Training in progress, epoch 9, checkpoint

checkpoint-32391/adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0c1e7adb550d5476a0bbfbbb194adde1ef42e4fa00271003acb388ab5389f7ff
+oid sha256:0e2caf34709c383443f6ae6cc2159e556ff6af7579d645eb7d2263c1c3eacbe5
 size 541459256

checkpoint-32391/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e1088be08ef1a69522aaa5f2634ff138bdb4f226ecab5df5db6255d7e2f88879
+oid sha256:56b27b6bdeaca18c913363d3e63beea60a5606d0dd553d366da6354b9e87e16b
 size 33662074

checkpoint-32391/rng_state_0.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4782a110a81ee9be2e3f67ef9761d8c4feaf5cfd83a28b304db753a061a426d5
+oid sha256:f05e91e2ebdb9f66dabd55f4a5a11e5dd1712e33f56b9a1ced22d4b37693f618
 size 15024

checkpoint-32391/rng_state_1.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1b405771dd0efddca4128096ed2517b61c8404000792e566690682d74ef1be9a
+oid sha256:57606bda571ed2343e131e3a8f29c83008f104925afea1e46399c58e4e94652d
 size 15024

checkpoint-32391/rng_state_2.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0faafcc9239c28a6a635ab68d7497576b26359552c6c0358832588be5aec95ea
+oid sha256:f1c642c6f4d96743c980e0c808f8f1cd50171206a008e9f1b3bb131a8b702185
 size 15024

checkpoint-32391/rng_state_3.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2755df21357a8b4fb6cecbd8e3df57ac4e8db475645b7a5072187ec9fcebd429
+oid sha256:cd08b8c4d1bfb3d6110707e5f47a02d44e9af34725668e9fbe5a52389ec47f37
 size 15024

checkpoint-32391/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:cc7189ac699c1d4873a4c3fedac86620e0a2a925c9532fef89aab9c677e6a89d
+oid sha256:65176024d1d967afe90959c4508869e0026cecd6b05dfede84f85d0b70d6d1bc
 size 1064

checkpoint-32391/trainer_state.json

@@ -10,802 +10,802 @@
   "log_history": [
     {
       "epoch": 0.13892747985551543,
-      "grad_norm": 0.6994414925575256,
-      "learning_rate": 0.00039537834583680656,
-      "loss": 0.2943,
+      "grad_norm": 0.4299773573875427,
+      "learning_rate": 0.0003953968695007873,
+      "loss": 0.2967,
       "step": 500
     },
     {
       "epoch": 0.2000555709919422,
-      "eval_loss": 0.21401312947273254,
-      "eval_runtime": 16.5317,
-      "eval_samples_per_second": 30.245,
-      "eval_steps_per_second": 3.811,
+      "eval_loss": 0.1983761489391327,
+      "eval_runtime": 16.4512,
+      "eval_samples_per_second": 30.393,
+      "eval_steps_per_second": 3.83,
       "step": 720
     },
     {
       "epoch": 0.27785495971103086,
-      "grad_norm": 0.5945971608161926,
-      "learning_rate": 0.00039075669167361305,
-      "loss": 0.1986,
+      "grad_norm": 0.5706949830055237,
+      "learning_rate": 0.0003907659535056034,
+      "loss": 0.1959,
       "step": 1000
     },
     {
       "epoch": 0.4001111419838844,
-      "eval_loss": 0.17625106871128082,
-      "eval_runtime": 16.4823,
-      "eval_samples_per_second": 30.336,
+      "eval_loss": 0.17383727431297302,
+      "eval_runtime": 16.4842,
+      "eval_samples_per_second": 30.332,
       "eval_steps_per_second": 3.822,
       "step": 1440
     },
     {
       "epoch": 0.41678243956654626,
-      "grad_norm": 0.3619824945926666,
-      "learning_rate": 0.0003861257756784292,
-      "loss": 0.1802,
+      "grad_norm": 0.5392869710922241,
+      "learning_rate": 0.0003861350375104196,
+      "loss": 0.1723,
       "step": 1500
     },
     {
       "epoch": 0.5557099194220617,
-      "grad_norm": 0.40362638235092163,
-      "learning_rate": 0.00038149485968324534,
-      "loss": 0.1686,
+      "grad_norm": 0.5005412697792053,
+      "learning_rate": 0.0003815041215152357,
+      "loss": 0.1636,
       "step": 2000
     },
     {
       "epoch": 0.6001667129758266,
-      "eval_loss": 0.16156445443630219,
-      "eval_runtime": 16.3535,
-      "eval_samples_per_second": 30.575,
-      "eval_steps_per_second": 3.852,
+      "eval_loss": 0.16229495406150818,
+      "eval_runtime": 16.4963,
+      "eval_samples_per_second": 30.31,
+      "eval_steps_per_second": 3.819,
       "step": 2160
     },
     {
       "epoch": 0.6946373992775771,
-      "grad_norm": 0.4962793290615082,
-      "learning_rate": 0.0003768639436880615,
-      "loss": 0.1577,
+      "grad_norm": 0.370914489030838,
+      "learning_rate": 0.0003768732055200519,
+      "loss": 0.1539,
       "step": 2500
     },
     {
       "epoch": 0.8002222839677688,
-      "eval_loss": 0.15043224394321442,
-      "eval_runtime": 16.3325,
-      "eval_samples_per_second": 30.614,
-      "eval_steps_per_second": 3.857,
