Upload task output 5846ca31-dcde-4604-917f-3e562a84c4c9

loss.txt

@@ -1 +1 @@
-500,1.7186222076416016
+500,1.7299695014953613

model-00001-of-00002.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:dcf0d1f5d96482f408af15b01b6d19f378a7bd5154218f2835fa69faab1d7e09
+oid sha256:629e8b76c1c295975bfb8c5c5c3df75fc0140b7313f8002310b18b6c205c8dbf
 size 4972489328

model-00002-of-00002.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8565e07d5a821eb16a7ca49c252695ea04afe37ec0569210658b08a606ac5a71
+oid sha256:2421360f3ef0ad1f3b58523a49046c3e0f166978c4d13c64afb2bfaa9afc3fa8
 size 2669692552

trainer_state.json

@@ -11,710 +11,710 @@
   "log_history": [
     {
       "epoch": 0.004537205081669692,
-      "grad_norm": 1.8671875,
-      "learning_rate": 8.816326530612247e-06,
-      "loss": 2.0754,
+      "grad_norm": 1.4140625,
+      "learning_rate": 1.7632653061224493e-05,
+      "loss": 2.0708,
       "step": 5
     },
     {
       "epoch": 0.009074410163339383,
-      "grad_norm": 0.94921875,
-      "learning_rate": 1.983673469387755e-05,
-      "loss": 2.0129,
+      "grad_norm": 1.1796875,
+      "learning_rate": 3.96734693877551e-05,
+      "loss": 1.9893,
       "step": 10
     },
     {
       "epoch": 0.013611615245009074,
-      "grad_norm": 0.859375,
-      "learning_rate": 3.085714285714286e-05,
-      "loss": 1.9803,
+      "grad_norm": 0.85546875,
+      "learning_rate": 6.171428571428573e-05,
+      "loss": 1.955,
       "step": 15
     },
     {
       "epoch": 0.018148820326678767,
-      "grad_norm": 0.85546875,
-      "learning_rate": 4.187755102040817e-05,
-      "loss": 1.9228,
+      "grad_norm": 1.0,
+      "learning_rate": 8.375510204081634e-05,
+      "loss": 1.9002,
       "step": 20
     },
     {
       "epoch": 0.022686025408348458,
-      "grad_norm": 0.76171875,
-      "learning_rate": 5.289795918367348e-05,
-      "loss": 1.8937,
+      "grad_norm": 0.78515625,
+      "learning_rate": 0.00010579591836734696,
+      "loss": 1.8666,
       "step": 25
     },
     {
       "epoch": 0.02722323049001815,
       "grad_norm": 0.71875,
-      "learning_rate": 6.391836734693879e-05,
-      "loss": 1.9657,
+      "learning_rate": 0.00012783673469387758,
+      "loss": 1.9462,
       "step": 30
     },
     {
       "epoch": 0.03176043557168784,
-      "grad_norm": 0.953125,
-      "learning_rate": 7.493877551020409e-05,
-      "loss": 1.9228,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.00014987755102040818,
+      "loss": 1.9157,
       "step": 35
     },
     {
       "epoch": 0.036297640653357534,
-      "grad_norm": 0.70703125,
-      "learning_rate": 7.714264366388776e-05,
-      "loss": 1.9006,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.00015428528732777552,
+      "loss": 1.9,
       "step": 40
     },
     {
       "epoch": 0.04083484573502722,
-      "grad_norm": 0.73828125,
-      "learning_rate": 7.714177641097452e-05,
-      "loss": 1.8964,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.00015428355282194904,
+      "loss": 1.9086,
       "step": 45
     },
     {
       "epoch": 0.045372050816696916,
-      "grad_norm": 0.66796875,
-      "learning_rate": 7.714024206166619e-05,
-      "loss": 1.9056,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.00015428048412333237,
+      "loss": 1.918,
       "step": 50
     },
     {
       "epoch": 0.0499092558983666,
-      "grad_norm": 0.640625,
-      "learning_rate": 7.713804065134638e-05,
-      "loss": 1.8664,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.00015427608130269275,
+      "loss": 1.8871,
       "step": 55
     },
     {
       "epoch": 0.0544464609800363,
-      "grad_norm": 0.703125,
-      "learning_rate": 7.713517223078173e-05,
-      "loss": 1.8595,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.00015427034446156347,
+      "loss": 1.879,
       "step": 60
     },
     {
       "epoch": 0.05898366606170599,
-      "grad_norm": 0.71875,
-      "learning_rate": 7.713163686612089e-05,
