Training in progress, epoch 9, checkpoint

checkpoint-32391/adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2947462f551402f6c524aff34e2903fc37de1ccc53c6dc4318dd885af9a5cc91
+oid sha256:0c1e7adb550d5476a0bbfbbb194adde1ef42e4fa00271003acb388ab5389f7ff
 size 541459256

checkpoint-32391/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d39f27602f4fbfc5ab748e49745072893b30e6a5c974d9be7d622b0574051f50
+oid sha256:e1088be08ef1a69522aaa5f2634ff138bdb4f226ecab5df5db6255d7e2f88879
 size 33662074

checkpoint-32391/rng_state_0.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6ccc10f724be38240d3f94f0d9d254a98b2d8018ee89e1cf995f323af372041a
+oid sha256:4782a110a81ee9be2e3f67ef9761d8c4feaf5cfd83a28b304db753a061a426d5
 size 15024

checkpoint-32391/rng_state_1.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:de9a0fe74f9515a7071cf9efedad04d23098ea6521119fe7ae755ea0dceee802
+oid sha256:1b405771dd0efddca4128096ed2517b61c8404000792e566690682d74ef1be9a
 size 15024

checkpoint-32391/rng_state_2.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:33bf17125afbe7cfbdade29669a7dc68b6e5c21c43082e3d35ade740b692db5b
+oid sha256:0faafcc9239c28a6a635ab68d7497576b26359552c6c0358832588be5aec95ea
 size 15024

checkpoint-32391/rng_state_3.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f2e59d43374dfc728b281ca66655d1f6811731c5ff935e4e976d891a337b6d9c
+oid sha256:2755df21357a8b4fb6cecbd8e3df57ac4e8db475645b7a5072187ec9fcebd429
 size 15024

checkpoint-32391/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:87c292156116e9deaf3157f793714b76d77d0bbae6d6578d153076864d4be18b
+oid sha256:cc7189ac699c1d4873a4c3fedac86620e0a2a925c9532fef89aab9c677e6a89d
 size 1064

checkpoint-32391/trainer_state.json

@@ -10,802 +10,802 @@
   "log_history": [
     {
       "epoch": 0.13892747985551543,
-      "grad_norm": 0.5752785205841064,
-      "learning_rate": 0.00039538760766879693,
-      "loss": 0.2947,
+      "grad_norm": 0.6994414925575256,
+      "learning_rate": 0.00039537834583680656,
+      "loss": 0.2943,
       "step": 500
     },
     {
       "epoch": 0.2000555709919422,
-      "eval_loss": 0.19290830194950104,
-      "eval_runtime": 6.2591,
-      "eval_samples_per_second": 79.884,
-      "eval_steps_per_second": 10.065,
+      "eval_loss": 0.21401312947273254,
+      "eval_runtime": 16.5317,
+      "eval_samples_per_second": 30.245,
+      "eval_steps_per_second": 3.811,
       "step": 720
     },
     {
       "epoch": 0.27785495971103086,
-      "grad_norm": 0.4608103632926941,
+      "grad_norm": 0.5945971608161926,
       "learning_rate": 0.00039075669167361305,
-      "loss": 0.1972,
+      "loss": 0.1986,
       "step": 1000
     },
     {
       "epoch": 0.4001111419838844,
-      "eval_loss": 0.17835241556167603,
-      "eval_runtime": 6.2534,
-      "eval_samples_per_second": 79.957,
-      "eval_steps_per_second": 10.075,
+      "eval_loss": 0.17625106871128082,
+      "eval_runtime": 16.4823,
+      "eval_samples_per_second": 30.336,
+      "eval_steps_per_second": 3.822,
       "step": 1440
     },
     {
       "epoch": 0.41678243956654626,
-      "grad_norm": 0.44341567158699036,
+      "grad_norm": 0.3619824945926666,
       "learning_rate": 0.0003861257756784292,
-      "loss": 0.1817,
+      "loss": 0.1802,
       "step": 1500
     },
     {
       "epoch": 0.5557099194220617,
-      "grad_norm": 0.3840184509754181,
+      "grad_norm": 0.40362638235092163,
       "learning_rate": 0.00038149485968324534,
-      "loss": 0.1679,
+      "loss": 0.1686,
       "step": 2000
     },
     {
       "epoch": 0.6001667129758266,
-      "eval_loss": 0.16081829369068146,
-      "eval_runtime": 6.259,
-      "eval_samples_per_second": 79.885,
-      "eval_steps_per_second": 10.065,
+      "eval_loss": 0.16156445443630219,
+      "eval_runtime": 16.3535,
+      "eval_samples_per_second": 30.575,
+      "eval_steps_per_second": 3.852,
       "step": 2160
     },
     {
       "epoch": 0.6946373992775771,
-      "grad_norm": 0.4211641848087311,
+      "grad_norm": 0.4962793290615082,
       "learning_rate": 0.0003768639436880615,
-      "loss": 0.1584,
+      "loss": 0.1577,
       "step": 2500
     },
     {
       "epoch": 0.8002222839677688,
-      "eval_loss": 0.15788592398166656,
-      "eval_runtime": 6.2616,
-      "eval_samples_per_second": 79.852,
-      "eval_steps_per_second": 10.061,
+      "eval_loss": 0.15043224394321442,
+      "eval_runtime": 16.3325,
+      "eval_samples_per_second": 30.614,
+      "eval_steps_per_second": 3.857,
       "step": 2880
     },
     {
