Training in progress, epoch 1, checkpoint

last-checkpoint/model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:bc57420a3f9426bcaf177fea89f587ffe199692cb6b6daa1a7c14992d6327cc1
+oid sha256:ca3a26cb4f2b02fdf78505dcba360fa41760b9d0f1d205bee72034eb88ceb909
 size 368871908

last-checkpoint/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:32996fbbfffabbaa0422ca2104929b4ab522fdedbade44c3c5b8d4666df0d633
+oid sha256:70005fdc1fcc0837ee00cab26f99ba01422caf0930cdb9c70748a1a3ef0046b1
 size 737866955

last-checkpoint/rng_state.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:580fb8da3e81024a65fdeba249660e74f25c2a71a809eebb9ff29aeaf9dd2f66
+oid sha256:309a654c3ca5b1a9a6479d7c9d6a0fc2a1057af3810e209d98a0aba0d920ad55
 size 14645

last-checkpoint/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ed01cb3ead55d9aead6eb70916572ffd5fc6d58c2675a2442748ceab27932dec
+oid sha256:2408b440f534d6bf141e5e3eee030bf035f8ee033aa65c33a5266ae6766544ca
 size 1465

last-checkpoint/trainer_state.json

@@ -2,703 +2,354 @@
   "best_global_step": 2393,
   "best_metric": 0.7003869916491275,
   "best_model_checkpoint": "./results_deberta_framing/checkpoint-2393",
-  "epoch": 2.0,
+  "epoch": 1.0,
   "eval_steps": 500,
-  "global_step": 4786,
+  "global_step": 2393,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
       "epoch": 0.020894274968658588,
-      "grad_norm": 2.328125,
+      "grad_norm": 2.255859375,
       "learning_rate": 1.1368909512761022e-06,
-      "loss": 0.6834236145019531,
+      "loss": 0.6489646911621094,
       "step": 50
     },
     {
       "epoch": 0.041788549937317176,
-      "grad_norm": 5.69140625,
+      "grad_norm": 3.923828125,
       "learning_rate": 2.296983758700696e-06,
-      "loss": 0.4080801773071289,
+      "loss": 0.6197348022460938,
       "step": 100
     },
     {
       "epoch": 0.06268282490597576,
-      "grad_norm": 2.078125,
+      "grad_norm": 3.962890625,
       "learning_rate": 3.4570765661252904e-06,
-      "loss": 0.6873448181152344,
+      "loss": 0.6036038208007812,
       "step": 150
     },
     {
       "epoch": 0.08357709987463435,
-      "grad_norm": 1.177734375,
+      "grad_norm": 8.7890625,
       "learning_rate": 4.617169373549885e-06,
-      "loss": 0.7146883392333985,
+      "loss": 0.7426194000244141,
       "step": 200
     },
     {
       "epoch": 0.10447137484329294,
-      "grad_norm": 2.884765625,
+      "grad_norm": 48.9375,
       "learning_rate": 5.7772621809744785e-06,
-      "loss": 0.7065704345703125,
+      "loss": 0.8149568939208984,
       "step": 250
     },
     {
       "epoch": 0.12536564981195153,
-      "grad_norm": 0.93994140625,
+      "grad_norm": 2.2421875,
       "learning_rate": 6.937354988399072e-06,
-      "loss": 0.7347864532470703,
+      "loss": 0.7858116912841797,
       "step": 300
     },
     {
       "epoch": 0.14625992478061012,
-      "grad_norm": 1.400390625,
+      "grad_norm": 3.033203125,
       "learning_rate": 8.097447795823666e-06,
-      "loss": 0.7101815032958985,
+      "loss": 0.7603302764892578,
       "step": 350
     },
     {
       "epoch": 0.1671541997492687,
-      "grad_norm": 6.38671875,
+      "grad_norm": 3.1796875,
       "learning_rate": 9.257540603248261e-06,
-      "loss": 0.8512984466552734,
+      "loss": 0.7251170349121093,
       "step": 400
     },
     {