-      "loss": 1.8637,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.00015426327373224179,
+      "loss": 1.8881,
       "step": 65
     },
     {
       "epoch": 0.06352087114337568,
       "grad_norm": 0.69921875,
-      "learning_rate": 7.712743463889284e-05,
-      "loss": 1.8421,
+      "learning_rate": 0.00015425486927778568,
+      "loss": 1.8637,
       "step": 70
     },
     {
       "epoch": 0.06805807622504537,
-      "grad_norm": 0.8125,
-      "learning_rate": 7.712256564600505e-05,
-      "loss": 1.9107,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.0001542451312920101,
+      "loss": 1.9391,
       "step": 75
     },
     {
       "epoch": 0.07259528130671507,
-      "grad_norm": 0.89453125,
-      "learning_rate": 7.71170299997413e-05,
-      "loss": 1.8998,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.0001542340599994826,
+      "loss": 1.9262,
       "step": 80
     },
     {
       "epoch": 0.07713248638838476,
-      "grad_norm": 0.62890625,
-      "learning_rate": 7.711082782775905e-05,
-      "loss": 1.892,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.0001542216556555181,
+      "loss": 1.9177,
       "step": 85
     },
     {
       "epoch": 0.08166969147005444,
-      "grad_norm": 0.6796875,
-      "learning_rate": 7.710395927308647e-05,
-      "loss": 1.879,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.00015420791854617294,
+      "loss": 1.9026,
       "step": 90
     },
     {
       "epoch": 0.08620689655172414,
-      "grad_norm": 0.71875,
-      "learning_rate": 7.70964244941192e-05,
-      "loss": 1.8459,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0001541928489882384,
+      "loss": 1.868,
       "step": 95
     },
     {
       "epoch": 0.09074410163339383,
-      "grad_norm": 0.66015625,
-      "learning_rate": 7.708822366461664e-05,
-      "loss": 1.8314,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.0001541764473292333,
+      "loss": 1.8556,
       "step": 100
     },
     {
       "epoch": 0.09528130671506352,
-      "grad_norm": 0.66796875,
-      "learning_rate": 7.707935697369802e-05,
-      "loss": 1.8655,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.00015415871394739605,
+      "loss": 1.8941,
       "step": 105
     },
     {
       "epoch": 0.0998185117967332,
-      "grad_norm": 0.66796875,
-      "learning_rate": 7.706982462583793e-05,
-      "loss": 1.8556,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.00015413964925167585,
+      "loss": 1.8843,
       "step": 110
     },
     {
       "epoch": 0.10435571687840291,
-      "grad_norm": 0.66796875,
-      "learning_rate": 7.705962684086168e-05,
-      "loss": 1.8623,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.00015411925368172337,
+      "loss": 1.8913,
       "step": 115
     },
     {
       "epoch": 0.1088929219600726,
-      "grad_norm": 0.703125,
-      "learning_rate": 7.704876385394025e-05,
-      "loss": 1.8076,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.0001540975277078805,
+      "loss": 1.8362,
       "step": 120
     },
     {
       "epoch": 0.11343012704174228,
-      "grad_norm": 0.6875,
-      "learning_rate": 7.70372359155848e-05,
-      "loss": 1.8231,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.0001540744718311696,
+      "loss": 1.851,
       "step": 125
     },
     {
       "epoch": 0.11796733212341198,
-      "grad_norm": 0.6796875,
-      "learning_rate": 7.702504329164092e-05,
-      "loss": 1.8412,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.00015405008658328184,
+      "loss": 1.8704,
       "step": 130
     },
     {
       "epoch": 0.12250453720508167,
-      "grad_norm": 0.64453125,
-      "learning_rate": 7.70121862632825e-05,
-      "loss": 1.8033,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.000154024372526565,
+      "loss": 1.8304,
       "step": 135
     },
     {
       "epoch": 0.12704174228675136,
-      "grad_norm": 0.6484375,
-      "learning_rate": 7.69986651270053e-05,