       "epoch": 0.8335648791330925,
-      "grad_norm": 0.3697059750556946,
-      "learning_rate": 0.0003722515513568584,
-      "loss": 0.1501,
+      "grad_norm": 0.41315382719039917,
+      "learning_rate": 0.0003722330276928777,
+      "loss": 0.1493,
       "step": 3000
     },
     {
       "epoch": 0.972492358988608,
-      "grad_norm": 0.420885294675827,
-      "learning_rate": 0.00036762063536167455,
-      "loss": 0.1478,
+      "grad_norm": 0.37953054904937744,
+      "learning_rate": 0.0003676113735296842,
+      "loss": 0.1477,
       "step": 3500
     },
     {
       "epoch": 1.000277854959711,
-      "eval_loss": 0.14726293087005615,
-      "eval_runtime": 6.3796,
-      "eval_samples_per_second": 78.375,
-      "eval_steps_per_second": 9.875,
+      "eval_loss": 0.14538082480430603,
+      "eval_runtime": 16.4466,
+      "eval_samples_per_second": 30.401,
+      "eval_steps_per_second": 3.831,
       "step": 3600
     },
     {
       "epoch": 1.1114198388441234,
-      "grad_norm": 0.3687162697315216,
-      "learning_rate": 0.0003629897193664907,
-      "loss": 0.1315,
+      "grad_norm": 0.4146048128604889,
+      "learning_rate": 0.00036298045753450036,
+      "loss": 0.1318,
       "step": 4000
     },
     {
       "epoch": 1.2003334259516532,
-      "eval_loss": 0.14390353858470917,
-      "eval_runtime": 6.2741,
-      "eval_samples_per_second": 79.693,
-      "eval_steps_per_second": 10.041,
+      "eval_loss": 0.13961325585842133,
+      "eval_runtime": 16.5835,
+      "eval_samples_per_second": 30.15,
+      "eval_steps_per_second": 3.799,
       "step": 4320
     },
     {
       "epoch": 1.2503473186996388,
-      "grad_norm": 0.4367423951625824,
-      "learning_rate": 0.0003583588033713069,
-      "loss": 0.1309,
+      "grad_norm": 0.6914293766021729,
+      "learning_rate": 0.0003583495415393165,
+      "loss": 0.1303,
       "step": 4500
     },
     {
       "epoch": 1.3892747985551543,
-      "grad_norm": 0.40452277660369873,
-      "learning_rate": 0.000353727887376123,
-      "loss": 0.128,
+      "grad_norm": 0.39365154504776,
+      "learning_rate": 0.00035371862554413265,
+      "loss": 0.1269,
       "step": 5000
     },
     {
       "epoch": 1.4003889969435954,
-      "eval_loss": 0.1357639729976654,
-      "eval_runtime": 6.2739,
-      "eval_samples_per_second": 79.695,
-      "eval_steps_per_second": 10.042,
+      "eval_loss": 0.14090509712696075,
+      "eval_runtime": 16.4413,
+      "eval_samples_per_second": 30.411,
+      "eval_steps_per_second": 3.832,
       "step": 5040
     },
     {
       "epoch": 1.5282022784106695,
-      "grad_norm": 0.6852627396583557,
+      "grad_norm": 0.5620034337043762,
       "learning_rate": 0.00034909697138093914,
-      "loss": 0.126,
+      "loss": 0.1263,
       "step": 5500
     },
     {
       "epoch": 1.6004445679355377,
-      "eval_loss": 0.13292476534843445,
-      "eval_runtime": 6.2659,
-      "eval_samples_per_second": 79.797,
-      "eval_steps_per_second": 10.054,
+      "eval_loss": 0.13119302690029144,
+      "eval_runtime": 16.5618,
+      "eval_samples_per_second": 30.19,
+      "eval_steps_per_second": 3.804,
       "step": 5760
     },
     {
       "epoch": 1.667129758266185,
-      "grad_norm": 0.30270153284072876,
+      "grad_norm": 1.5726152658462524,
       "learning_rate": 0.0003444660553857553,
-      "loss": 0.1255,
+      "loss": 0.1258,
       "step": 6000
     },
     {
       "epoch": 1.8005001389274797,
-      "eval_loss": 0.13217522203922272,
-      "eval_runtime": 6.2758,
-      "eval_samples_per_second": 79.671,
-      "eval_steps_per_second": 10.038,
+      "eval_loss": 0.12919241189956665,
+      "eval_runtime": 16.4031,
+      "eval_samples_per_second": 30.482,
+      "eval_steps_per_second": 3.841,
       "step": 6480
     },
     {
       "epoch": 1.8060572381217006,
-      "grad_norm": 0.394003301858902,
+      "grad_norm": 0.9997938871383667,
       "learning_rate": 0.0003398351393905715,
-      "loss": 0.1214,
+      "loss": 0.1212,
       "step": 6500
     },
     {
       "epoch": 1.9449847179772157,
-      "grad_norm": 0.48431047797203064,
+      "grad_norm": 0.4134954810142517,
       "learning_rate": 0.00033520422339538766,
-      "loss": 0.1199,
+      "loss": 0.1201,
       "step": 7000
     },
     {
       "epoch": 2.000555709919422,
-      "eval_loss": 0.1269044131040573,
-      "eval_runtime": 6.422,
-      "eval_samples_per_second": 77.857,
-      "eval_steps_per_second": 9.81,
+      "eval_loss": 0.12256824225187302,
+      "eval_runtime": 16.8158,
