Training in progress, epoch 1, checkpoint

last-checkpoint/model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ee43cc787e4bbdf3b74113157c9557cc3d4eb0246845be2bc07fb9cb991c0e49
-size 1450506728
+oid sha256:4cac443dc31ef4fa509e8a533d609447e9408e9fcc6a4c34574adc713af26700
+size 1450506320

last-checkpoint/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f08bd408cd0095fb58cfc3775ff2ec9590d9677439cedc7321f7510f80768e38
-size 2901193658
+oid sha256:fc1881d8684e4da3b4d7efe825ccff51224f8ed8c542abc8b927dbc46478c012
+size 2901190266

last-checkpoint/rng_state.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5daace352b191b76e3c99c762305288cdf737ab918956dd1b4d849f72bed01cc
+oid sha256:ef4640a2a0153d48cf229598ea37a7fd35ecf11723df7b18bc022ac74b7b6c01
 size 14244

last-checkpoint/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:56f1d9ce5cbe4f6e6e248b69695a1b0c3f76ed7a81860142c34e6f660cddfc4f
+oid sha256:05e2abcc921b08d0d496adddf795c861e55680b7b20c40929c30d9d2032af4e8
 size 1064

last-checkpoint/trainer_state.json

@@ -1,547 +1,715 @@
 {
-  "best_global_step": 370,
-  "best_metric": 0.3884802460670471,
-  "best_model_checkpoint": "./VulnSentry/checkpoint-370",
+  "best_global_step": 493,
+  "best_metric": 0.4431517720222473,
+  "best_model_checkpoint": "./VulnSentry/checkpoint-493",
   "epoch": 1.0,
   "eval_steps": 500,
-  "global_step": 370,
+  "global_step": 493,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.013544018058690745,
-      "grad_norm": 61.920562744140625,
-      "learning_rate": 4.210526315789474e-06,
-      "loss": 0.7503,
+      "epoch": 0.01015916017609211,
+      "grad_norm": 124.9168472290039,
+      "learning_rate": 1.0810810810810812e-06,
+      "loss": 0.7082,
       "step": 5
     },
     {
-      "epoch": 0.02708803611738149,
-      "grad_norm": 53.86005783081055,
-      "learning_rate": 9.473684210526315e-06,
-      "loss": 0.5629,
+      "epoch": 0.02031832035218422,
+      "grad_norm": 216.048095703125,
+      "learning_rate": 2.432432432432433e-06,
+      "loss": 0.6676,
       "step": 10
     },
     {
-      "epoch": 0.040632054176072234,
-      "grad_norm": 52.824466705322266,
-      "learning_rate": 1.4736842105263159e-05,
-      "loss": 0.5162,
+      "epoch": 0.03047748052827633,
+      "grad_norm": 129.1244354248047,
+      "learning_rate": 3.7837837837837844e-06,
+      "loss": 0.6235,
       "step": 15
     },
     {
-      "epoch": 0.05417607223476298,
-      "grad_norm": 49.85675048828125,
-      "learning_rate": 2e-05,
-      "loss": 0.2927,
+      "epoch": 0.04063664070436844,
+      "grad_norm": 99.75704956054688,
+      "learning_rate": 5.135135135135135e-06,
+      "loss": 0.6237,
       "step": 20
     },
     {
-      "epoch": 0.06772009029345373,
-      "grad_norm": 40.95595169067383,
-      "learning_rate": 1.9999629996754254e-05,
-      "loss": 0.4022,
+      "epoch": 0.05079580088046055,
+      "grad_norm": 99.64442443847656,
+      "learning_rate": 6.486486486486487e-06,
+      "loss": 0.507,
       "step": 25
     },
     {
-      "epoch": 0.08126410835214447,
-      "grad_norm": 40.73978042602539,
-      "learning_rate": 1.9998520014397486e-05,
-      "loss": 0.495,
+      "epoch": 0.06095496105655266,
+      "grad_norm": 130.50547790527344,
+      "learning_rate": 7.837837837837838e-06,
+      "loss": 0.4867,
       "step": 30
     },
     {
-      "epoch": 0.09480812641083522,
-      "grad_norm": 25.020273208618164,
-      "learning_rate": 1.9996670135069123e-05,
-      "loss": 0.4785,
+      "epoch": 0.07111412123264477,
+      "grad_norm": 480.7685241699219,
+      "learning_rate": 9.189189189189191e-06,
+      "loss": 0.4644,
       "step": 35
     },
     {
-      "epoch": 0.10835214446952596,
-      "grad_norm": 26.943941116333008,
-      "learning_rate": 1.999408049566142e-05,
-      "loss": 0.432,