-      "loss": 1.7894,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.0001539973302540106,
+      "loss": 1.8121,
       "step": 140
     },
     {
       "epoch": 0.13157894736842105,
-      "grad_norm": 0.66796875,
-      "learning_rate": 7.698448019462e-05,
-      "loss": 1.8372,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.00015396896038924,
+      "loss": 1.861,
       "step": 145
     },
     {
       "epoch": 0.13611615245009073,
-      "grad_norm": 0.71484375,
-      "learning_rate": 7.69696317932451e-05,
-      "loss": 1.828,
+      "grad_norm": 0.96484375,
+      "learning_rate": 0.0001539392635864902,
+      "loss": 1.858,
       "step": 150
     },
     {
       "epoch": 0.14065335753176045,
-      "grad_norm": 0.66015625,
-      "learning_rate": 7.695412026529934e-05,
-      "loss": 1.8035,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.00015390824053059868,
+      "loss": 1.8305,
       "step": 155
     },
     {
       "epoch": 0.14519056261343014,
-      "grad_norm": 0.65234375,
-      "learning_rate": 7.693794596849383e-05,
-      "loss": 1.8647,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.00015387589193698766,
+      "loss": 1.8998,
       "step": 160
     },
     {
       "epoch": 0.14972776769509982,
-      "grad_norm": 0.78515625,
-      "learning_rate": 7.692110927582376e-05,
-      "loss": 1.8389,
+      "grad_norm": 0.921875,
+      "learning_rate": 0.00015384221855164752,
+      "loss": 1.8725,
       "step": 165
     },
     {
       "epoch": 0.1542649727767695,
-      "grad_norm": 0.65234375,
-      "learning_rate": 7.690361057555983e-05,
-      "loss": 1.8177,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.00015380722115111966,
+      "loss": 1.8505,
       "step": 170
     },
     {
       "epoch": 0.1588021778584392,
-      "grad_norm": 0.671875,
-      "learning_rate": 7.688545027123926e-05,
-      "loss": 1.8568,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.00015377090054247852,
+      "loss": 1.8817,
       "step": 175
     },
     {
       "epoch": 0.16333938294010888,
-      "grad_norm": 0.640625,
-      "learning_rate": 7.686662878165656e-05,
-      "loss": 1.8251,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.00015373325756331313,
+      "loss": 1.8494,
       "step": 180
     },
     {
       "epoch": 0.16787658802177857,
-      "grad_norm": 0.66015625,
-      "learning_rate": 7.684714654085381e-05,
-      "loss": 1.8902,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.00015369429308170762,
+      "loss": 1.9201,
       "step": 185
     },
     {
       "epoch": 0.1724137931034483,
-      "grad_norm": 0.6640625,
-      "learning_rate": 7.682700399811061e-05,
-      "loss": 1.8233,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.00015365400799622123,
+      "loss": 1.8467,
       "step": 190
     },
     {
       "epoch": 0.17695099818511797,
-      "grad_norm": 0.64453125,
-      "learning_rate": 7.680620161793384e-05,
-      "loss": 1.8327,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.00015361240323586768,
+      "loss": 1.8867,
       "step": 195
     },
     {
       "epoch": 0.18148820326678766,
-      "grad_norm": 0.64453125,
-      "learning_rate": 7.678473988004684e-05,
-      "loss": 1.8102,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.00015356947976009368,
+      "loss": 1.8469,
       "step": 200
     },
     {
       "epoch": 0.18602540834845735,
-      "grad_norm": 0.671875,
-      "learning_rate": 7.676261927937842e-05,
-      "loss": 1.8359,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.00015352523855875685,
+      "loss": 1.8678,
       "step": 205
     },
     {
       "epoch": 0.19056261343012704,
-      "grad_norm": 0.65234375,
-      "learning_rate": 7.673984032605136e-05,
-      "loss": 1.8503,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.00015347968065210273,
+      "loss": 1.8783,
       "step": 210
     },
     {
       "epoch": 0.19509981851179672,
-      "grad_norm": 0.64453125,
-      "learning_rate": 7.671640354537077e-05,
