795568,
+        "recall": 0.9232750535658111,
+        "support": 29841.0
+      },
+      "eval_runtime": 4.6834,
+      "eval_samples_per_second": 17.082,
+      "eval_steps_per_second": 2.135,
       "eval_weighted avg": {
-        "f1-score": 0.9433584006451857,
-        "precision": 0.943457440568685,
-        "recall": 0.9433519516145025,
-        "support": 30257.0
+        "f1-score": 0.9418744743217934,
+        "precision": 0.9418125843993292,
+        "recall": 0.9420260715123487,
+        "support": 29841.0
       },
       "step": 1134
     },
     {
       "epoch": 15.0,
       "eval_B": {
-        "f1-score": 0.8844750709363599,
-        "precision": 0.8463925523661753,
-        "recall": 0.9261460101867572,
-        "support": 1178.0
+        "f1-score": 0.8935681470137825,
+        "precision": 0.8861047835990888,
+        "recall": 0.9011583011583012,
+        "support": 1295.0
       },
       "eval_I": {
-        "f1-score": 0.9554282158321882,
-        "precision": 0.9367500508440105,
-        "recall": 0.974866395047357,
-        "support": 18899.0
+        "f1-score": 0.9633398694432723,
+        "precision": 0.9595530063291139,
+        "recall": 0.9671567405930726,
+        "support": 20065.0
       },
       "eval_O": {
-        "f1-score": 0.9105749486652978,
-        "precision": 0.9536559139784946,
-        "recall": 0.8712180746561886,
-        "support": 10180.0
+        "f1-score": 0.9100768726535964,
+        "precision": 0.92,
+        "recall": 0.9003655229336163,
+        "support": 8481.0
       },
-      "eval_accuracy": 0.9380969692963612,
-      "eval_loss": 0.4276476800441742,
+      "eval_accuracy": 0.9453101437619382,
+      "eval_loss": 0.39179423451423645,
       "eval_macro avg": {
-        "f1-score": 0.9168260784779486,
-        "precision": 0.9122661723962269,
-        "recall": 0.924076826630101,
-        "support": 30257.0
-      },
-      "eval_runtime": 4.6973,
-      "eval_samples_per_second": 17.244,
-      "eval_steps_per_second": 2.342,
+        "f1-score": 0.9223282963702171,
+        "precision": 0.9218859299760677,
+        "recall": 0.9228935215616634,
+        "support": 29841.0
+      },
+      "eval_runtime": 4.6645,
+      "eval_samples_per_second": 17.151,
+      "eval_steps_per_second": 2.144,
       "eval_weighted avg": {
-        "f1-score": 0.9375748574540862,
-        "precision": 0.9389201454833387,
-        "recall": 0.9380969692963612,
-        "support": 30257.0
+        "f1-score": 0.945174330208011,
+        "precision": 0.9451243847979119,
+        "recall": 0.9453101437619382,
+        "support": 29841.0
       },
       "step": 1215
     },
     {
       "epoch": 16.0,
       "eval_B": {
-        "f1-score": 0.8840336134453782,
-        "precision": 0.8752079866888519,
-        "recall": 0.8930390492359932,
-        "support": 1178.0
+        "f1-score": 0.9022614028363357,
+        "precision": 0.895738203957382,
+        "recall": 0.9088803088803089,
+        "support": 1295.0
       },
       "eval_I": {
-        "f1-score": 0.9589266737513285,
-        "precision": 0.9629688917346994,
-        "recall": 0.9549182496428382,
-        "support": 18899.0
+        "f1-score": 0.9632045864531594,
+        "precision": 0.9552494853445741,
+        "recall": 0.9712932967854473,
+        "support": 20065.0
       },
       "eval_O": {
-        "f1-score": 0.923489801893237,
-        "precision": 0.9174907892185379,
-        "recall": 0.9295677799607073,
-        "support": 10180.0
+        "f1-score": 0.9083463808262073,
+        "precision": 0.9282461538461538,
+        "recall": 0.8892819243013795,
+        "support": 8481.0
       },
-      "eval_accuracy": 0.9439799054764186,
-      "eval_loss": 0.3447544574737549,
+      "eval_accuracy": 0.9452766328206159,
+      "eval_loss": 0.44052162766456604,
       "eval_macro avg": {
-        "f1-score": 0.9221500296966479,
-        "precision": 0.9185558892140296,
-        "recall": 0.9258416929465129,
-        "support": 30257.0
-      },
-      "eval_runtime": 4.7359,
-      "eval_samples_per_second": 17.103,
-      "eval_steps_per_second": 2.323,