+      "eval_loss": 0.1537286937236786,
+      "eval_runtime": 16.4662,
+      "eval_samples_per_second": 30.365,
+      "eval_steps_per_second": 3.826,
       "step": 2880
     },
     {
       "epoch": 0.8335648791330925,
-      "grad_norm": 0.41315382719039917,
-      "learning_rate": 0.0003722330276928777,
-      "loss": 0.1493,
+      "grad_norm": 0.41378697752952576,
+      "learning_rate": 0.000372242289524868,
+      "loss": 0.1445,
       "step": 3000
     },
     {
       "epoch": 0.972492358988608,
-      "grad_norm": 0.37953054904937744,
+      "grad_norm": 0.4000154137611389,
       "learning_rate": 0.0003676113735296842,
-      "loss": 0.1477,
+      "loss": 0.1384,
       "step": 3500
     },
     {
       "epoch": 1.000277854959711,
-      "eval_loss": 0.14538082480430603,
-      "eval_runtime": 16.4466,
-      "eval_samples_per_second": 30.401,
-      "eval_steps_per_second": 3.831,
+      "eval_loss": 0.15115150809288025,
+      "eval_runtime": 16.8021,
+      "eval_samples_per_second": 29.758,
+      "eval_steps_per_second": 3.75,
       "step": 3600
     },
     {
       "epoch": 1.1114198388441234,
-      "grad_norm": 0.4146048128604889,
+      "grad_norm": 0.4774770140647888,
       "learning_rate": 0.00036298045753450036,
-      "loss": 0.1318,
+      "loss": 0.1317,
       "step": 4000
     },
     {
       "epoch": 1.2003334259516532,
-      "eval_loss": 0.13961325585842133,
-      "eval_runtime": 16.5835,
-      "eval_samples_per_second": 30.15,
-      "eval_steps_per_second": 3.799,
+      "eval_loss": 0.14330309629440308,
+      "eval_runtime": 16.4814,
+      "eval_samples_per_second": 30.337,
+      "eval_steps_per_second": 3.822,
       "step": 4320
     },
     {
       "epoch": 1.2503473186996388,
-      "grad_norm": 0.6914293766021729,
+      "grad_norm": 0.4988621175289154,
       "learning_rate": 0.0003583495415393165,
-      "loss": 0.1303,
+      "loss": 0.1283,
       "step": 4500
     },
     {
       "epoch": 1.3892747985551543,
-      "grad_norm": 0.39365154504776,
-      "learning_rate": 0.00035371862554413265,
-      "loss": 0.1269,
+      "grad_norm": 0.5688238739967346,
+      "learning_rate": 0.000353727887376123,
+      "loss": 0.1237,
       "step": 5000
     },
     {
       "epoch": 1.4003889969435954,
-      "eval_loss": 0.14090509712696075,
-      "eval_runtime": 16.4413,
-      "eval_samples_per_second": 30.411,
-      "eval_steps_per_second": 3.832,
+      "eval_loss": 0.13683326542377472,
+      "eval_runtime": 16.5065,
+      "eval_samples_per_second": 30.291,
+      "eval_steps_per_second": 3.817,
       "step": 5040
     },
     {
       "epoch": 1.5282022784106695,
-      "grad_norm": 0.5620034337043762,
+      "grad_norm": 0.5433902740478516,
       "learning_rate": 0.00034909697138093914,
-      "loss": 0.1263,
+      "loss": 0.1174,
       "step": 5500
     },
     {
       "epoch": 1.6004445679355377,
-      "eval_loss": 0.13119302690029144,
-      "eval_runtime": 16.5618,
-      "eval_samples_per_second": 30.19,
-      "eval_steps_per_second": 3.804,
+      "eval_loss": 0.13955478370189667,
+      "eval_runtime": 16.4989,
+      "eval_samples_per_second": 30.305,
+      "eval_steps_per_second": 3.818,
       "step": 5760
     },
     {
       "epoch": 1.667129758266185,
-      "grad_norm": 1.5726152658462524,
+      "grad_norm": 0.5300644040107727,
       "learning_rate": 0.0003444660553857553,
-      "loss": 0.1258,
+      "loss": 0.1178,
       "step": 6000
     },
     {
       "epoch": 1.8005001389274797,
-      "eval_loss": 0.12919241189956665,
-      "eval_runtime": 16.4031,
-      "eval_samples_per_second": 30.482,
-      "eval_steps_per_second": 3.841,
+      "eval_loss": 0.13158197700977325,
+      "eval_runtime": 16.533,
+      "eval_samples_per_second": 30.243,
+      "eval_steps_per_second": 3.811,
       "step": 6480
     },
     {
       "epoch": 1.8060572381217006,
-      "grad_norm": 0.9997938871383667,
+      "grad_norm": 0.3573897182941437,
       "learning_rate": 0.0003398351393905715,
-      "loss": 0.1212,
+      "loss": 0.113,
       "step": 6500
     },
     {
       "epoch": 1.9449847179772157,
-      "grad_norm": 0.4134954810142517,
+      "grad_norm": 0.5019258260726929,
       "learning_rate": 0.00033520422339538766,
-      "loss": 0.1201,
+      "loss": 0.1134,
       "step": 7000
     },
     {
       "epoch": 2.000555709919422,
-      "eval_loss": 0.12256824225187302,
-      "eval_runtime": 16.8158,