-      "loss": 1.8366,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.00015343280709074155,
+      "loss": 1.8663,
       "step": 215
     },
     {
       "epoch": 0.1996370235934664,
-      "grad_norm": 0.671875,
-      "learning_rate": 7.669230947781184e-05,
-      "loss": 1.8637,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.0001533846189556237,
+      "loss": 1.8891,
       "step": 220
     },
     {
       "epoch": 0.20417422867513613,
-      "grad_norm": 0.6875,
-      "learning_rate": 7.66675586790075e-05,
-      "loss": 1.8397,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.000153335117358015,
+      "loss": 1.87,
       "step": 225
     },
     {
       "epoch": 0.20871143375680581,
-      "grad_norm": 0.7421875,
-      "learning_rate": 7.664215171973552e-05,
-      "loss": 1.7521,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.00015328430343947104,
+      "loss": 1.7773,
       "step": 230
     },
     {
       "epoch": 0.2132486388384755,
-      "grad_norm": 0.63671875,
-      "learning_rate": 7.661608918590534e-05,
-      "loss": 1.8029,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.00015323217837181068,
+      "loss": 1.8276,
       "step": 235
     },
     {
       "epoch": 0.2177858439201452,
-      "grad_norm": 0.64453125,
-      "learning_rate": 7.658937167854468e-05,
-      "loss": 1.81,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.00015317874335708936,
+      "loss": 1.8301,
       "step": 240
     },
     {
       "epoch": 0.22232304900181488,
-      "grad_norm": 0.6796875,
-      "learning_rate": 7.656199981378549e-05,
-      "loss": 1.8231,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.00015312399962757098,
+      "loss": 1.8491,
       "step": 245
     },
     {
       "epoch": 0.22686025408348456,
-      "grad_norm": 0.64453125,
-      "learning_rate": 7.653397422284996e-05,
-      "loss": 1.8603,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.0001530679484456999,
+      "loss": 1.8976,
       "step": 250
     },
     {
       "epoch": 0.23139745916515425,
-      "grad_norm": 0.62890625,
-      "learning_rate": 7.650529555203574e-05,
-      "loss": 1.771,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.00015301059110407148,
+      "loss": 1.7919,
       "step": 255
     },
     {
       "epoch": 0.23593466424682397,
-      "grad_norm": 0.64453125,
-      "learning_rate": 7.647596446270122e-05,
-      "loss": 1.8081,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.00015295192892540244,
+      "loss": 1.8364,
       "step": 260
     },
     {
       "epoch": 0.24047186932849365,
-      "grad_norm": 0.7421875,
-      "learning_rate": 7.644598163125018e-05,
-      "loss": 1.8215,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.00015289196326250036,
+      "loss": 1.846,
       "step": 265
     },
     {
       "epoch": 0.24500907441016334,
-      "grad_norm": 0.6328125,
-      "learning_rate": 7.641534774911618e-05,
-      "loss": 1.7823,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.00015283069549823237,
+      "loss": 1.8033,
       "step": 270
     },
     {
       "epoch": 0.24954627949183303,
-      "grad_norm": 0.65234375,
-      "learning_rate": 7.638406352274673e-05,
-      "loss": 1.8182,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.00015276812704549347,
+      "loss": 1.8463,
       "step": 275
     },
     {
       "epoch": 0.2540834845735027,
-      "grad_norm": 0.64453125,
-      "learning_rate": 7.635212967358684e-05,
-      "loss": 1.7561,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.00015270425934717368,
+      "loss": 1.7821,
       "step": 280
     },
     {
       "epoch": 0.25862068965517243,
-      "grad_norm": 0.64453125,
-      "learning_rate": 7.631954693806248e-05,
-      "loss": 1.7742,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.00015263909387612496,
+      "loss": 1.7916,
       "step": 285
     },
     {
       "epoch": 0.2631578947368421,
       "grad_norm": 0.671875,