+        "f1-score": 0.9246041233719008,
+        "precision": 0.92641128104937,
+        "recall": 0.9231518433223785,
+        "support": 29841.0
+      },
+      "eval_runtime": 4.6376,
+      "eval_samples_per_second": 17.25,
+      "eval_steps_per_second": 2.156,
       "eval_weighted avg": {
-        "f1-score": 0.9440880783665984,
-        "precision": 0.9442509279657028,
-        "recall": 0.9439799054764186,
-        "support": 30257.0
+        "f1-score": 0.9449688080038458,
+        "precision": 0.944992377210312,
+        "recall": 0.9452766328206159,
+        "support": 29841.0
       },
       "step": 1296
     },
     {
       "epoch": 17.0,
       "eval_B": {
-        "f1-score": 0.8751040799333889,
-        "precision": 0.8586601307189542,
-        "recall": 0.8921901528013583,
-        "support": 1178.0
+        "f1-score": 0.8923649906890131,
+        "precision": 0.8618705035971223,
+        "recall": 0.9250965250965251,
+        "support": 1295.0
       },
       "eval_I": {
-        "f1-score": 0.9534606521336053,
-        "precision": 0.954598493688342,
-        "recall": 0.9523255198687761,
-        "support": 18899.0
+        "f1-score": 0.9599605522682445,
+        "precision": 0.9498902171261283,
+        "recall": 0.97024669823075,
+        "support": 20065.0
       },
       "eval_O": {
-        "f1-score": 0.9118326047448303,
-        "precision": 0.9118773946360154,
-        "recall": 0.9117878192534381,
-        "support": 10180.0
+        "f1-score": 0.9001642635517431,
+        "precision": 0.9298642533936652,
+        "recall": 0.8723027944817828,
+        "support": 8481.0
       },
-      "eval_accuracy": 0.9363453085236474,
-      "eval_loss": 0.3998367488384247,
+      "eval_accuracy": 0.9404510572701987,
+      "eval_loss": 0.4657233655452728,
       "eval_macro avg": {
-        "f1-score": 0.9134657789372748,
-        "precision": 0.9083786730144373,
-        "recall": 0.9187678306411908,
-        "support": 30257.0
-      },
-      "eval_runtime": 4.7114,
-      "eval_samples_per_second": 17.192,
-      "eval_steps_per_second": 2.335,
+        "f1-score": 0.9174966021696669,
+        "precision": 0.9138749913723053,
+        "recall": 0.9225486726030193,
+        "support": 29841.0
+      },
+      "eval_runtime": 4.6396,
+      "eval_samples_per_second": 17.243,
+      "eval_steps_per_second": 2.155,
       "eval_weighted avg": {
-        "f1-score": 0.9364041837306049,
-        "precision": 0.936489752572877,
-        "recall": 0.9363453085236474,
-        "support": 30257.0
+        "f1-score": 0.940032648483192,
+        "precision": 0.9403789498282803,
+        "recall": 0.9404510572701987,
+        "support": 29841.0
       },
       "step": 1377
     },
     {
       "epoch": 18.0,
       "eval_B": {
-        "f1-score": 0.8807644370585792,
-        "precision": 0.8624898291293734,
-        "recall": 0.8998302207130731,
-        "support": 1178.0
+        "f1-score": 0.9014302280633939,
+        "precision": 0.9024767801857585,
+        "recall": 0.9003861003861003,
+        "support": 1295.0
       },
       "eval_I": {
-        "f1-score": 0.958089514470979,
-        "precision": 0.9545669415410474,
-        "recall": 0.961638181914387,
-        "support": 18899.0
+        "f1-score": 0.9638021281301453,
+        "precision": 0.9506496024820632,
+        "recall": 0.9773236979815599,
+        "support": 20065.0
       },
       "eval_O": {
-        "f1-score": 0.9206207546234321,
-        "precision": 0.9294223646010612,
-        "recall": 0.9119842829076621,
-        "support": 10180.0
+        "f1-score": 0.9083038653822705,
+        "precision": 0.9404115641964399,
+        "recall": 0.87831623629289,
+        "support": 8481.0
       },
-      "eval_accuracy": 0.9425256965330336,
-      "eval_loss": 0.3564019203186035,
+      "eval_accuracy": 0.9458463188230958,
+      "eval_loss": 0.4731404185295105,
       "eval_macro avg": {
-        "f1-score": 0.9198249020509968,
-        "precision": 0.915493045090494,
-        "recall": 0.9244842285117074,
-        "support": 30257.0
-      },
-      "eval_runtime": 4.6986,