-      "eval_samples_per_second": 29.734,
-      "eval_steps_per_second": 3.746,
+      "eval_loss": 0.13124322891235352,
+      "eval_runtime": 16.0696,
+      "eval_samples_per_second": 31.115,
+      "eval_steps_per_second": 3.92,
       "step": 7200
     },
     {
       "epoch": 2.0839121978327313,
-      "grad_norm": 0.40068477392196655,
+      "grad_norm": 0.4890081286430359,
       "learning_rate": 0.0003305733074002038,
-      "loss": 0.108,
+      "loss": 0.1075,
       "step": 7500
     },
     {
       "epoch": 2.2006112809113643,
-      "eval_loss": 0.12798596918582916,
-      "eval_runtime": 16.5301,
-      "eval_samples_per_second": 30.248,
-      "eval_steps_per_second": 3.811,
+      "eval_loss": 0.12690122425556183,
+      "eval_runtime": 15.8697,
+      "eval_samples_per_second": 31.507,
+      "eval_steps_per_second": 3.97,
       "step": 7920
     },
     {
       "epoch": 2.222839677688247,
-      "grad_norm": 0.39021220803260803,
+      "grad_norm": 0.5411983132362366,
       "learning_rate": 0.0003259423914050199,
-      "loss": 0.1061,
+      "loss": 0.104,
       "step": 8000
     },
     {
       "epoch": 2.361767157543762,
-      "grad_norm": 0.4577147960662842,
-      "learning_rate": 0.0003213114754098361,
-      "loss": 0.1055,
+      "grad_norm": 0.892245352268219,
+      "learning_rate": 0.00032132073724182645,
+      "loss": 0.1018,
       "step": 8500
     },
     {
       "epoch": 2.4006668519033063,
-      "eval_loss": 0.1270737648010254,
-      "eval_runtime": 16.5227,
-      "eval_samples_per_second": 30.261,
-      "eval_steps_per_second": 3.813,
+      "eval_loss": 0.1253955215215683,
+      "eval_runtime": 15.9212,
+      "eval_samples_per_second": 31.405,
+      "eval_steps_per_second": 3.957,
       "step": 8640
     },
     {
       "epoch": 2.5006946373992776,
-      "grad_norm": 0.5035166144371033,
-      "learning_rate": 0.00031668055941465225,
-      "loss": 0.1033,
+      "grad_norm": 0.5154420137405396,
+      "learning_rate": 0.0003166898212466426,
+      "loss": 0.1018,
       "step": 9000
     },
     {
       "epoch": 2.600722422895249,
-      "eval_loss": 0.12219922989606857,
-      "eval_runtime": 16.637,
-      "eval_samples_per_second": 30.054,
-      "eval_steps_per_second": 3.787,
+      "eval_loss": 0.1270376443862915,
+      "eval_runtime": 15.8556,
+      "eval_samples_per_second": 31.535,
+      "eval_steps_per_second": 3.973,
       "step": 9360
     },
     {
       "epoch": 2.639622117254793,
-      "grad_norm": 0.6268121004104614,
-      "learning_rate": 0.00031205890525145874,
-      "loss": 0.1026,
+      "grad_norm": 0.4247698187828064,
+      "learning_rate": 0.0003120681670834491,
+      "loss": 0.0988,
       "step": 9500
     },
     {
       "epoch": 2.7785495971103087,
-      "grad_norm": 0.37843775749206543,
-      "learning_rate": 0.0003074279892562749,
-      "loss": 0.1042,
+      "grad_norm": 0.6174339652061462,
+      "learning_rate": 0.0003074372510882653,
+      "loss": 0.0931,
       "step": 10000
     },
     {
       "epoch": 2.800777993887191,
-      "eval_loss": 0.11693067103624344,
-      "eval_runtime": 16.4326,
-      "eval_samples_per_second": 30.427,
-      "eval_steps_per_second": 3.834,
+      "eval_loss": 0.12492711842060089,
+      "eval_runtime": 15.9536,
+      "eval_samples_per_second": 31.341,
+      "eval_steps_per_second": 3.949,
       "step": 10080
     },
     {
       "epoch": 2.917477076965824,
-      "grad_norm": 0.47878336906433105,
-      "learning_rate": 0.00030279707326109104,
-      "loss": 0.1032,
+      "grad_norm": 0.3945905864238739,
+      "learning_rate": 0.0003028063350930814,
+      "loss": 0.0924,
       "step": 10500
     },
     {
       "epoch": 3.000833564879133,
-      "eval_loss": 0.11544420570135117,
-      "eval_runtime": 16.6741,
-      "eval_samples_per_second": 29.987,
-      "eval_steps_per_second": 3.778,
+      "eval_loss": 0.12177152931690216,
+      "eval_runtime": 16.5123,
+      "eval_samples_per_second": 30.28,
+      "eval_steps_per_second": 3.815,
       "step": 10800
     },
     {
       "epoch": 3.0564045568213394,
-      "grad_norm": 0.6917030215263367,
-      "learning_rate": 0.0002981661572659072,
-      "loss": 0.0954,
+      "grad_norm": 0.4349508285522461,
+      "learning_rate": 0.0002981754190978976,
+      "loss": 0.0929,
       "step": 11000
     },
     {
       "epoch": 3.1953320366768545,
-      "grad_norm": 0.35634613037109375,
-      "learning_rate": 0.0002935352412707234,