+      "epoch": 0.08127328140873688,
+      "grad_norm": 499.7054748535156,
+      "learning_rate": 1.0540540540540541e-05,
+      "loss": 0.5031,
       "step": 40
     },
     {
-      "epoch": 0.12189616252821671,
-      "grad_norm": 22.082460403442383,
-      "learning_rate": 1.9990751287809393e-05,
-      "loss": 0.3905,
+      "epoch": 0.091432441584829,
+      "grad_norm": 152.89173889160156,
+      "learning_rate": 1.1891891891891894e-05,
+      "loss": 0.4018,
       "step": 45
     },
     {
-      "epoch": 0.13544018058690746,
-      "grad_norm": 26.54736328125,
-      "learning_rate": 1.9986682757876574e-05,
-      "loss": 0.4153,
+      "epoch": 0.1015916017609211,
+      "grad_norm": 134.70472717285156,
+      "learning_rate": 1.3243243243243244e-05,
+      "loss": 0.3914,
       "step": 50
     },
     {
-      "epoch": 0.1489841986455982,
-      "grad_norm": 17.708812713623047,
-      "learning_rate": 1.9981875206936818e-05,
-      "loss": 0.4651,
+      "epoch": 0.11175076193701321,
+      "grad_norm": 124.81039428710938,
+      "learning_rate": 1.4594594594594596e-05,
+      "loss": 0.4259,
       "step": 55
     },
     {
-      "epoch": 0.16252821670428894,
-      "grad_norm": 21.073911666870117,
-      "learning_rate": 1.9976328990752013e-05,
-      "loss": 0.4979,
+      "epoch": 0.12190992211310532,
+      "grad_norm": 417.4250793457031,
+      "learning_rate": 1.5945945945945947e-05,
+      "loss": 0.4314,
       "step": 60
     },
     {
-      "epoch": 0.17607223476297967,
-      "grad_norm": 17.107423782348633,
-      "learning_rate": 1.997004451974577e-05,
-      "loss": 0.3886,
+      "epoch": 0.13206908228919742,
+      "grad_norm": 111.6860580444336,
+      "learning_rate": 1.72972972972973e-05,
+      "loss": 0.3708,
       "step": 65
     },
     {
-      "epoch": 0.18961625282167044,
-      "grad_norm": 18.816625595092773,
-      "learning_rate": 1.9963022258973007e-05,
-      "loss": 0.5016,
+      "epoch": 0.14222824246528953,
+      "grad_norm": 57.337913513183594,
+      "learning_rate": 1.864864864864865e-05,
+      "loss": 0.3717,
       "step": 70
     },
     {
-      "epoch": 0.20316027088036118,
-      "grad_norm": 18.011642456054688,
-      "learning_rate": 1.9955262728085592e-05,
-      "loss": 0.4861,
+      "epoch": 0.15238740264138165,
+      "grad_norm": 82.06193542480469,
+      "learning_rate": 2e-05,
+      "loss": 0.4444,
       "step": 75
     },
     {
-      "epoch": 0.21670428893905191,
-      "grad_norm": 21.518707275390625,
-      "learning_rate": 1.9946766501293838e-05,
-      "loss": 0.4783,
+      "epoch": 0.16254656281747376,
+      "grad_norm": 62.36454772949219,
+      "learning_rate": 1.999978329580869e-05,
+      "loss": 0.4449,
       "step": 80
     },
     {
-      "epoch": 0.23024830699774265,
-      "grad_norm": 14.0570650100708,
-      "learning_rate": 1.9937534207324053e-05,
-      "loss": 0.4393,
+      "epoch": 0.17270572299356587,
+      "grad_norm": 51.05297088623047,
+      "learning_rate": 1.9999133192626893e-05,
+      "loss": 0.4272,
       "step": 85
     },
     {
-      "epoch": 0.24379232505643342,
-      "grad_norm": 11.293530464172363,
-      "learning_rate": 1.992756652937198e-05,
-      "loss": 0.4614,
+      "epoch": 0.182864883169658,
+      "grad_norm": 40.76633834838867,
+      "learning_rate": 1.999804971863063e-05,
+      "loss": 0.3714,
       "step": 90
     },
     {
-      "epoch": 0.25733634311512416,
-      "grad_norm": 10.388118743896484,
-      "learning_rate": 1.991686420505225e-05,
-      "loss": 0.515,
+      "epoch": 0.19302404334575007,
+      "grad_norm": 56.4720344543457,
+      "learning_rate": 1.999653292077857e-05,
+      "loss": 0.4217,
       "step": 95
     },
     {
-      "epoch": 0.2708803611738149,
-      "grad_norm": 9.720160484313965,
-      "learning_rate": 1.990542802634382e-05,