       "epoch": 0.1880484747179273,
-      "grad_norm": 20.890625,
+      "grad_norm": 1.8046875,
       "learning_rate": 9.973325429756967e-06,
-      "loss": 0.7889546203613281,
+      "loss": 0.716187973022461,
       "step": 450
     },
     {
       "epoch": 0.20894274968658588,
-      "grad_norm": 9.4921875,
+      "grad_norm": 4.421875,
       "learning_rate": 9.899229401304091e-06,
-      "loss": 0.7517801666259766,
+      "loss": 0.6938434600830078,
       "step": 500
     },
     {
       "epoch": 0.22983702465524447,
-      "grad_norm": 17.796875,
+      "grad_norm": 1.8232421875,
       "learning_rate": 9.825133372851215e-06,
-      "loss": 0.7498007965087891,
+      "loss": 0.6959276580810547,
       "step": 550
     },
     {
       "epoch": 0.25073129962390306,
-      "grad_norm": 6.5703125,
+      "grad_norm": 0.84716796875,
       "learning_rate": 9.751037344398341e-06,
-      "loss": 0.7610357666015625,
+      "loss": 0.696595458984375,
       "step": 600
     },
     {
       "epoch": 0.27162557459256165,
-      "grad_norm": 20.53125,
+      "grad_norm": 2.201171875,
       "learning_rate": 9.676941315945465e-06,
-      "loss": 0.7625791931152344,
+      "loss": 0.7023362731933593,
       "step": 650
     },
     {
       "epoch": 0.29251984956122024,
-      "grad_norm": 13.1015625,
+      "grad_norm": 2.84765625,
       "learning_rate": 9.602845287492591e-06,
-      "loss": 0.7331975555419922,
+      "loss": 0.6915609741210937,
       "step": 700
     },
     {
       "epoch": 0.3134141245298788,
-      "grad_norm": 10.71875,
+      "grad_norm": 3.876953125,
       "learning_rate": 9.528749259039716e-06,
-      "loss": 0.7403915405273438,
+      "loss": 0.6972248840332032,
       "step": 750
     },
     {
       "epoch": 0.3343083994985374,
-      "grad_norm": 5.87890625,
+      "grad_norm": 0.5517578125,
       "learning_rate": 9.454653230586842e-06,
-      "loss": 0.7098237609863282,
+      "loss": 0.69171142578125,
       "step": 800
     },
     {
       "epoch": 0.355202674467196,
-      "grad_norm": 4.44921875,
+      "grad_norm": 1.32421875,
       "learning_rate": 9.380557202133966e-06,
-      "loss": 0.7066761779785157,
+      "loss": 0.6903911590576172,
       "step": 850
     },
     {
       "epoch": 0.3760969494358546,
-      "grad_norm": 11.4140625,
+      "grad_norm": 2.158203125,
       "learning_rate": 9.306461173681092e-06,
-      "loss": 0.7206510162353515,
+      "loss": 0.6898154449462891,
       "step": 900
     },
     {
       "epoch": 0.3969912244045132,
-      "grad_norm": 13.9375,
+      "grad_norm": 1.0,
       "learning_rate": 9.232365145228218e-06,
-      "loss": 0.7132095336914063,
+      "loss": 0.6989466094970703,
       "step": 950
     },
     {
       "epoch": 0.41788549937317176,
-      "grad_norm": 17.90625,
+      "grad_norm": 1.8564453125,
       "learning_rate": 9.158269116775342e-06,
-      "loss": 0.7332618713378907,
+      "loss": 0.6927836608886718,
       "step": 1000
     },
     {
       "epoch": 0.43877977434183035,
-      "grad_norm": 9.4921875,
+      "grad_norm": 1.8994140625,
       "learning_rate": 9.084173088322466e-06,
-      "loss": 0.7324075317382812,
+      "loss": 0.6929279327392578,
       "step": 1050
     },
     {
       "epoch": 0.45967404931048894,
-      "grad_norm": 2.833984375,
+      "grad_norm": 1.439453125,