-      "loss": 0.0874,
+      "grad_norm": 0.5195356011390686,
+      "learning_rate": 0.00029354450310271375,
+      "loss": 0.0897,
       "step": 11500
     },
     {
       "epoch": 3.2008891358710754,
-      "eval_loss": 0.1193525642156601,
-      "eval_runtime": 16.401,
-      "eval_samples_per_second": 30.486,
-      "eval_steps_per_second": 3.841,
+      "eval_loss": 0.12157219648361206,
+      "eval_runtime": 15.887,
+      "eval_samples_per_second": 31.472,
+      "eval_steps_per_second": 3.965,
       "step": 11520
     },
     {
       "epoch": 3.33425951653237,
-      "grad_norm": 0.4005107283592224,
-      "learning_rate": 0.0002889043252755395,
-      "loss": 0.0902,
+      "grad_norm": 0.38773760199546814,
+      "learning_rate": 0.0002889135871075299,
+      "loss": 0.0868,
       "step": 12000
     },
     {
       "epoch": 3.4009447068630174,
-      "eval_loss": 0.11315633356571198,
-      "eval_runtime": 16.525,
-      "eval_samples_per_second": 30.257,
-      "eval_steps_per_second": 3.812,
+      "eval_loss": 0.12406055629253387,
+      "eval_runtime": 15.9444,
+      "eval_samples_per_second": 31.359,
+      "eval_steps_per_second": 3.951,
       "step": 12240
     },
     {
       "epoch": 3.4731869963878856,
-      "grad_norm": 0.348311185836792,
-      "learning_rate": 0.0002842734092803557,
-      "loss": 0.0891,
+      "grad_norm": 0.3054683804512024,
+      "learning_rate": 0.00028428267111234605,
+      "loss": 0.0865,
       "step": 12500
     },
     {
       "epoch": 3.6010002778549595,
-      "eval_loss": 0.11703498661518097,
-      "eval_runtime": 16.4464,
-      "eval_samples_per_second": 30.402,
-      "eval_steps_per_second": 3.831,
+      "eval_loss": 0.11476034671068192,
+      "eval_runtime": 15.9006,
+      "eval_samples_per_second": 31.445,
+      "eval_steps_per_second": 3.962,
       "step": 12960
     },
     {
       "epoch": 3.612114476243401,
-      "grad_norm": 0.3483453094959259,
-      "learning_rate": 0.0002796424932851718,
-      "loss": 0.0892,
+      "grad_norm": 0.5311923623085022,
+      "learning_rate": 0.0002796610169491526,
+      "loss": 0.0845,
       "step": 13000
     },
     {
       "epoch": 3.7510419560989163,
-      "grad_norm": 0.8891735672950745,
-      "learning_rate": 0.00027502083912197834,
-      "loss": 0.0878,
+      "grad_norm": 0.7641647458076477,
+      "learning_rate": 0.0002750301009539687,
+      "loss": 0.084,
       "step": 13500
     },
     {
       "epoch": 3.801055848846902,
-      "eval_loss": 0.11358808726072311,
-      "eval_runtime": 16.5506,
-      "eval_samples_per_second": 30.21,
-      "eval_steps_per_second": 3.807,
+      "eval_loss": 0.11587072908878326,
+      "eval_runtime": 15.933,
+      "eval_samples_per_second": 31.381,
+      "eval_steps_per_second": 3.954,
       "step": 13680
     },
     {
       "epoch": 3.889969435954432,
-      "grad_norm": 2.020261764526367,
-      "learning_rate": 0.00027038992312679446,
-      "loss": 0.088,
+      "grad_norm": 0.5842312574386597,
+      "learning_rate": 0.00027039918495878483,
+      "loss": 0.0815,
       "step": 14000
     },
     {
       "epoch": 4.001111419838844,
-      "eval_loss": 0.11110712587833405,
-      "eval_runtime": 16.9629,
-      "eval_samples_per_second": 29.476,
-      "eval_steps_per_second": 3.714,
+      "eval_loss": 0.11761430650949478,
+      "eval_runtime": 16.0803,
+      "eval_samples_per_second": 31.094,
+      "eval_steps_per_second": 3.918,
       "step": 14400
     },
     {
       "epoch": 4.0288969158099475,
-      "grad_norm": 0.4606301486492157,
-      "learning_rate": 0.00026575900713161064,
-      "loss": 0.0848,
+      "grad_norm": 0.5182059407234192,
+      "learning_rate": 0.000265768268963601,
+      "loss": 0.0823,
       "step": 14500
     },
     {
       "epoch": 4.167824395665463,
-      "grad_norm": 0.5307084321975708,
-      "learning_rate": 0.0002611280911364268,
-      "loss": 0.0754,
+      "grad_norm": 0.3954576253890991,
+      "learning_rate": 0.0002611373529684172,
+      "loss": 0.0753,
       "step": 15000
     },
     {
       "epoch": 4.201166990830786,
-      "eval_loss": 0.11462360620498657,
-      "eval_runtime": 16.4785,
-      "eval_samples_per_second": 30.343,
-      "eval_steps_per_second": 3.823,
+      "eval_loss": 0.11391445249319077,
+      "eval_runtime": 15.9483,
+      "eval_samples_per_second": 31.351,
+      "eval_steps_per_second": 3.95,
       "step": 15120
     },