-      "loss": 0.4896,
+      "epoch": 0.2031832035218422,
+      "grad_norm": 59.50137710571289,
+      "learning_rate": 1.9994582864810008e-05,
+      "loss": 0.4571,
       "step": 100
     },
     {
-      "epoch": 0.28442437923250563,
-      "grad_norm": 8.744847297668457,
-      "learning_rate": 1.9893258839531335e-05,
-      "loss": 0.4879,
+      "epoch": 0.2133423636979343,
+      "grad_norm": 35.567649841308594,
+      "learning_rate": 1.9992199635241997e-05,
+      "loss": 0.3531,
       "step": 105
     },
     {
-      "epoch": 0.2979683972911964,
-      "grad_norm": 11.211833953857422,
-      "learning_rate": 1.988035754514252e-05,
-      "loss": 0.4154,
+      "epoch": 0.22350152387402641,
+      "grad_norm": 46.55067825317383,
+      "learning_rate": 1.9989383335365713e-05,
+      "loss": 0.4312,
       "step": 110
     },
     {
-      "epoch": 0.3115124153498871,
-      "grad_norm": 9.039229393005371,
-      "learning_rate": 1.986672509788153e-05,
-      "loss": 0.3759,
+      "epoch": 0.23366068405011853,
+      "grad_norm": 36.95201110839844,
+      "learning_rate": 1.998613408724195e-05,
+      "loss": 0.4454,
       "step": 115
     },
     {
-      "epoch": 0.32505643340857787,
-      "grad_norm": 11.156192779541016,
-      "learning_rate": 1.985236250655832e-05,
-      "loss": 0.4234,
+      "epoch": 0.24381984422621064,
+      "grad_norm": 37.9529914855957,
+      "learning_rate": 1.9982452031695837e-05,
+      "loss": 0.4221,
       "step": 120
     },
     {
-      "epoch": 0.33860045146726864,
-      "grad_norm": 9.05034065246582,
-      "learning_rate": 1.983727083401396e-05,
-      "loss": 0.4786,
+      "epoch": 0.25397900440230275,
+      "grad_norm": 27.23141860961914,
+      "learning_rate": 1.997833732831076e-05,
+      "loss": 0.5048,
       "step": 125
     },
     {
-      "epoch": 0.35214446952595935,
-      "grad_norm": 15.79513931274414,
-      "learning_rate": 1.9821451197042028e-05,
-      "loss": 0.5549,
+      "epoch": 0.26413816457839484,
+      "grad_norm": 28.440961837768555,
+      "learning_rate": 1.9973790155421406e-05,
+      "loss": 0.3467,
       "step": 130
     },
     {
-      "epoch": 0.3656884875846501,
-      "grad_norm": 10.228875160217285,
-      "learning_rate": 1.980490476630592e-05,
-      "loss": 0.5006,
+      "epoch": 0.274297324754487,
+      "grad_norm": 25.352418899536133,
+      "learning_rate": 1.9968810710106065e-05,
+      "loss": 0.4385,
       "step": 135
     },
     {
-      "epoch": 0.3792325056433409,
-      "grad_norm": 5.66217041015625,
-      "learning_rate": 1.9787632766252256e-05,
-      "loss": 0.4001,
+      "epoch": 0.28445648493057907,
+      "grad_norm": 23.881067276000977,
+      "learning_rate": 1.9963399208178066e-05,
+      "loss": 0.4138,
       "step": 140
     },
     {
-      "epoch": 0.3927765237020316,
-      "grad_norm": 6.6009521484375,
-      "learning_rate": 1.976963647502025e-05,
-      "loss": 0.6198,
+      "epoch": 0.2946156451066712,
+      "grad_norm": 10.770444869995117,
+      "learning_rate": 1.995755588417644e-05,
+      "loss": 0.3605,
       "step": 145
     },
     {
-      "epoch": 0.40632054176072235,
-      "grad_norm": 6.579983234405518,
-      "learning_rate": 1.9750917224347136e-05,
-      "loss": 0.3937,
+      "epoch": 0.3047748052827633,
+      "grad_norm": 10.67106819152832,
+      "learning_rate": 1.995128099135575e-05,
+      "loss": 0.3031,
       "step": 150
     },
     {
-      "epoch": 0.4198645598194131,
-      "grad_norm": 9.299453735351562,
-      "learning_rate": 1.9731476399469615e-05,
-      "loss": 0.4485,
+      "epoch": 0.3149339654588554,
+      "grad_norm": 12.032747268676758,
+      "learning_rate": 1.9944574801675106e-05,
+      "loss": 0.4027,
       "step": 155
     },
     {
-      "epoch": 0.43340857787810383,
-      "grad_norm": 8.393924713134766,