       "learning_rate": 9.010077059869592e-06,
-      "loss": 0.7019334411621094,
+      "loss": 0.6930996704101563,
       "step": 1100
     },
     {
       "epoch": 0.48056832427914753,
-      "grad_norm": 2.498046875,
+      "grad_norm": 0.501953125,
       "learning_rate": 8.935981031416716e-06,
-      "loss": 0.7027674865722656,
+      "loss": 0.6932637786865234,
       "step": 1150
     },
     {
       "epoch": 0.5014625992478061,
-      "grad_norm": 6.46875,
+      "grad_norm": 2.6171875,
       "learning_rate": 8.86188500296384e-06,
-      "loss": 0.6919042205810547,
+      "loss": 0.687666015625,
       "step": 1200
     },
     {
       "epoch": 0.5223568742164647,
-      "grad_norm": 1.6796875,
+      "grad_norm": 0.8564453125,
       "learning_rate": 8.787788974510967e-06,
-      "loss": 0.7015120697021484,
+      "loss": 0.6918360137939453,
       "step": 1250
     },
     {
       "epoch": 0.5432511491851233,
-      "grad_norm": 1.26953125,
+      "grad_norm": 0.6552734375,
       "learning_rate": 8.713692946058093e-06,
-      "loss": 0.698077163696289,
+      "loss": 0.6918303680419922,
       "step": 1300
     },
     {
       "epoch": 0.5641454241537819,
-      "grad_norm": 11.375,
+      "grad_norm": 3.671875,
       "learning_rate": 8.639596917605217e-06,
-      "loss": 0.7053836059570312,
+      "loss": 0.6942510223388672,
       "step": 1350
     },
     {
       "epoch": 0.5850396991224405,
-      "grad_norm": 1.7216796875,
+      "grad_norm": 0.357177734375,
       "learning_rate": 8.565500889152341e-06,
-      "loss": 0.6991368865966797,
+      "loss": 0.6894574737548829,
       "step": 1400
     },
     {
       "epoch": 0.6059339740910991,
-      "grad_norm": 4.34765625,
+      "grad_norm": 1.0927734375,
       "learning_rate": 8.491404860699467e-06,
-      "loss": 0.6921279144287109,
+      "loss": 0.6866970825195312,
       "step": 1450
     },
     {
       "epoch": 0.6268282490597576,
-      "grad_norm": 1.767578125,
+      "grad_norm": 0.7958984375,
       "learning_rate": 8.417308832246593e-06,
-      "loss": 0.6929914855957031,
+      "loss": 0.6886749267578125,
       "step": 1500
     },
     {
       "epoch": 0.6477225240284162,
-      "grad_norm": 0.494140625,
+      "grad_norm": 0.447265625,
       "learning_rate": 8.343212803793717e-06,
-      "loss": 0.6972026062011719,
+      "loss": 0.6923796844482422,
       "step": 1550
     },
     {
       "epoch": 0.6686167989970748,
-      "grad_norm": 3.6796875,
+      "grad_norm": 1.814453125,
       "learning_rate": 8.269116775340842e-06,
-      "loss": 0.6942645263671875,
+      "loss": 0.6926873016357422,
       "step": 1600
     },
     {
       "epoch": 0.6895110739657334,
-      "grad_norm": 2.2421875,
+      "grad_norm": 0.85009765625,
       "learning_rate": 8.195020746887967e-06,
-      "loss": 0.688611831665039,
+      "loss": 0.6897069549560547,
       "step": 1650
     },
     {
       "epoch": 0.710405348934392,
-      "grad_norm": 1.4794921875,
+      "grad_norm": 0.4150390625,
       "learning_rate": 8.120924718435092e-06,
-      "loss": 0.6983576965332031,
+      "loss": 0.6941582489013672,
       "step": 1700
     },
     {
       "epoch": 0.7312996239030506,
-      "grad_norm": 1.568359375,
+      "grad_norm": 0.5458984375,
       "learning_rate": 8.046828689982218e-06,
-      "loss": 0.6902479553222656,