-      "eval_samples_per_second": 17.239,
-      "eval_steps_per_second": 2.341,
+        "f1-score": 0.9245120738586032,
+        "precision": 0.9311793156214204,
+        "recall": 0.9186753448868501,
+        "support": 29841.0
+      },
+      "eval_runtime": 4.6405,
+      "eval_samples_per_second": 17.24,
+      "eval_steps_per_second": 2.155,
       "eval_weighted avg": {
-        "f1-score": 0.9424726021386316,
-        "precision": 0.9425221706890129,
-        "recall": 0.9425256965330336,
-        "support": 30257.0
+        "f1-score": 0.9453224398840687,
+        "precision": 0.9456493475450943,
+        "recall": 0.9458463188230958,
+        "support": 29841.0
       },
       "step": 1458
     },
     {
       "epoch": 18.52,
-      "grad_norm": 0.01765240728855133,
+      "grad_norm": 1.548362135887146,
       "learning_rate": 1.2592592592592593e-05,
-      "loss": 0.0067,
+      "loss": 0.0038,
       "step": 1500
     },
     {
       "epoch": 19.0,
       "eval_B": {
-        "f1-score": 0.8756097560975611,
-        "precision": 0.8400936037441498,
-        "recall": 0.9142614601018676,
-        "support": 1178.0
+        "f1-score": 0.8891402714932127,
+        "precision": 0.868828297715549,
+        "recall": 0.9104247104247104,
+        "support": 1295.0
       },
       "eval_I": {
-        "f1-score": 0.9563048159235334,
-        "precision": 0.9492258770786633,
-        "recall": 0.9634901317530028,
-        "support": 18899.0
+        "f1-score": 0.9603099696984749,
+        "precision": 0.9571718572065158,
+        "recall": 0.9634687266384251,
+        "support": 20065.0
       },
       "eval_O": {
-        "f1-score": 0.9161826557180053,
-        "precision": 0.9343341503267973,
-        "recall": 0.8987229862475442,
-        "support": 10180.0
+        "f1-score": 0.9027910305343512,
+        "precision": 0.913358271992277,
+        "recall": 0.8924655111425539,
+        "support": 8481.0
       },
-      "eval_accuracy": 0.9397825296625574,
-      "eval_loss": 0.41976475715637207,
+      "eval_accuracy": 0.9409872323313562,
+      "eval_loss": 0.49517783522605896,
       "eval_macro avg": {
-        "f1-score": 0.9160324092463666,
-        "precision": 0.9078845437165368,
-        "recall": 0.9254915260341382,
-        "support": 30257.0
-      },
-      "eval_runtime": 4.7401,
-      "eval_samples_per_second": 17.088,
-      "eval_steps_per_second": 2.321,
+        "f1-score": 0.9174137572420129,
+        "precision": 0.9131194756381139,
+        "recall": 0.9221196494018965,
+        "support": 29841.0
+      },
+      "eval_runtime": 4.6377,
+      "eval_samples_per_second": 17.25,
+      "eval_steps_per_second": 2.156,
       "eval_weighted avg": {
-        "f1-score": 0.9396639602085823,
-        "precision": 0.9399666776761432,
-        "recall": 0.9397825296625574,
-        "support": 30257.0
+        "f1-score": 0.9408741973642117,
+        "precision": 0.9408859443435835,
+        "recall": 0.9409872323313562,
+        "support": 29841.0
       },
       "step": 1539
     },
     {
       "epoch": 20.0,
       "eval_B": {
-        "f1-score": 0.8851099128992119,
-        "precision": 0.8653690186536902,
-        "recall": 0.9057724957555179,
-        "support": 1178.0
+        "f1-score": 0.899884925201381,
+        "precision": 0.8940548780487805,
+        "recall": 0.9057915057915058,
+        "support": 1295.0
       },
       "eval_I": {
-        "f1-score": 0.9574706357267495,
-        "precision": 0.9427633603446507,
-        "recall": 0.972644055241018,
-        "support": 18899.0
+        "f1-score": 0.9634953846153845,
+        "precision": 0.951896887159533,
+        "recall": 0.9753800149514079,
+        "support": 20065.0
       },
       "eval_O": {
-        "f1-score": 0.9156602050137015,
-        "precision": 0.9470921688011757,
-        "recall": 0.8862475442043222,
-        "support": 10180.0
+        "f1-score": 0.9073556231003039,
+        "precision": 0.9365039528171665,
+        "recall": 0.8799669850253508,
+        "support": 8481.0
       },
-      "eval_accuracy": 0.9409723369798725,
-      "eval_loss": 0.4046960175037384,