     {
       "epoch": 4.306751875520978,
-      "grad_norm": 0.45674943923950195,
-      "learning_rate": 0.000256497175141243,
-      "loss": 0.076,
+      "grad_norm": 0.5974435210227966,
+      "learning_rate": 0.00025650643697323335,
+      "loss": 0.0762,
       "step": 15500
     },
     {
       "epoch": 4.4012225618227285,
-      "eval_loss": 0.11145643889904022,
-      "eval_runtime": 16.4715,
-      "eval_samples_per_second": 30.356,
-      "eval_steps_per_second": 3.825,
+      "eval_loss": 0.11403658986091614,
+      "eval_runtime": 15.92,
+      "eval_samples_per_second": 31.407,
+      "eval_steps_per_second": 3.957,
       "step": 15840
     },
     {
       "epoch": 4.445679355376494,
-      "grad_norm": 0.3333616554737091,
-      "learning_rate": 0.00025186625914605905,
-      "loss": 0.0768,
+      "grad_norm": 0.4496535360813141,
+      "learning_rate": 0.0002518755209780495,
+      "loss": 0.0737,
       "step": 16000
     },
     {
       "epoch": 4.584606835232009,
-      "grad_norm": 0.3716529905796051,
+      "grad_norm": 0.5617558360099792,
       "learning_rate": 0.0002472446049828656,
-      "loss": 0.0762,
+      "loss": 0.074,
       "step": 16500
     },
     {
       "epoch": 4.601278132814671,
-      "eval_loss": 0.1125354915857315,
-      "eval_runtime": 16.4547,
-      "eval_samples_per_second": 30.386,
-      "eval_steps_per_second": 3.829,
+      "eval_loss": 0.11306341737508774,
+      "eval_runtime": 15.9244,
+      "eval_samples_per_second": 31.398,
+      "eval_steps_per_second": 3.956,
       "step": 16560
     },
     {
       "epoch": 4.723534315087524,
-      "grad_norm": 0.5058280229568481,
+      "grad_norm": 0.5999208092689514,
       "learning_rate": 0.00024261368898768177,
-      "loss": 0.0771,
+      "loss": 0.0732,
       "step": 17000
     },
     {
       "epoch": 4.801333703806613,
-      "eval_loss": 0.10654650628566742,
-      "eval_runtime": 16.5091,
-      "eval_samples_per_second": 30.286,
-      "eval_steps_per_second": 3.816,
+      "eval_loss": 0.11077062785625458,
+      "eval_runtime": 15.9311,
+      "eval_samples_per_second": 31.385,
+      "eval_steps_per_second": 3.955,
       "step": 17280
     },
     {
       "epoch": 4.86246179494304,
-      "grad_norm": 0.37022796273231506,
+      "grad_norm": 0.3961442708969116,
       "learning_rate": 0.0002379920348244883,
-      "loss": 0.0763,
+      "loss": 0.0724,
       "step": 17500
     },
     {
       "epoch": 5.001389274798555,
-      "grad_norm": 0.3737078607082367,
+      "grad_norm": 0.507563054561615,
       "learning_rate": 0.00023336111882930443,
-      "loss": 0.0749,
+      "loss": 0.0685,
       "step": 18000
     },
     {
       "epoch": 5.001389274798555,
-      "eval_loss": 0.10845010727643967,
-      "eval_runtime": 16.6489,
-      "eval_samples_per_second": 30.032,
-      "eval_steps_per_second": 3.784,
+      "eval_loss": 0.11523561179637909,
+      "eval_runtime": 16.1329,
+      "eval_samples_per_second": 30.992,
+      "eval_steps_per_second": 3.905,
       "step": 18000
     },
     {
       "epoch": 5.14031675465407,
-      "grad_norm": 0.3841913640499115,
+      "grad_norm": 0.5651789307594299,
       "learning_rate": 0.00022873020283412058,
-      "loss": 0.0642,
+      "loss": 0.0655,
       "step": 18500
     },
     {
       "epoch": 5.201444845790498,
-      "eval_loss": 0.11235029250383377,
-      "eval_runtime": 16.682,
-      "eval_samples_per_second": 29.972,
-      "eval_steps_per_second": 3.777,
+      "eval_loss": 0.11398093402385712,
+      "eval_runtime": 15.9337,
+      "eval_samples_per_second": 31.38,
+      "eval_steps_per_second": 3.954,
       "step": 18720
     },
     {
       "epoch": 5.279244234509586,
-      "grad_norm": 0.31463760137557983,
+      "grad_norm": 0.619132399559021,
       "learning_rate": 0.00022409928683893675,
-      "loss": 0.0649,
+      "loss": 0.0664,
       "step": 19000
     },
     {
       "epoch": 5.401500416782439,
-      "eval_loss": 0.1092103123664856,
-      "eval_runtime": 16.4715,
-      "eval_samples_per_second": 30.355,
-      "eval_steps_per_second": 3.825,
+      "eval_loss": 0.11203750967979431,
+      "eval_runtime": 15.8871,
+      "eval_samples_per_second": 31.472,
+      "eval_steps_per_second": 3.965,
       "step": 19440
     },
     {
       "epoch": 5.418171714365101,
-      "grad_norm": 0.30956658720970154,
+      "grad_norm": 0.4724760353565216,
       "learning_rate": 0.0002194683708437529,