+      "loss": 0.6902919769287109,
       "step": 1750
     },
     {
       "epoch": 0.7521938988717092,
-      "grad_norm": 0.7119140625,
+      "grad_norm": 0.78515625,
       "learning_rate": 7.972732661529344e-06,
-      "loss": 0.6989839935302734,
+      "loss": 0.6972850036621093,
       "step": 1800
     },
     {
       "epoch": 0.7730881738403678,
-      "grad_norm": 5.09765625,
+      "grad_norm": 3.0390625,
       "learning_rate": 7.898636633076468e-06,
-      "loss": 0.6913361358642578,
+      "loss": 0.6896543884277344,
       "step": 1850
     },
     {
       "epoch": 0.7939824488090264,
-      "grad_norm": 1.63671875,
+      "grad_norm": 0.80908203125,
       "learning_rate": 7.824540604623592e-06,
-      "loss": 0.6869488525390625,
+      "loss": 0.6867472839355468,
       "step": 1900
     },
     {
       "epoch": 0.8148767237776849,
-      "grad_norm": 3.251953125,
+      "grad_norm": 4.5703125,
       "learning_rate": 7.750444576170718e-06,
-      "loss": 0.6954217529296876,
+      "loss": 0.694358901977539,
       "step": 1950
     },
     {
       "epoch": 0.8357709987463435,
-      "grad_norm": 2.732421875,
+      "grad_norm": 2.62109375,
       "learning_rate": 7.676348547717844e-06,
-      "loss": 0.6931136322021484,
+      "loss": 0.6951927185058594,
       "step": 2000
     },
     {
       "epoch": 0.8566652737150021,
-      "grad_norm": 1.0146484375,
+      "grad_norm": 1.0546875,
       "learning_rate": 7.6022525192649674e-06,
-      "loss": 0.6909190368652344,
+      "loss": 0.6904816436767578,
       "step": 2050
     },
     {
       "epoch": 0.8775595486836607,
-      "grad_norm": 7.515625,
+      "grad_norm": 4.55078125,
       "learning_rate": 7.528156490812093e-06,
-      "loss": 0.6913449859619141,
+      "loss": 0.6920033264160156,
       "step": 2100
     },
     {
       "epoch": 0.8984538236523193,
-      "grad_norm": 1.0673828125,
+      "grad_norm": 0.83544921875,
       "learning_rate": 7.4540604623592185e-06,
-      "loss": 0.6896768188476563,
+      "loss": 0.6891203308105469,
       "step": 2150
     },
     {
       "epoch": 0.9193480986209779,
-      "grad_norm": 1.865234375,
+      "grad_norm": 1.541015625,
       "learning_rate": 7.379964433906344e-06,
-      "loss": 0.6873223876953125,
+      "loss": 0.6877539825439453,
       "step": 2200
     },
     {
       "epoch": 0.9402423735896365,
-      "grad_norm": 2.818359375,
+      "grad_norm": 0.7763671875,
       "learning_rate": 7.305868405453468e-06,
-      "loss": 0.6821249389648437,
+      "loss": 0.6825115203857421,
       "step": 2250
     },
     {
       "epoch": 0.9611366485582951,
-      "grad_norm": 0.9091796875,
+      "grad_norm": 0.3837890625,
       "learning_rate": 7.231772377000593e-06,
-      "loss": 0.6942471313476563,
+      "loss": 0.6943415832519532,
       "step": 2300
     },
     {
       "epoch": 0.9820309235269536,
-      "grad_norm": 0.9287109375,
+      "grad_norm": 1.103515625,
       "learning_rate": 7.157676348547719e-06,
-      "loss": 0.6885271453857422,
+      "loss": 0.6877147674560546,
       "step": 2350
     },
     {
       "epoch": 1.0,
       "eval_accuracy": 0.5389196531187964,
       "eval_f1": 0.7003869916491275,
-      "eval_loss": 0.6903790235519409,
+      "eval_loss": 0.6902970671653748,
       "eval_precision": 0.5389196531187964,
