Training in progress, epoch 5, checkpoint

checkpoint-17995/adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:630a0b512ff09ec08378e175168045fd6dcd189b7ff4a12af8bd39d13d049daf
+oid sha256:188effeb8dc288e9df8d1c3f74f8619794ab5f0e21a02c26c492f9e53fe7db6d
 size 541459256

checkpoint-17995/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a78fbf4aec89f39dd28ca40b2973689e32045448fec6a3184fac5a6b8236f5aa
+oid sha256:02ee2648e66d09dbf88a9e4bd9f148bdd60a4e6ac5c2d99f1e5b6c16e63ac8e2
 size 33662074

checkpoint-17995/rng_state_0.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d313648e532b8d7a5f40a7f2e7380280fcc694ceca79b23f7de7e2210103497e
+oid sha256:0b0ada917fbf60666fe9208402628de9e50a0dc268ea796d31be5cbf56c844fa
 size 15024

checkpoint-17995/rng_state_1.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:26df90e0032f6bce8866cb076ac8af38b78c99bb13fffff021d7073ff7ad1b05
+oid sha256:b7f775019ed91c13a8840123984cf5b2622f67ddd877a7b8588b0f9f7bd67ef1
 size 15024

checkpoint-17995/rng_state_2.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:426dce54c6df13193cca221e3a69f6671bded5ae4f4d0703684a20d6e925989c
+oid sha256:edf75191176557363958c966b72042fce1c22fe4676902ec6638b02d9b331bd8
 size 15024

checkpoint-17995/rng_state_3.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2df45d2d95621c0076ebc7c031847020ffcfcfa280221dbd052332b1404d72a5
+oid sha256:1984736499222fd98d49ab6c9fae66179ee6f8516b2aea6bb474dd6a96db7f52
 size 15024

checkpoint-17995/trainer_state.json

@@ -10,439 +10,439 @@
   "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
     }
   ],
@@ -463,7 +463,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 3.364036190461231e+18,
+  "total_flos": 3.3736461282464236e+18,
   "train_batch_size": 4,
   "trial_name": null,
   "trial_params": null

checkpoint-17995/training_args.bin

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