+      "eval_accuracy": 0.9452431218792936,
+      "eval_loss": 0.45907479524612427,
       "eval_macro avg": {
-        "f1-score": 0.9194135845465543,
-        "precision": 0.9184081825998388,
-        "recall": 0.921554698400286,
-        "support": 30257.0
-      },
-      "eval_runtime": 4.6993,
-      "eval_samples_per_second": 17.237,
-      "eval_steps_per_second": 2.341,
+        "f1-score": 0.9235786443056897,
+        "precision": 0.9274852393418267,
+        "recall": 0.9203795019227549,
+        "support": 29841.0
+      },
+      "eval_runtime": 4.6356,
+      "eval_samples_per_second": 17.258,
+      "eval_steps_per_second": 2.157,
       "eval_weighted avg": {
-        "f1-score": 0.9405862414989786,
-        "precision": 0.9412065878812694,
-        "recall": 0.9409723369798725,
-        "support": 30257.0
+        "f1-score": 0.9447796290324438,
+        "precision": 0.9450119678218422,
+        "recall": 0.9452431218792936,
+        "support": 29841.0
       },
       "step": 1620
     },
     {
       "epoch": 21.0,
       "eval_B": {
-        "f1-score": 0.8877928483353884,
-        "precision": 0.8605577689243028,
-        "recall": 0.9168081494057725,
-        "support": 1178.0
+        "f1-score": 0.8944337811900192,
+        "precision": 0.8893129770992366,
+        "recall": 0.8996138996138996,
+        "support": 1295.0
       },
       "eval_I": {
-        "f1-score": 0.9589888818963709,
-        "precision": 0.9505640172584082,
-        "recall": 0.9675644213979576,
-        "support": 18899.0
+        "f1-score": 0.9624703322784811,
+        "precision": 0.9549624687239366,
+        "recall": 0.9700971841515076,
+        "support": 20065.0
       },
       "eval_O": {
-        "f1-score": 0.9206317372775131,
-        "precision": 0.9401945724526369,
-        "recall": 0.9018664047151277,
-        "support": 10180.0
+        "f1-score": 0.9066089361958025,
+        "precision": 0.9251350024545901,
+        "recall": 0.8888102818063908,
+        "support": 8481.0
       },
-      "eval_accuracy": 0.9434841524275375,
-      "eval_loss": 0.3954521417617798,
+      "eval_accuracy": 0.9439361951677223,
+      "eval_loss": 0.4614206850528717,
       "eval_macro avg": {
-        "f1-score": 0.9224711558364241,
-        "precision": 0.9171054528784492,
-        "recall": 0.9287463251729525,
-        "support": 30257.0
-      },
-      "eval_runtime": 4.703,
-      "eval_samples_per_second": 17.223,
-      "eval_steps_per_second": 2.339,
+        "f1-score": 0.9211710165547675,
+        "precision": 0.9231368160925877,
+        "recall": 0.9195071218572659,
+        "support": 29841.0
+      },
+      "eval_runtime": 4.6636,
+      "eval_samples_per_second": 17.154,
+      "eval_steps_per_second": 2.144,
       "eval_weighted avg": {
-        "f1-score": 0.9433116944767718,
-        "precision": 0.9435709806500092,
-        "recall": 0.9434841524275375,
-        "support": 30257.0
+        "f1-score": 0.9436416122678664,
+        "precision": 0.9436363458364894,
+        "recall": 0.9439361951677223,
+        "support": 29841.0
       },
       "step": 1701
     },
     {
       "epoch": 22.0,
       "eval_B": {
-        "f1-score": 0.882280049566295,
-        "precision": 0.8592115848753017,
-        "recall": 0.9066213921901528,
-        "support": 1178.0
+        "f1-score": 0.9011072928598702,
+        "precision": 0.8912386706948641,
+        "recall": 0.9111969111969112,
+        "support": 1295.0
       },
       "eval_I": {
-        "f1-score": 0.955072275035812,
-        "precision": 0.9404493229380386,
-        "recall": 0.9701571511720197,
-        "support": 18899.0
+        "f1-score": 0.9635141147970535,
+        "precision": 0.9558585511795576,
+        "recall": 0.9712932967854473,
+        "support": 20065.0
       },
       "eval_O": {
-        "f1-score": 0.9106508274951772,
-        "precision": 0.9423198150872032,
-        "recall": 0.881041257367387,
-        "support": 10180.0
+        "f1-score": 0.9095068938527304,
+        "precision": 0.9292568897637795,
+        "recall": 0.8905789411625987,
+        "support": 8481.0
       },