+      "eval_samples_per_second": 29.734,
+      "eval_steps_per_second": 3.746,
       "step": 7200
     },
     {
       "epoch": 2.0839121978327313,
-      "grad_norm": 0.35144302248954773,
+      "grad_norm": 0.40068477392196655,
       "learning_rate": 0.0003305733074002038,
       "loss": 0.108,
       "step": 7500
     },
     {
       "epoch": 2.2006112809113643,
-      "eval_loss": 0.12548324465751648,
-      "eval_runtime": 6.2942,
-      "eval_samples_per_second": 79.438,
-      "eval_steps_per_second": 10.009,
+      "eval_loss": 0.12798596918582916,
+      "eval_runtime": 16.5301,
+      "eval_samples_per_second": 30.248,
+      "eval_steps_per_second": 3.811,
       "step": 7920
     },
     {
       "epoch": 2.222839677688247,
-      "grad_norm": 0.5099118947982788,
-      "learning_rate": 0.00032595165323701027,
-      "loss": 0.1056,
+      "grad_norm": 0.39021220803260803,
+      "learning_rate": 0.0003259423914050199,
+      "loss": 0.1061,
       "step": 8000
     },
     {
       "epoch": 2.361767157543762,
-      "grad_norm": 0.5797615647315979,
-      "learning_rate": 0.00032132073724182645,
-      "loss": 0.1047,
+      "grad_norm": 0.4577147960662842,
+      "learning_rate": 0.0003213114754098361,
+      "loss": 0.1055,
       "step": 8500
     },
     {
       "epoch": 2.4006668519033063,
-      "eval_loss": 0.12244148552417755,
-      "eval_runtime": 6.2902,
-      "eval_samples_per_second": 79.489,
-      "eval_steps_per_second": 10.016,
+      "eval_loss": 0.1270737648010254,
+      "eval_runtime": 16.5227,
+      "eval_samples_per_second": 30.261,
+      "eval_steps_per_second": 3.813,
       "step": 8640
     },
     {
       "epoch": 2.5006946373992776,
-      "grad_norm": 0.40433964133262634,
-      "learning_rate": 0.0003166898212466426,
-      "loss": 0.1038,
+      "grad_norm": 0.5035166144371033,
+      "learning_rate": 0.00031668055941465225,
+      "loss": 0.1033,
       "step": 9000
     },
     {
       "epoch": 2.600722422895249,
-      "eval_loss": 0.12048304080963135,
-      "eval_runtime": 6.289,
-      "eval_samples_per_second": 79.503,
-      "eval_steps_per_second": 10.017,
+      "eval_loss": 0.12219922989606857,
+      "eval_runtime": 16.637,
+      "eval_samples_per_second": 30.054,
+      "eval_steps_per_second": 3.787,
       "step": 9360
     },
     {
       "epoch": 2.639622117254793,
-      "grad_norm": 0.7869251370429993,
+      "grad_norm": 0.6268121004104614,
       "learning_rate": 0.00031205890525145874,
-      "loss": 0.1023,
+      "loss": 0.1026,
       "step": 9500
     },
     {
       "epoch": 2.7785495971103087,
-      "grad_norm": 0.34468919038772583,
+      "grad_norm": 0.37843775749206543,
       "learning_rate": 0.0003074279892562749,
-      "loss": 0.1043,
+      "loss": 0.1042,
       "step": 10000
     },
     {
       "epoch": 2.800777993887191,
-      "eval_loss": 0.1157899871468544,
-      "eval_runtime": 6.2874,
-      "eval_samples_per_second": 79.524,
-      "eval_steps_per_second": 10.02,
+      "eval_loss": 0.11693067103624344,
+      "eval_runtime": 16.4326,
+      "eval_samples_per_second": 30.427,
+      "eval_steps_per_second": 3.834,
       "step": 10080
     },
     {
       "epoch": 2.917477076965824,
-      "grad_norm": 0.5141274929046631,
+      "grad_norm": 0.47878336906433105,
       "learning_rate": 0.00030279707326109104,
-      "loss": 0.1042,
+      "loss": 0.1032,
       "step": 10500
     },
     {
       "epoch": 3.000833564879133,
-      "eval_loss": 0.11479081958532333,
-      "eval_runtime": 6.4111,
-      "eval_samples_per_second": 77.99,
-      "eval_steps_per_second": 9.827,
+      "eval_loss": 0.11544420570135117,
+      "eval_runtime": 16.6741,
+      "eval_samples_per_second": 29.987,
+      "eval_steps_per_second": 3.778,
       "step": 10800
     },
     {
       "epoch": 3.0564045568213394,
-      "grad_norm": 0.4209577143192291,
-      "learning_rate": 0.0002981754190978976,
-      "loss": 0.0957,
+      "grad_norm": 0.6917030215263367,
+      "learning_rate": 0.0002981661572659072,
+      "loss": 0.0954,
       "step": 11000
     },
     {
       "epoch": 3.1953320366768545,
-      "grad_norm": 0.4159316122531891,
-      "learning_rate": 0.00029355376493470407,
-      "loss": 0.0884,
+      "grad_norm": 0.35634613037109375,
+      "learning_rate": 0.0002935352412707234,
+      "loss": 0.0874,
       "step": 11500
     },
     {
       "epoch": 3.2008891358710754,
-      "eval_loss": 0.11302317678928375,
-      "eval_runtime": 6.2864,
-      "eval_samples_per_second": 79.537,