-      "learning_rate": 1.9711315439021347e-05,
-      "loss": 0.4297,
+      "epoch": 0.3250931256349475,
+      "grad_norm": 13.55962085723877,
+      "learning_rate": 1.9937437605786395e-05,
+      "loss": 0.4205,
       "step": 160
     },
     {
-      "epoch": 0.4469525959367946,
-      "grad_norm": 6.152463912963867,
-      "learning_rate": 1.9690435834926495e-05,
-      "loss": 0.3706,
+      "epoch": 0.3352522858110396,
+      "grad_norm": 10.579209327697754,
+      "learning_rate": 1.9929869713021668e-05,
+      "loss": 0.4578,
       "step": 165
     },
     {
-      "epoch": 0.4604966139954853,
-      "grad_norm": 9.310385704040527,
-      "learning_rate": 1.9668839132289315e-05,
-      "loss": 0.4448,
+      "epoch": 0.34541144598713175,
+      "grad_norm": 9.225481033325195,
+      "learning_rate": 1.992187145137974e-05,
+      "loss": 0.3515,
       "step": 170
     },
     {
-      "epoch": 0.47404063205417607,
-      "grad_norm": 8.190108299255371,
-      "learning_rate": 1.964652692927982e-05,
-      "loss": 0.371,
+      "epoch": 0.35557060616322383,
+      "grad_norm": 11.829900741577148,
+      "learning_rate": 1.991344316751198e-05,
+      "loss": 0.4827,
       "step": 175
     },
     {
-      "epoch": 0.48758465011286684,
-      "grad_norm": 6.591668605804443,
-      "learning_rate": 1.962350087701552e-05,
-      "loss": 0.3555,
+      "epoch": 0.365729766339316,
+      "grad_norm": 8.982975006103516,
+      "learning_rate": 1.990458522670727e-05,
+      "loss": 0.3654,
       "step": 180
     },
     {
-      "epoch": 0.5011286681715575,
-      "grad_norm": 8.31998062133789,
-      "learning_rate": 1.959976267943923e-05,
-      "loss": 0.4765,
+      "epoch": 0.37588892651540806,
+      "grad_norm": 13.60408878326416,
+      "learning_rate": 1.9895298012876192e-05,
+      "loss": 0.4196,
       "step": 185
     },
     {
-      "epoch": 0.5146726862302483,
-      "grad_norm": 15.614466667175293,
-      "learning_rate": 1.9575314093192978e-05,
-      "loss": 0.3886,
+      "epoch": 0.38604808669150015,
+      "grad_norm": 15.972042083740234,
+      "learning_rate": 1.988558192853438e-05,
+      "loss": 0.4916,
       "step": 190
     },
     {
-      "epoch": 0.5282167042889391,
-      "grad_norm": 10.621317863464355,
-      "learning_rate": 1.9550156927488017e-05,
-      "loss": 0.4668,
+      "epoch": 0.3962072468675923,
+      "grad_norm": 13.422971725463867,
+      "learning_rate": 1.987543739478507e-05,
+      "loss": 0.4736,
       "step": 195
     },
     {
-      "epoch": 0.5417607223476298,
-      "grad_norm": 9.298853874206543,
-      "learning_rate": 1.9524293043970937e-05,
-      "loss": 0.4527,
+      "epoch": 0.4063664070436844,
+      "grad_norm": 7.696578502655029,
+      "learning_rate": 1.9864864851300863e-05,
+      "loss": 0.3668,
       "step": 200
     },
     {
-      "epoch": 0.5553047404063205,
-      "grad_norm": 5.578908920288086,
-      "learning_rate": 1.9497724356585913e-05,
-      "loss": 0.3918,
+      "epoch": 0.4165255672197765,
+      "grad_norm": 10.70732593536377,
+      "learning_rate": 1.9853864756304654e-05,
+      "loss": 0.3385,
       "step": 205
     },
     {
-      "epoch": 0.5688487584650113,
-      "grad_norm": 6.8293256759643555,
-      "learning_rate": 1.947045283143306e-05,
-      "loss": 0.4018,
+      "epoch": 0.4266847273958686,
+      "grad_norm": 11.150247573852539,
+      "learning_rate": 1.9842437586549783e-05,
+      "loss": 0.4407,
       "step": 210
     },
     {
-      "epoch": 0.582392776523702,
-      "grad_norm": 7.2509894371032715,
-      "learning_rate": 1.9442480486622933e-05,
-      "loss": 0.3758,
+      "epoch": 0.43684388757196074,
+      "grad_norm": 9.345061302185059,
+      "learning_rate": 1.9830583837299363e-05,
+      "loss": 0.363,
       "step": 215
     },