-      "eval_accuracy": 0.9377003668572562,
-      "eval_loss": 0.45334869623184204,
+      "eval_accuracy": 0.9457457859991287,
+      "eval_loss": 0.4724768102169037,
       "eval_macro avg": {
-        "f1-score": 0.9160010506990948,
-        "precision": 0.9139935743001812,
-        "recall": 0.9192732669098532,
-        "support": 30257.0
-      },
-      "eval_runtime": 4.7089,
-      "eval_samples_per_second": 17.201,
-      "eval_steps_per_second": 2.336,
+        "f1-score": 0.9247094338365515,
+        "precision": 0.9254513705460671,
+        "recall": 0.9243563830483191,
+        "support": 29841.0
+      },
+      "eval_runtime": 4.6333,
+      "eval_samples_per_second": 17.266,
+      "eval_steps_per_second": 2.158,
       "eval_weighted avg": {
-        "f1-score": 0.9372926016522395,
-        "precision": 0.9379158118378169,
-        "recall": 0.9377003668572562,
-        "support": 30257.0
+        "f1-score": 0.9454566745223489,
+        "precision": 0.9454939040264833,
+        "recall": 0.9457457859991287,
+        "support": 29841.0
       },
       "step": 1782
     },
     {
       "epoch": 23.0,
       "eval_B": {
-        "f1-score": 0.8881632653061224,
-        "precision": 0.8553459119496856,
-        "recall": 0.9235993208828522,
-        "support": 1178.0
+        "f1-score": 0.8956916099773242,
+        "precision": 0.8771280532938565,
+        "recall": 0.915057915057915,
+        "support": 1295.0
       },
       "eval_I": {
-        "f1-score": 0.9594668757350059,
-        "precision": 0.947898378601673,
-        "recall": 0.9713212339277211,
-        "support": 18899.0
+        "f1-score": 0.9624542808091362,
+        "precision": 0.9609957269204015,
+        "recall": 0.9639172688761525,
+        "support": 20065.0
       },
       "eval_O": {
-        "f1-score": 0.9211576342239507,
-        "precision": 0.9480195446512112,
-        "recall": 0.8957760314341847,
-        "support": 10180.0
+        "f1-score": 0.9084001187295934,
+        "precision": 0.914753706360593,
+        "recall": 0.9021341822898243,
+        "support": 8481.0
       },
-      "eval_accuracy": 0.9440460058829362,
-      "eval_loss": 0.43748563528060913,
+      "eval_accuracy": 0.9442377936396233,
+      "eval_loss": 0.47468113899230957,
       "eval_macro avg": {
-        "f1-score": 0.9229292584216929,
-        "precision": 0.9170879450675233,
-        "recall": 0.9302321954149194,
-        "support": 30257.0
-      },
-      "eval_runtime": 4.714,
-      "eval_samples_per_second": 17.183,
-      "eval_steps_per_second": 2.333,
+        "f1-score": 0.9221820031720179,
+        "precision": 0.9176258288582838,
+        "recall": 0.927036455407964,
+        "support": 29841.0
+      },
+      "eval_runtime": 4.6476,
+      "eval_samples_per_second": 17.213,
+      "eval_steps_per_second": 2.152,
       "eval_weighted avg": {
-        "f1-score": 0.9438016170620454,
-        "precision": 0.9443357869590202,
-        "recall": 0.9440460058829362,
-        "support": 30257.0
+        "f1-score": 0.9441944702356367,
+        "precision": 0.9442138759866489,
+        "recall": 0.9442377936396233,
+        "support": 29841.0
       },
       "step": 1863
     },
     {
       "epoch": 24.0,
       "eval_B": {
-        "f1-score": 0.8849630238290879,
-        "precision": 0.857484076433121,
-        "recall": 0.9142614601018676,
-        "support": 1178.0
+        "f1-score": 0.9008390541571321,
+        "precision": 0.8899773926149209,
+        "recall": 0.911969111969112,
+        "support": 1295.0
       },
       "eval_I": {
-        "f1-score": 0.960634026800146,
-        "precision": 0.9468112441543759,
-        "recall": 0.974866395047357,
-        "support": 18899.0
+        "f1-score": 0.9639726399486381,
+        "precision": 0.9553151918559123,
+        "recall": 0.9727884375778719,
+        "support": 20065.0
       },
       "eval_O": {
-        "f1-score": 0.9227258898691816,
-        "precision": 0.9535736742821211,
-        "recall": 0.8938113948919449,
-        "support": 10180.0
+        "f1-score": 0.9101008271448409,
+        "precision": 0.9325661964860184,