-      "eval_steps_per_second": 10.022,
+      "eval_loss": 0.1193525642156601,
+      "eval_runtime": 16.401,
+      "eval_samples_per_second": 30.486,
+      "eval_steps_per_second": 3.841,
       "step": 11520
     },
     {
       "epoch": 3.33425951653237,
-      "grad_norm": 0.613624632358551,
-      "learning_rate": 0.00028892284893952024,
-      "loss": 0.0903,
+      "grad_norm": 0.4005107283592224,
+      "learning_rate": 0.0002889043252755395,
+      "loss": 0.0902,
       "step": 12000
     },
     {
       "epoch": 3.4009447068630174,
-      "eval_loss": 0.11040342599153519,
-      "eval_runtime": 6.2939,
-      "eval_samples_per_second": 79.442,
-      "eval_steps_per_second": 10.01,
+      "eval_loss": 0.11315633356571198,
+      "eval_runtime": 16.525,
+      "eval_samples_per_second": 30.257,
+      "eval_steps_per_second": 3.812,
       "step": 12240
     },
     {
       "epoch": 3.4731869963878856,
-      "grad_norm": 0.7177799940109253,
-      "learning_rate": 0.0002842919329443364,
-      "loss": 0.0889,
+      "grad_norm": 0.348311185836792,
+      "learning_rate": 0.0002842734092803557,
+      "loss": 0.0891,
       "step": 12500
     },
     {
       "epoch": 3.6010002778549595,
-      "eval_loss": 0.11320961266756058,
-      "eval_runtime": 6.2914,
-      "eval_samples_per_second": 79.473,
-      "eval_steps_per_second": 10.014,
+      "eval_loss": 0.11703498661518097,
+      "eval_runtime": 16.4464,
+      "eval_samples_per_second": 30.402,
+      "eval_steps_per_second": 3.831,
       "step": 12960
     },
     {
       "epoch": 3.612114476243401,
-      "grad_norm": 0.5415129065513611,
-      "learning_rate": 0.0002796610169491526,
-      "loss": 0.0889,
+      "grad_norm": 0.3483453094959259,
+      "learning_rate": 0.0002796424932851718,
+      "loss": 0.0892,
       "step": 13000
     },
     {
       "epoch": 3.7510419560989163,
-      "grad_norm": 0.6058521866798401,
-      "learning_rate": 0.0002750301009539687,
-      "loss": 0.0885,
+      "grad_norm": 0.8891735672950745,
+      "learning_rate": 0.00027502083912197834,
+      "loss": 0.0878,
       "step": 13500
     },
     {
       "epoch": 3.801055848846902,
-      "eval_loss": 0.11218452453613281,
-      "eval_runtime": 6.2932,
-      "eval_samples_per_second": 79.45,
-      "eval_steps_per_second": 10.011,
+      "eval_loss": 0.11358808726072311,
+      "eval_runtime": 16.5506,
+      "eval_samples_per_second": 30.21,
+      "eval_steps_per_second": 3.807,
       "step": 13680
     },
     {
       "epoch": 3.889969435954432,
-      "grad_norm": 0.6304852366447449,
-      "learning_rate": 0.00027039918495878483,
-      "loss": 0.0882,
+      "grad_norm": 2.020261764526367,
+      "learning_rate": 0.00027038992312679446,
+      "loss": 0.088,
       "step": 14000
     },
     {
       "epoch": 4.001111419838844,
-      "eval_loss": 0.1118089109659195,
-      "eval_runtime": 6.4064,
-      "eval_samples_per_second": 78.047,
-      "eval_steps_per_second": 9.834,
+      "eval_loss": 0.11110712587833405,
+      "eval_runtime": 16.9629,
+      "eval_samples_per_second": 29.476,
+      "eval_steps_per_second": 3.714,
       "step": 14400
     },
     {
       "epoch": 4.0288969158099475,
-      "grad_norm": 0.46223384141921997,
-      "learning_rate": 0.000265768268963601,
-      "loss": 0.0858,
+      "grad_norm": 0.4606301486492157,
+      "learning_rate": 0.00026575900713161064,
+      "loss": 0.0848,
       "step": 14500
     },
     {
       "epoch": 4.167824395665463,
-      "grad_norm": 0.4947393238544464,
-      "learning_rate": 0.0002611373529684172,
-      "loss": 0.0753,
+      "grad_norm": 0.5307084321975708,
+      "learning_rate": 0.0002611280911364268,
+      "loss": 0.0754,
       "step": 15000
     },
     {
       "epoch": 4.201166990830786,
-      "eval_loss": 0.11774840950965881,
-      "eval_runtime": 6.2977,
-      "eval_samples_per_second": 79.394,
-      "eval_steps_per_second": 10.004,
+      "eval_loss": 0.11462360620498657,
+      "eval_runtime": 16.4785,
+      "eval_samples_per_second": 30.343,
+      "eval_steps_per_second": 3.823,
       "step": 15120
     },
     {
       "epoch": 4.306751875520978,
-      "grad_norm": 0.40220069885253906,
-      "learning_rate": 0.00025650643697323335,
-      "loss": 0.0764,
+      "grad_norm": 0.45674943923950195,
+      "learning_rate": 0.000256497175141243,
+      "loss": 0.076,
       "step": 15500
     },
     {
       "epoch": 4.4012225618227285,
-      "eval_loss": 0.11307726055383682,
-      "eval_runtime": 6.301,