       "eval_recall": 1.0,
       "eval_roc_auc": 0.5,
-      "eval_runtime": 349.2377,
-      "eval_samples_per_second": 27.405,
-      "eval_steps_per_second": 1.715,
+      "eval_runtime": 76.0306,
+      "eval_samples_per_second": 125.883,
+      "eval_steps_per_second": 7.878,
       "step": 2393
-    },
-    {
-      "epoch": 1.0029251984956122,
-      "grad_norm": 2.685546875,
-      "learning_rate": 7.083580320094844e-06,
-      "loss": 0.6907852172851563,
-      "step": 2400
-    },
-    {
-      "epoch": 1.0238194734642707,
-      "grad_norm": 4.87890625,
-      "learning_rate": 7.009484291641968e-06,
-      "loss": 0.6929679870605469,
-      "step": 2450
-    },
-    {
-      "epoch": 1.0447137484329294,
-      "grad_norm": 0.73681640625,
-      "learning_rate": 6.935388263189093e-06,
-      "loss": 0.6920276641845703,
-      "step": 2500
-    },
-    {
-      "epoch": 1.065608023401588,
-      "grad_norm": 1.796875,
-      "learning_rate": 6.8612922347362184e-06,
-      "loss": 0.6908602905273438,
-      "step": 2550
-    },
-    {
-      "epoch": 1.0865022983702466,
-      "grad_norm": 6.328125,
-      "learning_rate": 6.787196206283344e-06,
-      "loss": 0.6924705505371094,
-      "step": 2600
-    },
-    {
-      "epoch": 1.107396573338905,
-      "grad_norm": 0.7470703125,
-      "learning_rate": 6.713100177830469e-06,
-      "loss": 0.6856446838378907,
-      "step": 2650
-    },
-    {
-      "epoch": 1.1282908483075638,
-      "grad_norm": 2.78515625,
-      "learning_rate": 6.639004149377594e-06,
-      "loss": 0.6915918731689453,
-      "step": 2700
-    },
-    {
-      "epoch": 1.1491851232762222,
-      "grad_norm": 0.72802734375,
-      "learning_rate": 6.564908120924719e-06,
-      "loss": 0.6918558502197265,
-      "step": 2750
-    },
-    {
-      "epoch": 1.170079398244881,
-      "grad_norm": 1.701171875,
-      "learning_rate": 6.490812092471844e-06,
-      "loss": 0.6948513793945312,
-      "step": 2800
-    },
-    {
-      "epoch": 1.1909736732135394,
-      "grad_norm": 2.4375,
-      "learning_rate": 6.41671606401897e-06,
-      "loss": 0.6888138580322266,
-      "step": 2850
-    },
-    {
-      "epoch": 1.2118679481821981,
-      "grad_norm": 4.6640625,
-      "learning_rate": 6.342620035566094e-06,
-      "loss": 0.6906478881835938,
-      "step": 2900
-    },
-    {
-      "epoch": 1.2327622231508566,
-      "grad_norm": 2.67578125,
-      "learning_rate": 6.268524007113219e-06,
-      "loss": 0.6887605285644531,
-      "step": 2950
-    },
-    {
-      "epoch": 1.2536564981195153,
-      "grad_norm": 1.1572265625,
-      "learning_rate": 6.194427978660344e-06,
-      "loss": 0.6858110809326172,
-      "step": 3000
-    },
-    {
-      "epoch": 1.2745507730881738,
-      "grad_norm": 1.2041015625,
-      "learning_rate": 6.1203319502074694e-06,
-      "loss": 0.6906262969970703,
-      "step": 3050
-    },
-    {
-      "epoch": 1.2954450480568325,
-      "grad_norm": 3.1953125,
-      "learning_rate": 6.046235921754594e-06,
-      "loss": 0.6898616027832031,
-      "step": 3100
-    },
-    {
-      "epoch": 1.316339323025491,
-      "grad_norm": 0.80615234375,
-      "learning_rate": 5.97213989330172e-06,
-      "loss": 0.6915110778808594,
-      "step": 3150
-    },
-    {
-      "epoch": 1.3372335979941496,
-      "grad_norm": 0.95458984375,