-      "learning_rate": 7.62863160675636e-05,
-      "loss": 1.8146,
+      "learning_rate": 0.0001525726321351272,
+      "loss": 1.8298,
       "step": 290
     },
     {
       "epoch": 0.2676950998185118,
-      "grad_norm": 0.61328125,
-      "learning_rate": 7.625243782842677e-05,
-      "loss": 1.8276,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.00015250487565685353,
+      "loss": 1.8417,
       "step": 295
     },
     {
       "epoch": 0.27223230490018147,
-      "grad_norm": 0.6328125,
-      "learning_rate": 7.62179130019175e-05,
-      "loss": 1.8267,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.000152435826003835,
+      "loss": 1.8475,
       "step": 300
     },
     {
       "epoch": 0.2767695099818512,
-      "grad_norm": 0.78125,
-      "learning_rate": 7.618274238421226e-05,
-      "loss": 1.771,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.00015236548476842452,
+      "loss": 1.7997,
       "step": 305
     },
     {
       "epoch": 0.2813067150635209,
-      "grad_norm": 0.69140625,
-      "learning_rate": 7.614692678638012e-05,
-      "loss": 1.8036,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.00015229385357276023,
+      "loss": 1.8284,
       "step": 310
     },
     {
       "epoch": 0.28584392014519056,
-      "grad_norm": 0.66796875,
-      "learning_rate": 7.611046703436397e-05,
-      "loss": 1.8304,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.00015222093406872794,
+      "loss": 1.8628,
       "step": 315
     },
     {
       "epoch": 0.29038112522686027,
-      "grad_norm": 0.6640625,
-      "learning_rate": 7.607336396896157e-05,
-      "loss": 1.8091,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.00015214672793792313,
+      "loss": 1.8286,
       "step": 320
     },
     {
       "epoch": 0.29491833030852993,
-      "grad_norm": 0.7578125,
-      "learning_rate": 7.603561844580612e-05,
-      "loss": 1.7907,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.00015207123689161224,
+      "loss": 1.8207,
       "step": 325
     },
     {
       "epoch": 0.29945553539019965,
-      "grad_norm": 0.69921875,
-      "learning_rate": 7.599723133534652e-05,
-      "loss": 1.778,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.00015199446267069304,
+      "loss": 1.8005,
       "step": 330
     },
     {
       "epoch": 0.3039927404718693,
-      "grad_norm": 0.67578125,
-      "learning_rate": 7.59582035228273e-05,
-      "loss": 1.8424,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.0001519164070456546,
+      "loss": 1.8659,
       "step": 335
     },
     {
       "epoch": 0.308529945553539,
-      "grad_norm": 0.64453125,
-      "learning_rate": 7.591853590826822e-05,
-      "loss": 1.8394,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.00015183707181653643,
+      "loss": 1.8642,
       "step": 340
     },
     {
       "epoch": 0.31306715063520874,
-      "grad_norm": 0.6484375,
-      "learning_rate": 7.587822940644347e-05,
-      "loss": 1.7529,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.00015175645881288695,
+      "loss": 1.7679,
       "step": 345
     },
     {
       "epoch": 0.3176043557168784,
-      "grad_norm": 0.6328125,
-      "learning_rate": 7.583728494686066e-05,
-      "loss": 1.8048,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.00015167456989372132,
+      "loss": 1.8267,
       "step": 350
     },
     {
       "epoch": 0.3221415607985481,
-      "grad_norm": 0.6875,
-      "learning_rate": 7.579570347373932e-05,
-      "loss": 1.8103,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.00015159140694747864,
+      "loss": 1.826,
       "step": 355
     },
     {
       "epoch": 0.32667876588021777,
       "grad_norm": 0.6328125,
-      "learning_rate": 7.57534859459891e-05,
-      "loss": 1.8083,
+      "learning_rate": 0.0001515069718919782,
+      "loss": 1.8233,
       "step": 360
     },
     {
       "epoch": 0.3312159709618875,
-      "grad_norm": 0.625,
-      "learning_rate": 7.571063333718775e-05,