-      "eval_samples_per_second": 79.353,
-      "eval_steps_per_second": 9.998,
+      "eval_loss": 0.11145643889904022,
+      "eval_runtime": 16.4715,
+      "eval_samples_per_second": 30.356,
+      "eval_steps_per_second": 3.825,
       "step": 15840
     },
     {
       "epoch": 4.445679355376494,
-      "grad_norm": 0.38409939408302307,
-      "learning_rate": 0.00025188478281003984,
-      "loss": 0.0779,
+      "grad_norm": 0.3333616554737091,
+      "learning_rate": 0.00025186625914605905,
+      "loss": 0.0768,
       "step": 16000
     },
     {
       "epoch": 4.584606835232009,
-      "grad_norm": 0.4317833185195923,
-      "learning_rate": 0.00024725386681485596,
-      "loss": 0.0755,
+      "grad_norm": 0.3716529905796051,
+      "learning_rate": 0.0002472446049828656,
+      "loss": 0.0762,
       "step": 16500
     },
     {
       "epoch": 4.601278132814671,
-      "eval_loss": 0.11036152392625809,
-      "eval_runtime": 6.3093,
-      "eval_samples_per_second": 79.248,
-      "eval_steps_per_second": 9.985,
+      "eval_loss": 0.1125354915857315,
+      "eval_runtime": 16.4547,
+      "eval_samples_per_second": 30.386,
+      "eval_steps_per_second": 3.829,
       "step": 16560
     },
     {
       "epoch": 4.723534315087524,
-      "grad_norm": 0.5405837297439575,
-      "learning_rate": 0.00024262295081967214,
-      "loss": 0.077,
+      "grad_norm": 0.5058280229568481,
+      "learning_rate": 0.00024261368898768177,
+      "loss": 0.0771,
       "step": 17000
     },
     {
       "epoch": 4.801333703806613,
-      "eval_loss": 0.10895710438489914,
-      "eval_runtime": 6.3035,
-      "eval_samples_per_second": 79.321,
-      "eval_steps_per_second": 9.994,
+      "eval_loss": 0.10654650628566742,
+      "eval_runtime": 16.5091,
+      "eval_samples_per_second": 30.286,
+      "eval_steps_per_second": 3.816,
       "step": 17280
     },
     {
       "epoch": 4.86246179494304,
-      "grad_norm": 0.3864780366420746,
+      "grad_norm": 0.37022796273231506,
       "learning_rate": 0.0002379920348244883,
-      "loss": 0.0779,
+      "loss": 0.0763,
       "step": 17500
     },
     {
       "epoch": 5.001389274798555,
-      "grad_norm": 0.3517213761806488,
-      "learning_rate": 0.0002333703806612948,
-      "loss": 0.0745,
+      "grad_norm": 0.3737078607082367,
+      "learning_rate": 0.00023336111882930443,
+      "loss": 0.0749,
       "step": 18000
     },
     {
       "epoch": 5.001389274798555,
-      "eval_loss": 0.10910706967115402,
-      "eval_runtime": 6.5025,
-      "eval_samples_per_second": 76.894,
-      "eval_steps_per_second": 9.689,
+      "eval_loss": 0.10845010727643967,
+      "eval_runtime": 16.6489,
+      "eval_samples_per_second": 30.032,
+      "eval_steps_per_second": 3.784,
       "step": 18000
     },
     {
       "epoch": 5.14031675465407,
-      "grad_norm": 0.3627057671546936,
-      "learning_rate": 0.00022873946466611095,
-      "loss": 0.0641,
+      "grad_norm": 0.3841913640499115,
+      "learning_rate": 0.00022873020283412058,
+      "loss": 0.0642,
       "step": 18500
     },
     {
       "epoch": 5.201444845790498,
-      "eval_loss": 0.11183220893144608,
-      "eval_runtime": 6.3062,
-      "eval_samples_per_second": 79.287,
-      "eval_steps_per_second": 9.99,
+      "eval_loss": 0.11235029250383377,
+      "eval_runtime": 16.682,
+      "eval_samples_per_second": 29.972,
+      "eval_steps_per_second": 3.777,
       "step": 18720
     },
     {
       "epoch": 5.279244234509586,
-      "grad_norm": 0.4831227958202362,
-      "learning_rate": 0.00022410854867092712,
-      "loss": 0.0648,
+      "grad_norm": 0.31463760137557983,
+      "learning_rate": 0.00022409928683893675,
+      "loss": 0.0649,
       "step": 19000
     },
     {
       "epoch": 5.401500416782439,
-      "eval_loss": 0.10868020355701447,
-      "eval_runtime": 6.3155,
-      "eval_samples_per_second": 79.17,
-      "eval_steps_per_second": 9.975,
+      "eval_loss": 0.1092103123664856,
+      "eval_runtime": 16.4715,
+      "eval_samples_per_second": 30.355,
+      "eval_steps_per_second": 3.825,
       "step": 19440
     },
     {
       "epoch": 5.418171714365101,
-      "grad_norm": 0.43644019961357117,
-      "learning_rate": 0.0002194776326757433,
-      "loss": 0.066,
+      "grad_norm": 0.30956658720970154,
+      "learning_rate": 0.0002194683708437529,
+      "loss": 0.0651,
       "step": 19500
     },
     {
       "epoch": 5.5570991942206165,
-      "grad_norm": 0.4624837338924408,
-      "learning_rate": 0.00021484671668055944,