-      "learning_rate": 5.898043864848845e-06,
-      "loss": 0.69228271484375,
-      "step": 3200
-    },
-    {
-      "epoch": 1.3581278729628081,
-      "grad_norm": 6.22265625,
-      "learning_rate": 5.82394783639597e-06,
-      "loss": 0.6870991516113282,
-      "step": 3250
-    },
-    {
-      "epoch": 1.3790221479314668,
-      "grad_norm": 2.3359375,
-      "learning_rate": 5.749851807943094e-06,
-      "loss": 0.6957393646240234,
-      "step": 3300
-    },
-    {
-      "epoch": 1.3999164229001253,
-      "grad_norm": 0.86767578125,
-      "learning_rate": 5.675755779490219e-06,
-      "loss": 0.6911115264892578,
-      "step": 3350
-    },
-    {
-      "epoch": 1.420810697868784,
-      "grad_norm": 1.87109375,
-      "learning_rate": 5.601659751037345e-06,
-      "loss": 0.6849404907226563,
-      "step": 3400
-    },
-    {
-      "epoch": 1.4417049728374425,
-      "grad_norm": 1.0029296875,
-      "learning_rate": 5.52756372258447e-06,
-      "loss": 0.6912786102294922,
-      "step": 3450
-    },
-    {
-      "epoch": 1.4625992478061012,
-      "grad_norm": 1.4169921875,
-      "learning_rate": 5.4534676941315945e-06,
-      "loss": 0.6927188110351562,
-      "step": 3500
-    },
-    {
-      "epoch": 1.4834935227747597,
-      "grad_norm": 2.767578125,
-      "learning_rate": 5.37937166567872e-06,
-      "loss": 0.688505859375,
-      "step": 3550
-    },
-    {
-      "epoch": 1.5043877977434184,
-      "grad_norm": 0.49169921875,
-      "learning_rate": 5.3052756372258455e-06,
-      "loss": 0.6911637878417969,
-      "step": 3600
-    },
-    {
-      "epoch": 1.525282072712077,
-      "grad_norm": 2.513671875,
-      "learning_rate": 5.231179608772971e-06,
-      "loss": 0.6921497344970703,
-      "step": 3650
-    },
-    {
-      "epoch": 1.5461763476807355,
-      "grad_norm": 1.263671875,
-      "learning_rate": 5.157083580320096e-06,
-      "loss": 0.6940757751464843,
-      "step": 3700
-    },
-    {
-      "epoch": 1.567070622649394,
-      "grad_norm": 3.740234375,
-      "learning_rate": 5.08298755186722e-06,
-      "loss": 0.6916998291015625,
-      "step": 3750
-    },
-    {
-      "epoch": 1.5879648976180527,
-      "grad_norm": 1.34765625,
-      "learning_rate": 5.008891523414345e-06,
-      "loss": 0.6906791687011719,
-      "step": 3800
-    },
-    {
-      "epoch": 1.6088591725867114,
-      "grad_norm": 2.775390625,
-      "learning_rate": 4.934795494961471e-06,
-      "loss": 0.6903327941894531,
-      "step": 3850
-    },
-    {
-      "epoch": 1.6297534475553699,
-      "grad_norm": 3.341796875,
-      "learning_rate": 4.860699466508595e-06,
-      "loss": 0.6916295623779297,
-      "step": 3900
-    },
-    {
-      "epoch": 1.6506477225240284,
-      "grad_norm": 1.8291015625,
-      "learning_rate": 4.78660343805572e-06,
-      "loss": 0.6899072265625,
-      "step": 3950
-    },
-    {
-      "epoch": 1.671541997492687,
-      "grad_norm": 2.580078125,
-      "learning_rate": 4.7125074096028455e-06,
-      "loss": 0.6946562194824218,
-      "step": 4000
-    },
-    {
-      "epoch": 1.6924362724613455,
-      "grad_norm": 5.171875,
-      "learning_rate": 4.638411381149971e-06,
-      "loss": 0.692690200805664,
-      "step": 4050
-    },
-    {
-      "epoch": 1.713330547430004,
-      "grad_norm": 0.90478515625,