     {
-      "epoch": 0.5959367945823928,
-      "grad_norm": 7.862231254577637,
-      "learning_rate": 1.941380939212722e-05,
-      "loss": 0.4305,
+      "epoch": 0.44700304774805283,
+      "grad_norm": 10.30116081237793,
+      "learning_rate": 1.9818304022304824e-05,
+      "loss": 0.3223,
       "step": 220
     },
     {
-      "epoch": 0.6094808126410836,
-      "grad_norm": 6.373060703277588,
-      "learning_rate": 1.9384441669625514e-05,
-      "loss": 0.3768,
+      "epoch": 0.4571622079241449,
+      "grad_norm": 11.17568588256836,
+      "learning_rate": 1.9805598673783644e-05,
+      "loss": 0.4883,
       "step": 225
     },
     {
-      "epoch": 0.6230248306997742,
-      "grad_norm": 10.84733772277832,
-      "learning_rate": 1.9354379492348348e-05,
-      "loss": 0.5301,
+      "epoch": 0.46732136810023706,
+      "grad_norm": 7.653155326843262,
+      "learning_rate": 1.9792468342396277e-05,
+      "loss": 0.3687,
       "step": 230
     },
     {
-      "epoch": 0.636568848758465,
-      "grad_norm": 8.134037971496582,
-      "learning_rate": 1.9323625084916358e-05,
-      "loss": 0.4017,
+      "epoch": 0.47748052827632914,
+      "grad_norm": 8.249720573425293,
+      "learning_rate": 1.977891359722229e-05,
+      "loss": 0.3696,
       "step": 235
     },
     {
-      "epoch": 0.6501128668171557,
-      "grad_norm": 7.003602504730225,
-      "learning_rate": 1.9292180723175656e-05,
-      "loss": 0.4416,
+      "epoch": 0.4876396884524213,
+      "grad_norm": 6.872019290924072,
+      "learning_rate": 1.9764935025735704e-05,
+      "loss": 0.2978,
       "step": 240
     },
     {
-      "epoch": 0.6636568848758465,
-      "grad_norm": 9.637020111083984,
-      "learning_rate": 1.926004873402942e-05,
-      "loss": 0.4691,
+      "epoch": 0.49779884862851337,
+      "grad_norm": 8.779520034790039,
+      "learning_rate": 1.975053323377952e-05,
+      "loss": 0.4099,
       "step": 245
     },
     {
-      "epoch": 0.6772009029345373,
-      "grad_norm": 9.159772872924805,
-      "learning_rate": 1.9227231495265707e-05,
-      "loss": 0.4566,
+      "epoch": 0.5079580088046055,
+      "grad_norm": 6.471978664398193,
+      "learning_rate": 1.9735708845539486e-05,
+      "loss": 0.3345,
       "step": 250
     },
     {
-      "epoch": 0.690744920993228,
-      "grad_norm": 8.606935501098633,
-      "learning_rate": 1.919373143538149e-05,
-      "loss": 0.389,
+      "epoch": 0.5181171689806976,
+      "grad_norm": 7.227390766143799,
+      "learning_rate": 1.9720462503517e-05,
+      "loss": 0.4044,
       "step": 255
     },
     {
-      "epoch": 0.7042889390519187,
-      "grad_norm": 6.0342607498168945,
-      "learning_rate": 1.9159551033402952e-05,
-      "loss": 0.3691,
+      "epoch": 0.5282763291567897,
+      "grad_norm": 10.536629676818848,
+      "learning_rate": 1.9704794868501314e-05,
+      "loss": 0.3444,
       "step": 260
     },
     {
-      "epoch": 0.7178329571106095,
-      "grad_norm": 12.473052978515625,
-      "learning_rate": 1.912469281870202e-05,
-      "loss": 0.4589,
+      "epoch": 0.5384354893328818,
+      "grad_norm": 5.935635566711426,
+      "learning_rate": 1.9688706619540863e-05,
+      "loss": 0.4022,
       "step": 265
     },
     {
-      "epoch": 0.7313769751693002,
-      "grad_norm": 11.965147018432617,
-      "learning_rate": 1.9089159370809213e-05,
-      "loss": 0.4183,
+      "epoch": 0.548594649508974,
+      "grad_norm": 10.507606506347656,
+      "learning_rate": 1.967219845391384e-05,
+      "loss": 0.453,
       "step": 270
     },
     {
-      "epoch": 0.744920993227991,
-      "grad_norm": 7.451316833496094,
-      "learning_rate": 1.9052953319222737e-05,
-      "loss": 0.3731,
+      "epoch": 0.558753809685066,
+      "grad_norm": 9.127860069274902,
+      "learning_rate": 1.965527108709798e-05,