-      "loss": 0.0661,
+      "grad_norm": 0.42891213297843933,
+      "learning_rate": 0.00021483745484856908,
+      "loss": 0.0656,
       "step": 20000
     },
     {
       "epoch": 5.601555987774382,
-      "eval_loss": 0.10925421118736267,
-      "eval_runtime": 6.3162,
-      "eval_samples_per_second": 79.161,
-      "eval_steps_per_second": 9.974,
+      "eval_loss": 0.1072624996304512,
+      "eval_runtime": 16.504,
+      "eval_samples_per_second": 30.296,
+      "eval_steps_per_second": 3.817,
       "step": 20160
     },
     {
       "epoch": 5.6960266740761325,
-      "grad_norm": 0.5039137601852417,
-      "learning_rate": 0.00021022506251736593,
-      "loss": 0.0666,
+      "grad_norm": 0.6161347031593323,
+      "learning_rate": 0.00021021580068537556,
+      "loss": 0.0673,
       "step": 20500
     },
     {
       "epoch": 5.801611558766324,
-      "eval_loss": 0.10597137361764908,
-      "eval_runtime": 6.3038,
-      "eval_samples_per_second": 79.317,
-      "eval_steps_per_second": 9.994,
+      "eval_loss": 0.10548041760921478,
+      "eval_runtime": 16.5903,
+      "eval_samples_per_second": 30.138,
+      "eval_steps_per_second": 3.797,
       "step": 20880
     },
     {
       "epoch": 5.834954153931648,
-      "grad_norm": 0.4911608397960663,
-      "learning_rate": 0.0002055941465221821,
-      "loss": 0.0671,
+      "grad_norm": 0.6631760597229004,
+      "learning_rate": 0.0002055848846901917,
+      "loss": 0.0654,
       "step": 21000
     },
     {
       "epoch": 5.973881633787163,
-      "grad_norm": 0.648589015007019,
-      "learning_rate": 0.00020096323052699825,
+      "grad_norm": 0.37077194452285767,
+      "learning_rate": 0.0002009539686950079,
       "loss": 0.065,
       "step": 21500
     },
     {
       "epoch": 6.001667129758266,
-      "eval_loss": 0.10192042589187622,
-      "eval_runtime": 6.5803,
-      "eval_samples_per_second": 75.985,
-      "eval_steps_per_second": 9.574,
+      "eval_loss": 0.10457777976989746,
+      "eval_runtime": 16.8365,
+      "eval_samples_per_second": 29.697,
+      "eval_steps_per_second": 3.742,
       "step": 21600
     },
     {
       "epoch": 6.112809113642679,
-      "grad_norm": 0.4207409918308258,
-      "learning_rate": 0.0001963323145318144,
+      "grad_norm": 0.4291210174560547,
+      "learning_rate": 0.00019632305269982403,
       "loss": 0.0565,
       "step": 22000
     },
     {
       "epoch": 6.201722700750208,
-      "eval_loss": 0.11030665040016174,
-      "eval_runtime": 6.3184,
-      "eval_samples_per_second": 79.133,
-      "eval_steps_per_second": 9.971,
+      "eval_loss": 0.11106568574905396,
+      "eval_runtime": 16.6145,
+      "eval_samples_per_second": 30.094,
+      "eval_steps_per_second": 3.792,
       "step": 22320
     },
     {
       "epoch": 6.251736593498194,
-      "grad_norm": 0.5802621245384216,
-      "learning_rate": 0.00019170139853663055,
-      "loss": 0.0554,
+      "grad_norm": 0.5859522223472595,
+      "learning_rate": 0.00019169213670464018,
+      "loss": 0.0551,
       "step": 22500
     },
     {
       "epoch": 6.390664073353709,
-      "grad_norm": 0.56964111328125,
-      "learning_rate": 0.0001870704825414467,
-      "loss": 0.0575,
+      "grad_norm": 0.6046123504638672,
+      "learning_rate": 0.00018706122070945633,
+      "loss": 0.0566,
       "step": 23000
     },
     {
       "epoch": 6.401778271742151,
-      "eval_loss": 0.10719279199838638,
-      "eval_runtime": 6.3115,
-      "eval_samples_per_second": 79.22,
-      "eval_steps_per_second": 9.982,
+      "eval_loss": 0.11063603311777115,
+      "eval_runtime": 16.5709,
+      "eval_samples_per_second": 30.173,
+      "eval_steps_per_second": 3.802,
       "step": 23040
     },
     {
       "epoch": 6.529591553209225,
-      "grad_norm": 0.5339217185974121,
-      "learning_rate": 0.00018243956654626287,
-      "loss": 0.057,
+      "grad_norm": 0.5555618405342102,
+      "learning_rate": 0.0001824303047142725,
+      "loss": 0.0567,
       "step": 23500
     },
     {
       "epoch": 6.601833842734093,
-      "eval_loss": 0.110511913895607,
-      "eval_runtime": 6.3148,
-      "eval_samples_per_second": 79.179,
-      "eval_steps_per_second": 9.977,
+      "eval_loss": 0.11322695761919022,
+      "eval_runtime": 16.5489,
+      "eval_samples_per_second": 30.213,
+      "eval_steps_per_second": 3.807,
       "step": 23760
     },
     {
       "epoch": 6.66851903306474,
-      "grad_norm": 0.6509028673171997,
-      "learning_rate": 0.000177808650551079,
+      "grad_norm": 0.641233503818512,