-      "loss": 0.0651,
+      "loss": 0.0636,
       "step": 19500
     },
     {
       "epoch": 5.5570991942206165,
-      "grad_norm": 0.42891213297843933,
+      "grad_norm": 0.5861866474151611,
       "learning_rate": 0.00021483745484856908,
-      "loss": 0.0656,
+      "loss": 0.0648,
       "step": 20000
     },
     {
       "epoch": 5.601555987774382,
-      "eval_loss": 0.1072624996304512,
-      "eval_runtime": 16.504,
-      "eval_samples_per_second": 30.296,
-      "eval_steps_per_second": 3.817,
+      "eval_loss": 0.1131061241030693,
+      "eval_runtime": 15.952,
+      "eval_samples_per_second": 31.344,
+      "eval_steps_per_second": 3.949,
       "step": 20160
     },
     {
       "epoch": 5.6960266740761325,
-      "grad_norm": 0.6161347031593323,
-      "learning_rate": 0.00021021580068537556,
-      "loss": 0.0673,
+      "grad_norm": 0.5262423157691956,
+      "learning_rate": 0.0002102065388533852,
+      "loss": 0.063,
       "step": 20500
     },
     {
       "epoch": 5.801611558766324,
-      "eval_loss": 0.10548041760921478,
-      "eval_runtime": 16.5903,
-      "eval_samples_per_second": 30.138,
-      "eval_steps_per_second": 3.797,
+      "eval_loss": 0.11248422414064407,
+      "eval_runtime": 15.9153,
+      "eval_samples_per_second": 31.416,
+      "eval_steps_per_second": 3.958,
       "step": 20880
     },
     {
       "epoch": 5.834954153931648,
-      "grad_norm": 0.6631760597229004,
-      "learning_rate": 0.0002055848846901917,
-      "loss": 0.0654,
+      "grad_norm": 0.6074294447898865,
+      "learning_rate": 0.00020557562285820135,
+      "loss": 0.0624,
       "step": 21000
     },
     {
       "epoch": 5.973881633787163,
-      "grad_norm": 0.37077194452285767,
+      "grad_norm": 0.5349674820899963,
       "learning_rate": 0.0002009539686950079,
-      "loss": 0.065,
+      "loss": 0.0609,
       "step": 21500
     },
     {
       "epoch": 6.001667129758266,
-      "eval_loss": 0.10457777976989746,
-      "eval_runtime": 16.8365,
-      "eval_samples_per_second": 29.697,
-      "eval_steps_per_second": 3.742,
+      "eval_loss": 0.11405794322490692,
+      "eval_runtime": 16.1656,
+      "eval_samples_per_second": 30.93,
+      "eval_steps_per_second": 3.897,
       "step": 21600
     },
     {
       "epoch": 6.112809113642679,
-      "grad_norm": 0.4291210174560547,
+      "grad_norm": 0.36713194847106934,
       "learning_rate": 0.00019632305269982403,
-      "loss": 0.0565,
+      "loss": 0.0576,
       "step": 22000
     },
     {
       "epoch": 6.201722700750208,
-      "eval_loss": 0.11106568574905396,
-      "eval_runtime": 16.6145,
-      "eval_samples_per_second": 30.094,
-      "eval_steps_per_second": 3.792,
+      "eval_loss": 0.11051186919212341,
+      "eval_runtime": 15.9395,
+      "eval_samples_per_second": 31.369,
+      "eval_steps_per_second": 3.952,
       "step": 22320
     },
     {
       "epoch": 6.251736593498194,
-      "grad_norm": 0.5859522223472595,
+      "grad_norm": 0.4714512526988983,
       "learning_rate": 0.00019169213670464018,
-      "loss": 0.0551,
+      "loss": 0.057,
       "step": 22500
     },
     {
       "epoch": 6.390664073353709,
-      "grad_norm": 0.6046123504638672,
+      "grad_norm": 0.3419685363769531,
       "learning_rate": 0.00018706122070945633,
-      "loss": 0.0566,
+      "loss": 0.0572,
       "step": 23000
     },
     {
       "epoch": 6.401778271742151,
-      "eval_loss": 0.11063603311777115,
-      "eval_runtime": 16.5709,
-      "eval_samples_per_second": 30.173,
-      "eval_steps_per_second": 3.802,
+      "eval_loss": 0.1142740249633789,
+      "eval_runtime": 15.9346,
+      "eval_samples_per_second": 31.378,
+      "eval_steps_per_second": 3.954,
       "step": 23040
     },
     {
       "epoch": 6.529591553209225,
-      "grad_norm": 0.5555618405342102,
+      "grad_norm": 0.43148958683013916,
       "learning_rate": 0.0001824303047142725,
-      "loss": 0.0567,
+      "loss": 0.0554,
       "step": 23500
     },
     {
       "epoch": 6.601833842734093,
-      "eval_loss": 0.11322695761919022,
-      "eval_runtime": 16.5489,
-      "eval_samples_per_second": 30.213,
-      "eval_steps_per_second": 3.807,
+      "eval_loss": 0.1115042194724083,
+      "eval_runtime": 15.9938,
+      "eval_samples_per_second": 31.262,
+      "eval_steps_per_second": 3.939,
       "step": 23760
     },
     {
       "epoch": 6.66851903306474,
-      "grad_norm": 0.641233503818512,
+      "grad_norm": 0.5623305439949036,