+      "loss": 0.3919,
       "step": 275
     },
     {
-      "epoch": 0.7584650112866818,
-      "grad_norm": 7.067313194274902,
-      "learning_rate": 1.901607734321392e-05,
-      "loss": 0.4241,
+      "epoch": 0.5689129698611581,
+      "grad_norm": 8.938516616821289,
+      "learning_rate": 1.963792525273956e-05,
+      "loss": 0.4063,
       "step": 280
     },
     {
-      "epoch": 0.7720090293453724,
-      "grad_norm": 8.900699615478516,
-      "learning_rate": 1.897853417162892e-05,
-      "loss": 0.4662,
+      "epoch": 0.5790721300372502,
+      "grad_norm": 5.574091911315918,
+      "learning_rate": 1.962016170262157e-05,
+      "loss": 0.3903,
       "step": 285
     },
     {
-      "epoch": 0.7855530474040632,
-      "grad_norm": 8.984304428100586,
-      "learning_rate": 1.894032658268681e-05,
-      "loss": 0.4368,
+      "epoch": 0.5892312902133424,
+      "grad_norm": 8.818571090698242,
+      "learning_rate": 1.960198120663117e-05,
+      "loss": 0.3645,
       "step": 290
     },
     {
-      "epoch": 0.7990970654627539,
-      "grad_norm": 17.687589645385742,
-      "learning_rate": 1.890145740377397e-05,
-      "loss": 0.4358,
+      "epoch": 0.5993904503894345,
+      "grad_norm": 10.282978057861328,
+      "learning_rate": 1.9583384552726294e-05,
+      "loss": 0.4503,
       "step": 295
     },
     {
-      "epoch": 0.8126410835214447,
-      "grad_norm": 10.592819213867188,
-      "learning_rate": 1.886192951123487e-05,
-      "loss": 0.4406,
+      "epoch": 0.6095496105655266,
+      "grad_norm": 11.942185401916504,
+      "learning_rate": 1.9564372546901512e-05,
+      "loss": 0.3296,
       "step": 300
     },
     {
-      "epoch": 0.8261851015801355,
-      "grad_norm": 7.463470458984375,
-      "learning_rate": 1.882174583015922e-05,
-      "loss": 0.5127,
+      "epoch": 0.6197087707416187,
+      "grad_norm": 15.513096809387207,
+      "learning_rate": 1.9544946013153093e-05,
+      "loss": 0.4272,
       "step": 305
     },
     {
-      "epoch": 0.8397291196388262,
-      "grad_norm": 9.488760948181152,
-      "learning_rate": 1.8780909334165505e-05,
-      "loss": 0.4632,
+      "epoch": 0.6298679309177108,
+      "grad_norm": 13.280257225036621,
+      "learning_rate": 1.9525105793443288e-05,
+      "loss": 0.4149,
       "step": 310
     },
     {
-      "epoch": 0.8532731376975169,
-      "grad_norm": 7.275017738342285,
-      "learning_rate": 1.8739423045180936e-05,
-      "loss": 0.3711,
+      "epoch": 0.640027091093803,
+      "grad_norm": 14.556711196899414,
+      "learning_rate": 1.9504852747663862e-05,
+      "loss": 0.371,
       "step": 315
     },
     {
-      "epoch": 0.8668171557562077,
-      "grad_norm": 10.391792297363281,
-      "learning_rate": 1.8697290033217834e-05,
-      "loss": 0.3747,
+      "epoch": 0.650186251269895,
+      "grad_norm": 11.565403938293457,
+      "learning_rate": 1.948418775359879e-05,
+      "loss": 0.3603,
       "step": 320
     },
     {
-      "epoch": 0.8803611738148984,
-      "grad_norm": 10.058419227600098,
-      "learning_rate": 1.8654513416146428e-05,
-      "loss": 0.4147,
+      "epoch": 0.6603454114459871,
+      "grad_norm": 7.955371379852295,
+      "learning_rate": 1.9463111706886234e-05,
+      "loss": 0.4072,
       "step": 325
     },
     {
-      "epoch": 0.8939051918735892,
-      "grad_norm": 15.269502639770508,
-      "learning_rate": 1.861109635946415e-05,
-      "loss": 0.425,
+      "epoch": 0.6705045716220792,
+      "grad_norm": 10.183494567871094,
+      "learning_rate": 1.9441625520979736e-05,
+      "loss": 0.4365,
       "step": 330
     },
     {
-      "epoch": 0.90744920993228,
-      "grad_norm": 5.774105548858643,
-      "learning_rate": 1.8567042076061382e-05,
-      "loss": 0.4217,
+      "epoch": 0.6806637317981713,
+      "grad_norm": 7.960467338562012,