+        "recall": 0.8886923711826435,
+        "support": 8481.0
       },
-      "eval_accuracy": 0.9452358132002512,
-      "eval_loss": 0.4326893985271454,
+      "eval_accuracy": 0.9462484501189639,
+      "eval_loss": 0.4712923467159271,
       "eval_macro avg": {
-        "f1-score": 0.9227743134994718,
-        "precision": 0.9192896649565393,
-        "recall": 0.9276464166803898,
-        "support": 30257.0
+        "f1-score": 0.9249708404168704,
+        "precision": 0.9259529269856172,
+        "recall": 0.9244833069098758,
+        "support": 29841.0
       },
-      "eval_runtime": 4.6967,
-      "eval_samples_per_second": 17.246,
-      "eval_steps_per_second": 2.342,
+      "eval_runtime": 4.6509,
+      "eval_samples_per_second": 17.201,
+      "eval_steps_per_second": 2.15,
       "eval_weighted avg": {
-        "f1-score": 0.9449336838891791,
-        "precision": 0.9456086839245054,
-        "recall": 0.9452358132002512,
-        "support": 30257.0
+        "f1-score": 0.9459221443892064,
+        "precision": 0.9460143413566277,
+        "recall": 0.9462484501189639,
+        "support": 29841.0
       },
       "step": 1944
     },
     {
       "epoch": 24.69,
-      "grad_norm": 0.07635599374771118,
+      "grad_norm": 0.04072243347764015,
       "learning_rate": 1.0123456790123458e-05,
-      "loss": 0.0035,
+      "loss": 0.0014,
       "step": 2000
     },
     {
       "epoch": 25.0,
       "eval_B": {
-        "f1-score": 0.8899958488999585,
-        "precision": 0.8708367181153533,
-        "recall": 0.9100169779286927,
-        "support": 1178.0
+        "f1-score": 0.8943768996960487,
+        "precision": 0.8803290949887809,
+        "recall": 0.9088803088803089,
+        "support": 1295.0
       },
       "eval_I": {
-        "f1-score": 0.9602969235996759,
-        "precision": 0.9488636363636364,
-        "recall": 0.9720091010106355,
-        "support": 18899.0
+        "f1-score": 0.9629187495336865,
+        "precision": 0.9610305798252582,
+        "recall": 0.9648143533516073,
+        "support": 20065.0
       },
       "eval_O": {
-        "f1-score": 0.92240249924418,
-        "precision": 0.946927374301676,
-        "recall": 0.8991159135559922,
-        "support": 10180.0
+        "f1-score": 0.9089721512974288,
+        "precision": 0.9155502392344498,
+        "recall": 0.9024879141610659,
+        "support": 8481.0
       },
-      "eval_accuracy": 0.9450705621839575,
-      "eval_loss": 0.41382941603660583,
+      "eval_accuracy": 0.9446734358768137,
+      "eval_loss": 0.48565673828125,
       "eval_macro avg": {
-        "f1-score": 0.9242317572479383,
-        "precision": 0.9222092429268886,
-        "recall": 0.9270473308317735,
-        "support": 30257.0
-      },
-      "eval_runtime": 4.755,
-      "eval_samples_per_second": 17.035,
-      "eval_steps_per_second": 2.313,
+        "f1-score": 0.922089266842388,
+        "precision": 0.9189699713494962,
+        "recall": 0.925394192130994,
+        "support": 29841.0
+      },
+      "eval_runtime": 4.6314,
+      "eval_samples_per_second": 17.274,
+      "eval_steps_per_second": 2.159,
       "eval_weighted avg": {
-        "f1-score": 0.9448102624655512,
-        "precision": 0.9451743460345479,
-        "recall": 0.9450705621839575,
-        "support": 30257.0
+        "f1-score": 0.9446122988389564,
+        "precision": 0.944602605179171,
+        "recall": 0.9446734358768137,
+        "support": 29841.0
       },
       "step": 2025
     },
     {
       "epoch": 26.0,
       "eval_B": {
-        "f1-score": 0.885489871847871,
-        "precision": 0.863013698630137,
-        "recall": 0.9091680814940577,
-        "support": 1178.0
+        "f1-score": 0.897679726131609,
+        "precision": 0.8845577211394303,
+        "recall": 0.9111969111969112,
+        "support": 1295.0
       },
       "eval_I": {
-        "f1-score": 0.9624506968183014,
-        "precision": 0.9566149181955987,
-        "recall": 0.9683581141859358,
-        "support": 18899.0
+        "f1-score": 0.962990450744644,
+        "precision": 0.9560839023431743,
+        "recall": 0.9699975080986792,