       "learning_rate": 0.00017779938871908863,
-      "loss": 0.057,
+      "loss": 0.0538,
       "step": 24000
     },
     {
       "epoch": 6.801889413726035,
-      "eval_loss": 0.10657580196857452,
-      "eval_runtime": 16.5793,
-      "eval_samples_per_second": 30.158,
-      "eval_steps_per_second": 3.8,
+      "eval_loss": 0.11134042590856552,
+      "eval_runtime": 15.9452,
+      "eval_samples_per_second": 31.357,
+      "eval_steps_per_second": 3.951,
       "step": 24480
     },
     {
       "epoch": 6.807446512920255,
-      "grad_norm": 0.36209815740585327,
+      "grad_norm": 0.4066413640975952,
       "learning_rate": 0.0001731684727239048,
-      "loss": 0.0574,
+      "loss": 0.0534,
       "step": 24500
     },
     {
       "epoch": 6.946373992775771,
-      "grad_norm": 0.49614134430885315,
+      "grad_norm": 0.4021354019641876,
       "learning_rate": 0.00016853755672872095,
-      "loss": 0.0564,
+      "loss": 0.052,
       "step": 25000
     },
     {
       "epoch": 7.001944984717977,
-      "eval_loss": 0.10740524530410767,
-      "eval_runtime": 16.6216,
-      "eval_samples_per_second": 30.081,
-      "eval_steps_per_second": 3.79,
+      "eval_loss": 0.1132456511259079,
+      "eval_runtime": 16.1483,
+      "eval_samples_per_second": 30.963,
+      "eval_steps_per_second": 3.901,
       "step": 25200
     },
     {
       "epoch": 7.085301472631286,
-      "grad_norm": 0.518583357334137,
-      "learning_rate": 0.0001639066407335371,
-      "loss": 0.049,
+      "grad_norm": 0.46669623255729675,
+      "learning_rate": 0.00016391590256552746,
+      "loss": 0.0498,
       "step": 25500
     },
     {
       "epoch": 7.20200055570992,
-      "eval_loss": 0.11705406755208969,
-      "eval_runtime": 16.5295,
-      "eval_samples_per_second": 30.249,
-      "eval_steps_per_second": 3.811,
+      "eval_loss": 0.11319959908723831,
+      "eval_runtime": 15.9896,
+      "eval_samples_per_second": 31.27,
+      "eval_steps_per_second": 3.94,
       "step": 25920
     },
     {
       "epoch": 7.2242289524868015,
-      "grad_norm": 0.31886982917785645,
+      "grad_norm": 0.2583458125591278,
       "learning_rate": 0.00015928498657034364,
-      "loss": 0.0469,
+      "loss": 0.0487,
       "step": 26000
     },
     {
       "epoch": 7.3631564323423175,
-      "grad_norm": 0.28966373205184937,
+      "grad_norm": 0.3225070536136627,
       "learning_rate": 0.00015465407057515976,
-      "loss": 0.0478,
+      "loss": 0.0485,
       "step": 26500
     },
     {
       "epoch": 7.402056126701861,
-      "eval_loss": 0.10964089632034302,
-      "eval_runtime": 16.5451,
-      "eval_samples_per_second": 30.22,
-      "eval_steps_per_second": 3.808,
+      "eval_loss": 0.11151115596294403,
+      "eval_runtime": 15.9848,
+      "eval_samples_per_second": 31.28,
+      "eval_steps_per_second": 3.941,
       "step": 26640
     },
     {
       "epoch": 7.502083912197833,
-      "grad_norm": 0.36664408445358276,
-      "learning_rate": 0.0001500324164119663,
-      "loss": 0.0481,
+      "grad_norm": 0.4772126376628876,
+      "learning_rate": 0.00015002315457997593,
+      "loss": 0.0483,
       "step": 27000
     },
     {
       "epoch": 7.602111697693804,
-      "eval_loss": 0.11513197422027588,
-      "eval_runtime": 16.6147,
-      "eval_samples_per_second": 30.094,
-      "eval_steps_per_second": 3.792,
+      "eval_loss": 0.11146976053714752,
+      "eval_runtime": 15.9333,
+      "eval_samples_per_second": 31.381,
+      "eval_steps_per_second": 3.954,
       "step": 27360
     },
     {
       "epoch": 7.641011392053348,
-      "grad_norm": 0.32852914929389954,
+      "grad_norm": 0.7825577855110168,
       "learning_rate": 0.00014540150041678245,
-      "loss": 0.048,
+      "loss": 0.0478,
       "step": 27500
     },
     {
       "epoch": 7.779938871908864,
-      "grad_norm": 0.5736936330795288,
+      "grad_norm": 0.465191513299942,
       "learning_rate": 0.0001407705844215986,
-      "loss": 0.0483,
+      "loss": 0.0469,
       "step": 28000
     },
     {
       "epoch": 7.802167268685746,
-      "eval_loss": 0.10632374882698059,
-      "eval_runtime": 16.6173,
-      "eval_samples_per_second": 30.089,
-      "eval_steps_per_second": 3.791,
+      "eval_loss": 0.11259140819311142,
+      "eval_runtime": 15.9752,
+      "eval_samples_per_second": 31.299,
+      "eval_steps_per_second": 3.944,
       "step": 28080
     },
     {
       "epoch": 7.918866351764379,