-      "learning_rate": 4.564315352697096e-06,
-      "loss": 0.6910479736328125,
-      "step": 4100
-    },
-    {
-      "epoch": 1.7342248223986627,
-      "grad_norm": 1.8486328125,
-      "learning_rate": 4.490219324244221e-06,
-      "loss": 0.6891459655761719,
-      "step": 4150
-    },
-    {
-      "epoch": 1.7551190973673214,
-      "grad_norm": 5.5546875,
-      "learning_rate": 4.416123295791346e-06,
-      "loss": 0.6910189056396484,
-      "step": 4200
-    },
-    {
-      "epoch": 1.77601337233598,
-      "grad_norm": 1.09765625,
-      "learning_rate": 4.342027267338471e-06,
-      "loss": 0.6903411865234375,
-      "step": 4250
-    },
-    {
-      "epoch": 1.7969076473046384,
-      "grad_norm": 1.7626953125,
-      "learning_rate": 4.267931238885596e-06,
-      "loss": 0.6911215209960937,
-      "step": 4300
-    },
-    {
-      "epoch": 1.817801922273297,
-      "grad_norm": 7.09765625,
-      "learning_rate": 4.193835210432721e-06,
-      "loss": 0.6897867584228515,
-      "step": 4350
-    },
-    {
-      "epoch": 1.8386961972419558,
-      "grad_norm": 2.671875,
-      "learning_rate": 4.119739181979846e-06,
-      "loss": 0.6900591278076171,
-      "step": 4400
-    },
-    {
-      "epoch": 1.8595904722106142,
-      "grad_norm": 2.34375,
-      "learning_rate": 4.045643153526971e-06,
-      "loss": 0.6915322875976563,
-      "step": 4450
-    },
-    {
-      "epoch": 1.8804847471792727,
-      "grad_norm": 1.58984375,
-      "learning_rate": 3.9715471250740965e-06,
-      "loss": 0.6882122039794922,
-      "step": 4500
-    },
-    {
-      "epoch": 1.9013790221479314,
-      "grad_norm": 1.25,
-      "learning_rate": 3.897451096621222e-06,
-      "loss": 0.6915691375732422,
-      "step": 4550
-    },
-    {
-      "epoch": 1.9222732971165901,
-      "grad_norm": 4.87109375,
-      "learning_rate": 3.823355068168346e-06,
-      "loss": 0.6912251281738281,
-      "step": 4600
-    },
-    {
-      "epoch": 1.9431675720852486,
-      "grad_norm": 6.890625,
-      "learning_rate": 3.7492590397154714e-06,
-      "loss": 0.6866284942626953,
-      "step": 4650
-    },
-    {
-      "epoch": 1.964061847053907,
-      "grad_norm": 3.189453125,
-      "learning_rate": 3.675163011262597e-06,
-      "loss": 0.6898875427246094,
-      "step": 4700
-    },
-    {
-      "epoch": 1.9849561220225658,
-      "grad_norm": 5.70703125,
-      "learning_rate": 3.6010669828097216e-06,
-      "loss": 0.6908906555175781,
-      "step": 4750
-    },
-    {
-      "epoch": 2.0,
-      "eval_accuracy": 0.5389196531187964,
-      "eval_f1": 0.7003869916491275,
-      "eval_loss": 0.6901159882545471,
-      "eval_precision": 0.5389196531187964,
-      "eval_recall": 1.0,
-      "eval_roc_auc": 0.5,
-      "eval_runtime": 347.1335,
-      "eval_samples_per_second": 27.572,
-      "eval_steps_per_second": 1.726,
-      "step": 4786
     }
   ],
   "logging_steps": 50,
@@ -713,7 +364,7 @@
         "early_stopping_threshold": 0.0
       },
       "attributes": {
-        "early_stopping_patience_counter": 1
+        "early_stopping_patience_counter": 0
       }
     },
     "TrainerControl": {
@@ -727,7 +378,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 1.3062259262328e+16,
+  "total_flos": 6539240997132864.0,
   "train_batch_size": 16,
   "trial_name": null,
   "trial_params": null