+      "learning_rate": 1.941973012710859e-05,
+      "loss": 0.4313,
       "step": 335
     },
     {
-      "epoch": 0.9209932279909706,
-      "grad_norm": 12.779688835144043,
-      "learning_rate": 1.8522353825983696e-05,
-      "loss": 0.3445,
+      "epoch": 0.6908228919742635,
+      "grad_norm": 7.371551990509033,
+      "learning_rate": 1.9397426474237538e-05,
+      "loss": 0.4297,
       "step": 340
     },
     {
-      "epoch": 0.9345372460496614,
-      "grad_norm": 18.34588050842285,
-      "learning_rate": 1.8477034916190605e-05,
-      "loss": 0.4231,
+      "epoch": 0.7009820521503556,
+      "grad_norm": 7.921060562133789,
+      "learning_rate": 1.9374715529025575e-05,
+      "loss": 0.3439,
       "step": 345
     },
     {
-      "epoch": 0.9480812641083521,
-      "grad_norm": 6.060946941375732,
-      "learning_rate": 1.8431088700310846e-05,
-      "loss": 0.3402,
+      "epoch": 0.7111412123264477,
+      "grad_norm": 7.862534999847412,
+      "learning_rate": 1.9351598275784116e-05,
+      "loss": 0.3239,
       "step": 350
     },
     {
-      "epoch": 0.9616252821670429,
-      "grad_norm": 8.50345230102539,
-      "learning_rate": 1.8384518578394228e-05,
-      "loss": 0.4596,
+      "epoch": 0.7213003725025398,
+      "grad_norm": 7.013054847717285,
+      "learning_rate": 1.9328075716434287e-05,
+      "loss": 0.3831,
       "step": 355
     },
     {
-      "epoch": 0.9751693002257337,
-      "grad_norm": 4.315084934234619,
-      "learning_rate": 1.833732799666e-05,
-      "loss": 0.346,
+      "epoch": 0.731459532678632,
+      "grad_norm": 8.324191093444824,
+      "learning_rate": 1.9304148870463534e-05,
+      "loss": 0.4026,
       "step": 360
     },
     {
-      "epoch": 0.9887133182844243,
-      "grad_norm": 12.581374168395996,
-      "learning_rate": 1.8289520447241846e-05,
-      "loss": 0.4715,
+      "epoch": 0.741618692854724,
+      "grad_norm": 6.676064491271973,
+      "learning_rate": 1.9279818774881418e-05,
+      "loss": 0.3658,
       "step": 365
     },
     {
-      "epoch": 1.0,
-      "grad_norm": 2.780374765396118,
-      "learning_rate": 1.8241099467929455e-05,
-      "loss": 0.3656,
+      "epoch": 0.7517778530308161,
+      "grad_norm": 9.717394828796387,
+      "learning_rate": 1.925508648417467e-05,
+      "loss": 0.4352,
       "step": 370
     },
+    {
+      "epoch": 0.7619370132069082,
+      "grad_norm": 11.164401054382324,
+      "learning_rate": 1.922995307026151e-05,
+      "loss": 0.3615,
+      "step": 375
+    },
+    {
+      "epoch": 0.7720961733830003,
+      "grad_norm": 11.754335403442383,
+      "learning_rate": 1.9204419622445157e-05,
+      "loss": 0.418,
+      "step": 380
+    },
+    {
+      "epoch": 0.7822553335590925,
+      "grad_norm": 12.489151000976562,
+      "learning_rate": 1.9178487247366652e-05,
+      "loss": 0.4442,
+      "step": 385
+    },
+    {
+      "epoch": 0.7924144937351846,
+      "grad_norm": 9.152201652526855,
+      "learning_rate": 1.9152157068956863e-05,
+      "loss": 0.433,
+      "step": 390
+    },
+    {
+      "epoch": 0.8025736539112767,
+      "grad_norm": 11.61486530303955,
+      "learning_rate": 1.9125430228387794e-05,
+      "loss": 0.3619,
+      "step": 395
+    },
+    {
+      "epoch": 0.8127328140873687,
+      "grad_norm": 11.761738777160645,
+      "learning_rate": 1.9098307884023122e-05,
+      "loss": 0.4475,
+      "step": 400
+    },
+    {
+      "epoch": 0.8228919742634608,
+      "grad_norm": 13.330485343933105,
+      "learning_rate": 1.9070791211367984e-05,
+      "loss": 0.4375,
+      "step": 405
+    },
+    {
+      "epoch": 0.833051134439553,
+      "grad_norm": 11.111174583435059,
+      "learning_rate": 1.9042881403018044e-05,
+      "loss": 0.4828,
+      "step": 410
+    },
+    {
+      "epoch": 0.8432102946156451,