-      "grad_norm": 0.335509717464447,
+      "grad_norm": 0.2701134979724884,
       "learning_rate": 0.00013613966842641474,
-      "loss": 0.0487,
+      "loss": 0.0443,
       "step": 28500
     },
     {
       "epoch": 8.002222839677689,
-      "eval_loss": 0.11238289624452591,
-      "eval_runtime": 16.6415,
-      "eval_samples_per_second": 30.045,
-      "eval_steps_per_second": 3.786,
+      "eval_loss": 0.11337699741125107,
+      "eval_runtime": 16.0812,
+      "eval_samples_per_second": 31.092,
+      "eval_steps_per_second": 3.918,
       "step": 28800
     },
     {
       "epoch": 8.057793831619895,
-      "grad_norm": 0.8646371364593506,
-      "learning_rate": 0.00013150875243123092,
-      "loss": 0.0442,
+      "grad_norm": 0.3994615375995636,
+      "learning_rate": 0.00013151801426322128,
+      "loss": 0.044,
       "step": 29000
     },
     {
       "epoch": 8.19672131147541,
-      "grad_norm": 0.5025931000709534,
-      "learning_rate": 0.00012687783643604704,
-      "loss": 0.0411,
+      "grad_norm": 0.46412038803100586,
+      "learning_rate": 0.00012688709826803743,
+      "loss": 0.0421,
       "step": 29500
     },
     {
       "epoch": 8.20227841066963,
-      "eval_loss": 0.1167159229516983,
-      "eval_runtime": 16.2769,
-      "eval_samples_per_second": 30.718,
-      "eval_steps_per_second": 3.871,
+      "eval_loss": 0.11495082080364227,
+      "eval_runtime": 15.9654,
+      "eval_samples_per_second": 31.318,
+      "eval_steps_per_second": 3.946,
       "step": 29520
     },
     {
       "epoch": 8.335648791330925,
-      "grad_norm": 0.3653465509414673,
+      "grad_norm": 0.3810461759567261,
       "learning_rate": 0.00012225618227285358,
-      "loss": 0.0398,
+      "loss": 0.0411,
       "step": 30000
     },
     {
       "epoch": 8.402333981661572,
-      "eval_loss": 0.11505118012428284,
-      "eval_runtime": 16.3723,
-      "eval_samples_per_second": 30.539,
-      "eval_steps_per_second": 3.848,
+      "eval_loss": 0.11439384520053864,
+      "eval_runtime": 16.014,
+      "eval_samples_per_second": 31.223,
+      "eval_steps_per_second": 3.934,
       "step": 30240
     },
     {
       "epoch": 8.474576271186441,
-      "grad_norm": 0.4740363359451294,
+      "grad_norm": 0.4397641122341156,
       "learning_rate": 0.00011762526627766973,
-      "loss": 0.0401,
+      "loss": 0.0412,
       "step": 30500
     },
     {
       "epoch": 8.602389552653515,
-      "eval_loss": 0.11725138872861862,
-      "eval_runtime": 16.2874,
-      "eval_samples_per_second": 30.699,
-      "eval_steps_per_second": 3.868,
+      "eval_loss": 0.11167102307081223,
+      "eval_runtime": 15.9143,
+      "eval_samples_per_second": 31.418,
+      "eval_steps_per_second": 3.959,
       "step": 30960
     },
     {
       "epoch": 8.613503751041955,
-      "grad_norm": 0.4850703775882721,
-      "learning_rate": 0.00011299435028248589,
-      "loss": 0.0411,
+      "grad_norm": 0.7023443579673767,
+      "learning_rate": 0.00011300361211447625,
+      "loss": 0.041,
       "step": 31000
     },
     {
       "epoch": 8.752431230897471,
-      "grad_norm": 0.4242146611213684,
-      "learning_rate": 0.00010836343428730202,
-      "loss": 0.0411,
+      "grad_norm": 0.5792316198348999,
+      "learning_rate": 0.00010837269611929239,
+      "loss": 0.0391,
       "step": 31500
     },
     {
       "epoch": 8.802445123645457,
-      "eval_loss": 0.11391962319612503,
-      "eval_runtime": 16.5683,
-      "eval_samples_per_second": 30.178,
-      "eval_steps_per_second": 3.802,
+      "eval_loss": 0.11271476745605469,
+      "eval_runtime": 15.9859,
+      "eval_samples_per_second": 31.278,
+      "eval_steps_per_second": 3.941,
       "step": 31680
     },
     {
       "epoch": 8.891358710752987,
-      "grad_norm": 0.44564756751060486,
-      "learning_rate": 0.00010373251829211818,
-      "loss": 0.0399,
+      "grad_norm": 0.44151026010513306,
+      "learning_rate": 0.00010374178012410855,
+      "loss": 0.0403,
       "step": 32000
     }
   ],
@@ -826,7 +826,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 6.055901979589214e+18,
+  "total_flos": 6.074020652791103e+18,
   "train_batch_size": 4,
   "trial_name": null,
   "trial_params": null

checkpoint-32391/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:58052fbf3a2b07f2a6024b5cc28db88f1f0e48109a11483aa716d00657e9906e
+oid sha256:b2784cc9ebb113293b9d7c5af564dbf5463d67b520bf149c8840105fec4706ec
 size 5496