+        "support": 20065.0
       },
       "eval_O": {
-        "f1-score": 0.9278843757787192,
-        "precision": 0.9417298937784522,
-        "recall": 0.9144400785854617,
-        "support": 10180.0
+        "f1-score": 0.9077024833142927,
+        "precision": 0.9261349693251534,
+        "recall": 0.8899893880438627,
+        "support": 8481.0
       },
-      "eval_accuracy": 0.9479128796642099,
-      "eval_loss": 0.37765103578567505,
+      "eval_accuracy": 0.9447069468181362,
+      "eval_loss": 0.4857862889766693,
       "eval_macro avg": {
-        "f1-score": 0.9252749814816305,
-        "precision": 0.9204528368680626,
-        "recall": 0.9306554247551517,
-        "support": 30257.0
-      },
-      "eval_runtime": 4.7261,
-      "eval_samples_per_second": 17.139,
-      "eval_steps_per_second": 2.327,
+        "f1-score": 0.9227908867301818,
+        "precision": 0.9222588642692527,
+        "recall": 0.9237279357798177,
+        "support": 29841.0
+      },
+      "eval_runtime": 4.7085,
+      "eval_samples_per_second": 16.991,
+      "eval_steps_per_second": 2.124,
       "eval_weighted avg": {
-        "f1-score": 0.9478244946172202,
-        "precision": 0.9479626464827831,
-        "recall": 0.9479128796642099,
-        "support": 30257.0
+        "f1-score": 0.94444299455515,
+        "precision": 0.9444682290887698,
+        "recall": 0.9447069468181362,
+        "support": 29841.0
       },
       "step": 2106
     },
     {
       "epoch": 27.0,
       "eval_B": {
-        "f1-score": 0.8870500620604055,
-        "precision": 0.8652138821630347,
-        "recall": 0.9100169779286927,
-        "support": 1178.0
+        "f1-score": 0.8986280487804879,
+        "precision": 0.8871331828442438,
+        "recall": 0.9104247104247104,
+        "support": 1295.0
       },
       "eval_I": {
-        "f1-score": 0.9591804653112554,
-        "precision": 0.949971456743993,
-        "recall": 0.9685697655960633,
-        "support": 18899.0
+        "f1-score": 0.9620497586334943,
+        "precision": 0.9557796360059027,
+        "recall": 0.9684026912534264,
+        "support": 20065.0
       },
       "eval_O": {
-        "f1-score": 0.9205680164584273,
-        "precision": 0.940917017129962,
-        "recall": 0.9010805500982318,
-        "support": 10180.0
+        "f1-score": 0.9053591790193843,
+        "precision": 0.9219017355169885,
+        "recall": 0.8893998349251268,
+        "support": 8481.0
       },
-      "eval_accuracy": 0.9435833030373136,
-      "eval_loss": 0.44144001603126526,
+      "eval_accuracy": 0.9434335310478872,
+      "eval_loss": 0.4996240735054016,
       "eval_macro avg": {
-        "f1-score": 0.922266181276696,
-        "precision": 0.9187007853456634,
-        "recall": 0.926555764540996,
-        "support": 30257.0
-      },
-      "eval_runtime": 4.7203,
-      "eval_samples_per_second": 17.16,
-      "eval_steps_per_second": 2.33,
+        "f1-score": 0.9220123288111223,
+        "precision": 0.9216048514557116,
+        "recall": 0.9227424122010879,
+        "support": 29841.0
+      },
+      "eval_runtime": 4.6631,
+      "eval_samples_per_second": 17.156,
+      "eval_steps_per_second": 2.145,
       "eval_weighted avg": {
-        "f1-score": 0.9433810025637493,
-        "precision": 0.9436252023854246,
-        "recall": 0.9435833030373136,
-        "support": 30257.0
+        "f1-score": 0.9431856481825407,
+        "precision": 0.9431722960745722,
+        "recall": 0.9434335310478872,
+        "support": 29841.0
       },
       "step": 2187
     },
     {
       "epoch": 28.0,
       "eval_B": {
-        "f1-score": 0.8864954432477217,
-        "precision": 0.8656957928802589,
-        "recall": 0.9083191850594228,
-        "support": 1178.0
+        "f1-score": 0.8919434898816342,
+        "precision": 0.8821752265861027,
+        "recall": 0.901930501930502,
+        "support": 1295.0
       },
       "eval_I": {
-        "f1-score": 0.9565172351280657,
-        "precision": 0.9367454600791316,
-        "recall": 0.9771416477062278,
-        "support": 18899.0
+        "f1-score": 0.9604717405427016,