+      "learning_rate": 0.00017779938871908863,
       "loss": 0.057,
       "step": 24000
     },
     {
       "epoch": 6.801889413726035,
-      "eval_loss": 0.11028529703617096,
-      "eval_runtime": 6.2947,
-      "eval_samples_per_second": 79.432,
-      "eval_steps_per_second": 10.008,
+      "eval_loss": 0.10657580196857452,
+      "eval_runtime": 16.5793,
+      "eval_samples_per_second": 30.158,
+      "eval_steps_per_second": 3.8,
       "step": 24480
     },
     {
       "epoch": 6.807446512920255,
-      "grad_norm": 0.4752148389816284,
-      "learning_rate": 0.00017318699638788553,
-      "loss": 0.0564,
+      "grad_norm": 0.36209815740585327,
+      "learning_rate": 0.0001731684727239048,
+      "loss": 0.0574,
       "step": 24500
     },
     {
       "epoch": 6.946373992775771,
-      "grad_norm": 0.38218364119529724,
-      "learning_rate": 0.00016856534222469205,
-      "loss": 0.0566,
+      "grad_norm": 0.49614134430885315,
+      "learning_rate": 0.00016853755672872095,
+      "loss": 0.0564,
       "step": 25000
     },
     {
       "epoch": 7.001944984717977,
-      "eval_loss": 0.1053631454706192,
-      "eval_runtime": 6.5447,
-      "eval_samples_per_second": 76.398,
-      "eval_steps_per_second": 9.626,
+      "eval_loss": 0.10740524530410767,
+      "eval_runtime": 16.6216,
+      "eval_samples_per_second": 30.081,
+      "eval_steps_per_second": 3.79,
       "step": 25200
     },
     {
       "epoch": 7.085301472631286,
-      "grad_norm": 0.3690716028213501,
-      "learning_rate": 0.0001639344262295082,
+      "grad_norm": 0.518583357334137,
+      "learning_rate": 0.0001639066407335371,
       "loss": 0.049,
       "step": 25500
     },
     {
       "epoch": 7.20200055570992,
-      "eval_loss": 0.11346188932657242,
-      "eval_runtime": 6.3148,
-      "eval_samples_per_second": 79.179,
-      "eval_steps_per_second": 9.977,
+      "eval_loss": 0.11705406755208969,
+      "eval_runtime": 16.5295,
+      "eval_samples_per_second": 30.249,
+      "eval_steps_per_second": 3.811,
       "step": 25920
     },
     {
       "epoch": 7.2242289524868015,
-      "grad_norm": 0.6985990405082703,
-      "learning_rate": 0.00015930351023432434,
-      "loss": 0.0471,
+      "grad_norm": 0.31886982917785645,
+      "learning_rate": 0.00015928498657034364,
+      "loss": 0.0469,
       "step": 26000
     },
     {
       "epoch": 7.3631564323423175,
-      "grad_norm": 0.4522312581539154,
-      "learning_rate": 0.00015467259423914052,
-      "loss": 0.0471,
+      "grad_norm": 0.28966373205184937,
+      "learning_rate": 0.00015465407057515976,
+      "loss": 0.0478,
       "step": 26500
     },
     {
       "epoch": 7.402056126701861,
-      "eval_loss": 0.11016041040420532,
-      "eval_runtime": 6.3157,
-      "eval_samples_per_second": 79.168,
-      "eval_steps_per_second": 9.975,
+      "eval_loss": 0.10964089632034302,
+      "eval_runtime": 16.5451,
+      "eval_samples_per_second": 30.22,
+      "eval_steps_per_second": 3.808,
       "step": 26640
     },
     {
       "epoch": 7.502083912197833,
-      "grad_norm": 0.3933652341365814,
-      "learning_rate": 0.00015004167824395664,
+      "grad_norm": 0.36664408445358276,
+      "learning_rate": 0.0001500324164119663,
       "loss": 0.0481,
       "step": 27000
     },
     {
       "epoch": 7.602111697693804,
-      "eval_loss": 0.11197081953287125,
-      "eval_runtime": 6.3099,
-      "eval_samples_per_second": 79.24,
-      "eval_steps_per_second": 9.984,
+      "eval_loss": 0.11513197422027588,
+      "eval_runtime": 16.6147,
+      "eval_samples_per_second": 30.094,
+      "eval_steps_per_second": 3.792,
       "step": 27360
     },
     {
       "epoch": 7.641011392053348,
-      "grad_norm": 0.42119812965393066,
-      "learning_rate": 0.00014542002408076318,
-      "loss": 0.0484,
+      "grad_norm": 0.32852914929389954,
+      "learning_rate": 0.00014540150041678245,
+      "loss": 0.048,
       "step": 27500
     },
     {
       "epoch": 7.779938871908864,
-      "grad_norm": 0.44935351610183716,
-      "learning_rate": 0.00014078910808557933,
-      "loss": 0.0481,
+      "grad_norm": 0.5736936330795288,
+      "learning_rate": 0.0001407705844215986,
+      "loss": 0.0483,
       "step": 28000
     },
     {
       "epoch": 7.802167268685746,
-      "eval_loss": 0.10714428871870041,
-      "eval_runtime": 6.3046,
-      "eval_samples_per_second": 79.307,
-      "eval_steps_per_second": 9.993,
+      "eval_loss": 0.10632374882698059,
+      "eval_runtime": 16.6173,