+      "grad_norm": 9.631218910217285,
+      "learning_rate": 1.901457966860779e-05,
+      "loss": 0.3723,
+      "step": 415
+    },
+    {
+      "epoch": 0.8533694547917372,
+      "grad_norm": 9.732425689697266,
+      "learning_rate": 1.898588723475811e-05,
+      "loss": 0.4157,
+      "step": 420
+    },
+    {
+      "epoch": 0.8635286149678293,
+      "grad_norm": 7.526575565338135,
+      "learning_rate": 1.8956805345023145e-05,
+      "loss": 0.3623,
+      "step": 425
+    },
+    {
+      "epoch": 0.8736877751439215,
+      "grad_norm": 2.7879254817962646,
+      "learning_rate": 1.8927335259836376e-05,
+      "loss": 0.3487,
+      "step": 430
+    },
+    {
+      "epoch": 0.8838469353200136,
+      "grad_norm": 7.137866497039795,
+      "learning_rate": 1.889747825645599e-05,
+      "loss": 0.5294,
+      "step": 435
+    },
+    {
+      "epoch": 0.8940060954961057,
+      "grad_norm": 9.2273588180542,
+      "learning_rate": 1.8867235628909553e-05,
+      "loss": 0.4501,
+      "step": 440
+    },
+    {
+      "epoch": 0.9041652556721977,
+      "grad_norm": 5.851348876953125,
+      "learning_rate": 1.8836608687937883e-05,
+      "loss": 0.3622,
+      "step": 445
+    },
+    {
+      "epoch": 0.9143244158482898,
+      "grad_norm": 5.815377712249756,
+      "learning_rate": 1.8805598760938282e-05,
+      "loss": 0.3976,
+      "step": 450
+    },
+    {
+      "epoch": 0.924483576024382,
+      "grad_norm": 5.44211483001709,
+      "learning_rate": 1.8774207191906976e-05,
+      "loss": 0.385,
+      "step": 455
+    },
+    {
+      "epoch": 0.9346427362004741,
+      "grad_norm": 4.0405049324035645,
+      "learning_rate": 1.874243534138089e-05,
+      "loss": 0.3863,
+      "step": 460
+    },
+    {
+      "epoch": 0.9448018963765662,
+      "grad_norm": 5.252662658691406,
+      "learning_rate": 1.8710284586378645e-05,
+      "loss": 0.3869,
+      "step": 465
+    },
+    {
+      "epoch": 0.9549610565526583,
+      "grad_norm": 5.840423583984375,
+      "learning_rate": 1.8677756320340927e-05,
+      "loss": 0.4685,
+      "step": 470
+    },
+    {
+      "epoch": 0.9651202167287504,
+      "grad_norm": 10.367392539978027,
+      "learning_rate": 1.8644851953070045e-05,
+      "loss": 0.4045,
+      "step": 475
+    },
+    {
+      "epoch": 0.9752793769048426,
+      "grad_norm": 6.398050785064697,
+      "learning_rate": 1.8611572910668866e-05,
+      "loss": 0.2973,
+      "step": 480
+    },
+    {
+      "epoch": 0.9854385370809347,
+      "grad_norm": 11.543962478637695,
+      "learning_rate": 1.8577920635478976e-05,
+      "loss": 0.3855,
+      "step": 485
+    },
+    {
+      "epoch": 0.9955976972570267,
+      "grad_norm": 12.093887329101562,
+      "learning_rate": 1.85438965860182e-05,
+      "loss": 0.3772,
+      "step": 490
+    },
     {
       "epoch": 1.0,
       "eval_accuracy": 0.9081264108352144,
       "eval_f1": 0.0,
-      "eval_loss": 0.3884802460670471,
+      "eval_loss": 0.4431517720222473,
       "eval_precision": 0.0,
       "eval_recall": 0.0,
-      "eval_runtime": 241.8618,
-      "eval_samples_per_second": 18.316,
-      "eval_steps_per_second": 4.581,
-      "step": 370
+      "eval_runtime": 242.2358,
+      "eval_samples_per_second": 18.288,
+      "eval_steps_per_second": 4.574,
+      "step": 493
     }
   ],
   "logging_steps": 5,
-  "max_steps": 1845,
+  "max_steps": 2460,
   "num_input_tokens_seen": 0,
   "num_train_epochs": 5,
   "save_steps": 500,
@@ -558,7 +726,7 @@
     }
   },
   "total_flos": 0.0,
-  "train_batch_size": 4,
+  "train_batch_size": 3,
   "trial_name": null,
   "trial_params": null
 }

last-checkpoint/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:37a84c70d6770979b673065161611b4cc8e6918473e736a58dc10be5dba7f4a6
+oid sha256:9165fd87d622bef9592d08dfe6f3ba05ef5d0823e7e8902cf48a842306247f27
 size 5304