-      "loss": 1.8074,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.0001514212666743755,
+      "loss": 1.8402,
       "step": 365
     },
     {
       "epoch": 0.33575317604355714,
-      "grad_norm": 0.65234375,
-      "learning_rate": 7.566714663555857e-05,
-      "loss": 1.7981,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.00015133429327111715,
+      "loss": 1.8067,
       "step": 370
     },
     {
       "epoch": 0.34029038112522686,
-      "grad_norm": 0.6640625,
-      "learning_rate": 7.562302684394771e-05,
-      "loss": 1.7727,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.00015124605368789542,
+      "loss": 1.7867,
       "step": 375
     },
     {
       "epoch": 0.3448275862068966,
-      "grad_norm": 0.6953125,
-      "learning_rate": 7.557827497980093e-05,
-      "loss": 1.7885,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.00015115654995960187,
+      "loss": 1.8151,
       "step": 380
     },
     {
       "epoch": 0.34936479128856623,
-      "grad_norm": 0.7421875,
-      "learning_rate": 7.553289207514029e-05,
-      "loss": 1.7775,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.00015106578415028057,
+      "loss": 1.7844,
       "step": 385
     },
     {
       "epoch": 0.35390199637023595,
-      "grad_norm": 0.68359375,
-      "learning_rate": 7.548687917654018e-05,
-      "loss": 1.8054,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.00015097375835308036,
+      "loss": 1.818,
       "step": 390
     },
     {
       "epoch": 0.3584392014519056,
-      "grad_norm": 0.66015625,
-      "learning_rate": 7.544023734510334e-05,
-      "loss": 1.7976,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.00015088047469020668,
+      "loss": 1.8075,
       "step": 395
     },
     {
       "epoch": 0.3629764065335753,
-      "grad_norm": 0.66015625,
-      "learning_rate": 7.539296765643626e-05,
-      "loss": 1.8313,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.00015078593531287252,
+      "loss": 1.8531,
       "step": 400
     },
     {
       "epoch": 0.367513611615245,
-      "grad_norm": 0.8046875,
-      "learning_rate": 7.534507120062452e-05,
-      "loss": 1.7598,
+      "grad_norm": 0.73828125,
+      "learning_rate": 0.00015069014240124903,
+      "loss": 1.7792,
       "step": 405
     },
     {
       "epoch": 0.3720508166969147,
-      "grad_norm": 0.6875,
-      "learning_rate": 7.529654908220746e-05,
-      "loss": 1.7916,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.0001505930981644149,
+      "loss": 1.8016,
       "step": 410
     },
     {
       "epoch": 0.3765880217785844,
-      "grad_norm": 0.66796875,
-      "learning_rate": 7.524740242015287e-05,
-      "loss": 1.7769,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.00015049480484030574,
+      "loss": 1.7925,
       "step": 415
     },
     {
       "epoch": 0.3811252268602541,
-      "grad_norm": 0.66015625,
-      "learning_rate": 7.519763234783115e-05,
-      "loss": 1.7592,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.0001503952646956623,
+      "loss": 1.7902,
       "step": 420
     },
     {
       "epoch": 0.3856624319419238,
-      "grad_norm": 0.609375,
-      "learning_rate": 7.514724001298914e-05,
-      "loss": 1.7688,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.00015029448002597828,
+      "loss": 1.785,
       "step": 425
     },
     {
       "epoch": 0.39019963702359345,
-      "grad_norm": 0.69140625,
-      "learning_rate": 7.509622657772364e-05,
-      "loss": 1.8018,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.00015019245315544728,
+      "loss": 1.8234,
       "step": 430
     },
     {
       "epoch": 0.39473684210526316,
-      "grad_norm": 0.6328125,
-      "learning_rate": 7.504459321845466e-05,
-      "loss": 1.7811,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.00015008918643690932,
+      "loss": 1.7869,
       "step": 435
     },
     {
       "epoch": 0.3992740471869328,
-      "grad_norm": 0.69140625,