+        "precision": 0.9630706017938568,
+        "recall": 0.9578868676800398,
+        "support": 20065.0
       },
       "eval_O": {
-        "f1-score": 0.912197875506748,
-        "precision": 0.9549801224884495,
-        "recall": 0.8730844793713163,
-        "support": 10180.0
+        "f1-score": 0.9045243823719267,
+        "precision": 0.9003504672897197,
+        "recall": 0.9087371772196675,
+        "support": 8481.0
       },
-      "eval_accuracy": 0.9394520276299699,
-      "eval_loss": 0.5108960270881653,
+      "eval_accuracy": 0.9414898964511913,
+      "eval_loss": 0.4974842071533203,
       "eval_macro avg": {
-        "f1-score": 0.9184035179608451,
-        "precision": 0.9191404584826133,
-        "recall": 0.9195151040456556,
-        "support": 30257.0
-      },
-      "eval_runtime": 4.736,
-      "eval_samples_per_second": 17.103,
-      "eval_steps_per_second": 2.323,
+        "f1-score": 0.9189798709320876,
+        "precision": 0.9151987652232264,
+        "recall": 0.9228515156100698,
+        "support": 29841.0
+      },
+      "eval_runtime": 4.6318,
+      "eval_samples_per_second": 17.272,
+      "eval_steps_per_second": 2.159,
       "eval_weighted avg": {
-        "f1-score": 0.9388797710113306,
-        "precision": 0.9401143451426405,
-        "recall": 0.9394520276299699,
-        "support": 30257.0
+        "f1-score": 0.941597251442054,
+        "precision": 0.9417345550251954,
+        "recall": 0.9414898964511913,
+        "support": 29841.0
       },
       "step": 2268
     },
     {
       "epoch": 29.0,
       "eval_B": {
-        "f1-score": 0.8870292887029289,
-        "precision": 0.8745874587458746,
-        "recall": 0.8998302207130731,
-        "support": 1178.0
+        "f1-score": 0.9026415094339624,
+        "precision": 0.8826568265682657,
+        "recall": 0.9235521235521236,
+        "support": 1295.0
       },
       "eval_I": {
-        "f1-score": 0.9617674295541738,
-        "precision": 0.9618437764606266,
-        "recall": 0.9616910947669188,
-        "support": 18899.0
+        "f1-score": 0.963459587838005,
+        "precision": 0.9565730005895068,
+        "recall": 0.9704460503364066,
+        "support": 20065.0
       },
       "eval_O": {
-        "f1-score": 0.9281322249003886,
-        "precision": 0.9295497093309686,
-        "recall": 0.9267190569744598,
-        "support": 10180.0
+        "f1-score": 0.9098789958461262,
+        "precision": 0.929520295202952,
+        "recall": 0.8910505836575876,
+        "support": 8481.0
       },
-      "eval_accuracy": 0.9475162772251049,
-      "eval_loss": 0.41041621565818787,
+      "eval_accuracy": 0.9458463188230958,
+      "eval_loss": 0.4864692687988281,
       "eval_macro avg": {
-        "f1-score": 0.9256429810524972,
-        "precision": 0.9219936481791565,
-        "recall": 0.9294134574848173,
-        "support": 30257.0
-      },
-      "eval_runtime": 4.7029,
-      "eval_samples_per_second": 17.224,
-      "eval_steps_per_second": 2.339,
+        "f1-score": 0.9253266977060312,
+        "precision": 0.9229167074535748,
+        "recall": 0.9283495858487059,
+        "support": 29841.0
+      },
+      "eval_runtime": 4.6138,
+      "eval_samples_per_second": 17.339,
+      "eval_steps_per_second": 2.167,
       "eval_weighted avg": {
-        "f1-score": 0.94754103852736,
-        "precision": 0.9475812406623685,
-        "recall": 0.9475162772251049,
-        "support": 30257.0
+        "f1-score": 0.9455923443737659,
+        "precision": 0.9456767357277099,
+        "recall": 0.9458463188230958,
+        "support": 29841.0
       },
       "step": 2349
     }
@@ -1173,7 +1173,7 @@
   "num_input_tokens_seen": 0,
   "num_train_epochs": 50,
   "save_steps": 500,
-  "total_flos": 4156863199738200.0,
+  "total_flos": 4169812929332400.0,
   "train_batch_size": 4,
   "trial_name": null,
   "trial_params": null

checkpoint-2349/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a9f81281de988dbf2616f1c96e725c00f6e4415b289b670a7cb21136eb66e95f
+oid sha256:d229d7bc9127d222c880edfd3fa9418a9ccede656fe27f7f0aa04990c9605be4
 size 4856