+      "eval_samples_per_second": 30.089,
+      "eval_steps_per_second": 3.791,
       "step": 28080
     },
     {
       "epoch": 7.918866351764379,
-      "grad_norm": 0.31898295879364014,
-      "learning_rate": 0.00013615819209039548,
-      "loss": 0.0486,
+      "grad_norm": 0.335509717464447,
+      "learning_rate": 0.00013613966842641474,
+      "loss": 0.0487,
       "step": 28500
     },
     {
       "epoch": 8.002222839677689,
-      "eval_loss": 0.10903233289718628,
-      "eval_runtime": 6.4188,
-      "eval_samples_per_second": 77.896,
-      "eval_steps_per_second": 9.815,
+      "eval_loss": 0.11238289624452591,
+      "eval_runtime": 16.6415,
+      "eval_samples_per_second": 30.045,
+      "eval_steps_per_second": 3.786,
       "step": 28800
     },
     {
       "epoch": 8.057793831619895,
-      "grad_norm": 0.4274863004684448,
-      "learning_rate": 0.00013152727609521165,
-      "loss": 0.0443,
+      "grad_norm": 0.8646371364593506,
+      "learning_rate": 0.00013150875243123092,
+      "loss": 0.0442,
       "step": 29000
     },
     {
       "epoch": 8.19672131147541,
-      "grad_norm": 0.4764171242713928,
-      "learning_rate": 0.0001268963601000278,
-      "loss": 0.0404,
+      "grad_norm": 0.5025931000709534,
+      "learning_rate": 0.00012687783643604704,
+      "loss": 0.0411,
       "step": 29500
     },
     {
       "epoch": 8.20227841066963,
-      "eval_loss": 0.1156005710363388,
-      "eval_runtime": 6.3188,
-      "eval_samples_per_second": 79.129,
-      "eval_steps_per_second": 9.97,
+      "eval_loss": 0.1167159229516983,
+      "eval_runtime": 16.2769,
+      "eval_samples_per_second": 30.718,
+      "eval_steps_per_second": 3.871,
       "step": 29520
     },
     {
       "epoch": 8.335648791330925,
-      "grad_norm": 0.7897500395774841,
-      "learning_rate": 0.00012226544410484395,
-      "loss": 0.0396,
+      "grad_norm": 0.3653465509414673,
+      "learning_rate": 0.00012225618227285358,
+      "loss": 0.0398,
       "step": 30000
     },
     {
       "epoch": 8.402333981661572,
-      "eval_loss": 0.11201395094394684,
-      "eval_runtime": 6.3256,
-      "eval_samples_per_second": 79.044,
-      "eval_steps_per_second": 9.96,
+      "eval_loss": 0.11505118012428284,
+      "eval_runtime": 16.3723,
+      "eval_samples_per_second": 30.539,
+      "eval_steps_per_second": 3.848,
       "step": 30240
     },
     {
       "epoch": 8.474576271186441,
-      "grad_norm": 0.40140944719314575,
-      "learning_rate": 0.0001176345281096601,
-      "loss": 0.0406,
+      "grad_norm": 0.4740363359451294,
+      "learning_rate": 0.00011762526627766973,
+      "loss": 0.0401,
       "step": 30500
     },
     {
       "epoch": 8.602389552653515,
-      "eval_loss": 0.11844490468502045,
-      "eval_runtime": 6.3191,
-      "eval_samples_per_second": 79.125,
-      "eval_steps_per_second": 9.97,
+      "eval_loss": 0.11725138872861862,
+      "eval_runtime": 16.2874,
+      "eval_samples_per_second": 30.699,
+      "eval_steps_per_second": 3.868,
       "step": 30960
     },
     {
       "epoch": 8.613503751041955,
-      "grad_norm": 0.4029806852340698,
-      "learning_rate": 0.00011301287394646662,
-      "loss": 0.0415,
+      "grad_norm": 0.4850703775882721,
+      "learning_rate": 0.00011299435028248589,
+      "loss": 0.0411,
       "step": 31000
     },
     {
       "epoch": 8.752431230897471,
-      "grad_norm": 0.41953355073928833,
-      "learning_rate": 0.00010839121978327314,
-      "loss": 0.0413,
+      "grad_norm": 0.4242146611213684,
+      "learning_rate": 0.00010836343428730202,
+      "loss": 0.0411,
       "step": 31500
     },
     {
       "epoch": 8.802445123645457,
-      "eval_loss": 0.11227717995643616,
-      "eval_runtime": 6.3242,
-      "eval_samples_per_second": 79.061,
-      "eval_steps_per_second": 9.962,
+      "eval_loss": 0.11391962319612503,
+      "eval_runtime": 16.5683,
+      "eval_samples_per_second": 30.178,
+      "eval_steps_per_second": 3.802,
       "step": 31680
     },
     {
       "epoch": 8.891358710752987,
-      "grad_norm": 0.6146722435951233,
-      "learning_rate": 0.0001037603037880893,
-      "loss": 0.0407,
+      "grad_norm": 0.44564756751060486,
+      "learning_rate": 0.00010373251829211818,
+      "loss": 0.0399,
       "step": 32000
     }
   ],

checkpoint-32391/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:dfac93b9b3b40952529bd84fbff166d6257cf3d6fa4cc937269a037e37d853d5
+oid sha256:58052fbf3a2b07f2a6024b5cc28db88f1f0e48109a11483aa716d00657e9906e
 size 5496