-      "learning_rate": 7.499234112589828e-05,
-      "loss": 1.8019,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.00014998468225179657,
+      "loss": 1.8319,
       "step": 440
     },
     {
       "epoch": 0.40381125226860254,
-      "grad_norm": 0.59765625,
-      "learning_rate": 7.493947150503912e-05,
-      "loss": 1.7838,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.00014987894301007825,
+      "loss": 1.7935,
       "step": 445
     },
     {
       "epoch": 0.40834845735027225,
-      "grad_norm": 0.65234375,
-      "learning_rate": 7.48859855751027e-05,
-      "loss": 1.7756,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.0001497719711502054,
+      "loss": 1.7998,
       "step": 450
     },
     {
       "epoch": 0.4128856624319419,
-      "grad_norm": 0.671875,
-      "learning_rate": 7.483188456952715e-05,
-      "loss": 1.8105,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.0001496637691390543,
+      "loss": 1.8234,
       "step": 455
     },
     {
       "epoch": 0.41742286751361163,
-      "grad_norm": 0.67578125,
-      "learning_rate": 7.477716973593491e-05,
-      "loss": 1.7537,
+      "grad_norm": 1.40625,
+      "learning_rate": 0.00014955433947186982,
+      "loss": 1.7724,
       "step": 460
     },
     {
       "epoch": 0.4219600725952813,
-      "grad_norm": 0.6328125,
-      "learning_rate": 7.472184233610385e-05,
-      "loss": 1.7794,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.0001494436846722077,
+      "loss": 1.7939,
       "step": 465
     },
     {
       "epoch": 0.426497277676951,
-      "grad_norm": 0.76953125,
-      "learning_rate": 7.466590364593826e-05,
-      "loss": 1.7735,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.00014933180729187652,
+      "loss": 1.7921,
       "step": 470
     },
     {
       "epoch": 0.43103448275862066,
-      "grad_norm": 0.5859375,
-      "learning_rate": 7.460935495543935e-05,
-      "loss": 1.8138,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.0001492187099108787,
+      "loss": 1.8269,
       "step": 475
     },
     {
       "epoch": 0.4355716878402904,
-      "grad_norm": 0.66015625,
-      "learning_rate": 7.455219756867557e-05,
-      "loss": 1.7884,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.00014910439513735114,
+      "loss": 1.8065,
       "step": 480
     },
     {
       "epoch": 0.4401088929219601,
-      "grad_norm": 0.65625,
-      "learning_rate": 7.449443280375251e-05,
-      "loss": 1.7829,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.00014898886560750502,
+      "loss": 1.793,
       "step": 485
     },
     {
       "epoch": 0.44464609800362975,
-      "grad_norm": 0.67578125,
-      "learning_rate": 7.443606199278244e-05,
-      "loss": 1.7891,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.00014887212398556488,
+      "loss": 1.8052,
       "step": 490
     },
     {
       "epoch": 0.44918330308529947,
-      "grad_norm": 0.65625,
-      "learning_rate": 7.437708648185371e-05,
-      "loss": 1.7422,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.00014875417296370742,
+      "loss": 1.7471,
       "step": 495
     },
     {
       "epoch": 0.4537205081669691,
-      "grad_norm": 0.65234375,
-      "learning_rate": 7.431750763099963e-05,
-      "loss": 1.7667,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.00014863501526199925,
+      "loss": 1.7734,
       "step": 500
     },
     {
       "epoch": 0.4537205081669691,
-      "eval_loss": 1.7186222076416016,
-      "eval_runtime": 4.1831,
-      "eval_samples_per_second": 15.3,
-      "eval_steps_per_second": 15.3,
+      "eval_loss": 1.7299695014953613,
+      "eval_runtime": 4.0995,
+      "eval_samples_per_second": 15.612,
+      "eval_steps_per_second": 15.612,
       "step": 500
     }
   ],

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:605949a497492ef753b7d89024f80e06bf5a04a31d7dca2d8f50b3fe799f9a24
+oid sha256:1d120edbf599876edd8f2c19617828bc8b987a2fd5b0b50a07010ccefb40085c
 size 5624