Training in progress, step 2500, checkpoint

checkpoint-2500/model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7e1d70425befb55c50aebf311762d5d4cf1178ad01bcb687111ba011b0a8c67f
+oid sha256:268a37a59eccf7246161641edc8144a419cdb9521e57aa2ede9307ce0e0a9251
 size 577789320

checkpoint-2500/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2fc76428f4ea70d440463c2ce7e7c401cba290e14550823d02eff7b0ff8e9173
+oid sha256:a4cb884a52eaeac9587013d61160a29a95236c5f2bfc9c71b126e0f3216a3943
 size 1155772233

checkpoint-2500/rng_state.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d4a3cb642536620c029c848ed8f7ad7d549e7a33785e73fb5eba6ad51f6f5eb9
+oid sha256:ca70df9171e4f867e64d5a8090320e44c1ff252fb2768b278d5e6edad58c47f3
 size 14244

checkpoint-2500/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ba68065c0ba5863ea7a5974f2a119aa320d107d98a915a0d39227618f3cb2905
+oid sha256:f21e44438b98faf063216590a5534fa1899dfb57fd798660776c8a525c03fb28
 size 1064

checkpoint-2500/trainer_state.json

@@ -1,7 +1,7 @@
 {
-  "best_metric": 0.41558554768562317,
+  "best_metric": 0.4210895597934723,
   "best_model_checkpoint": "mikhail-panzo/zlm_b64_le5_s8000/checkpoint-2500",
-  "epoch": 2.094240837696335,
+  "epoch": 2.0938023450586263,
   "eval_steps": 500,
   "global_step": 2500,
   "is_hyper_param_search": false,
@@ -9,393 +9,393 @@
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.041884816753926704,
-      "grad_norm": 11.218673706054688,
-      "learning_rate": 2.4500000000000004e-07,
-      "loss": 1.1196,
+      "epoch": 0.04187604690117253,
+      "grad_norm": 5.415271759033203,
+      "learning_rate": 2.4000000000000003e-07,
+      "loss": 1.1819,
       "step": 50
     },
     {
-      "epoch": 0.08376963350785341,
-      "grad_norm": 5.423076629638672,
-      "learning_rate": 4.95e-07,
-      "loss": 1.057,
+      "epoch": 0.08375209380234507,
+      "grad_norm": 16.256921768188477,
+      "learning_rate": 4.900000000000001e-07,
+      "loss": 1.0035,
       "step": 100
     },
     {
-      "epoch": 0.1256544502617801,
-      "grad_norm": 12.713382720947266,
-      "learning_rate": 7.4e-07,
-      "loss": 1.0037,
+      "epoch": 0.12562814070351758,
+      "grad_norm": 3.2439260482788086,
+      "learning_rate": 7.350000000000001e-07,
+      "loss": 0.9742,
       "step": 150
     },
     {
-      "epoch": 0.16753926701570682,
-      "grad_norm": 4.434875965118408,
-      "learning_rate": 9.9e-07,
-      "loss": 0.9381,
+      "epoch": 0.16750418760469013,
+      "grad_norm": 4.24526834487915,
+      "learning_rate": 9.85e-07,
+      "loss": 0.9361,
       "step": 200
     },
     {
-      "epoch": 0.2094240837696335,
-      "grad_norm": 3.668553590774536,
-      "learning_rate": 1.2400000000000002e-06,
-      "loss": 0.8591,
+      "epoch": 0.20938023450586266,
+      "grad_norm": 3.625537872314453,
+      "learning_rate": 1.235e-06,
+      "loss": 0.8733,
       "step": 250
     },
     {
-      "epoch": 0.2513089005235602,
-      "grad_norm": 2.2456483840942383,
-      "learning_rate": 1.4900000000000001e-06,
-      "loss": 0.8262,
+      "epoch": 0.25125628140703515,
+      "grad_norm": 4.1455464363098145,
+      "learning_rate": 1.485e-06,
+      "loss": 0.8471,
       "step": 300
     },
     {
-      "epoch": 0.2931937172774869,
-      "grad_norm": 2.70170259475708,
-      "learning_rate": 1.74e-06,
-      "loss": 0.7634,
+      "epoch": 0.2931323283082077,
+      "grad_norm": 3.164158582687378,
+      "learning_rate": 1.7350000000000001e-06,
+      "loss": 0.838,
       "step": 350
     },
     {
-      "epoch": 0.33507853403141363,
-      "grad_norm": 2.9946064949035645,
-      "learning_rate": 1.9900000000000004e-06,
-      "loss": 0.7336,
+      "epoch": 0.33500837520938026,
+      "grad_norm": 4.24931001663208,
+      "learning_rate": 1.985e-06,
+      "loss": 0.7467,
       "step": 400
     },
     {
-      "epoch": 0.3769633507853403,
-      "grad_norm": 2.2950398921966553,
-      "learning_rate": 2.24e-06,
-      "loss": 0.7121,
+      "epoch": 0.3768844221105528,
+      "grad_norm": 2.332688331604004,
+      "learning_rate": 2.235e-06,
+      "loss": 0.7389,
       "step": 450
     },
     {
-      "epoch": 0.418848167539267,
-      "grad_norm": 8.007049560546875,
-      "learning_rate": 2.4900000000000003e-06,
-      "loss": 0.6879,
+      "epoch": 0.4187604690117253,
+      "grad_norm": 2.370961904525757,
+      "learning_rate": 2.4850000000000003e-06,
+      "loss": 0.7074,
       "step": 500
     },
     {
-      "epoch": 0.418848167539267,
-      "eval_loss": 0.5903987884521484,
-      "eval_runtime": 258.509,
-      "eval_samples_per_second": 32.838,
-      "eval_steps_per_second": 4.108,
+      "epoch": 0.4187604690117253,
+      "eval_loss": 0.6029486060142517,
+      "eval_runtime": 220.2253,
+      "eval_samples_per_second": 38.547,
+      "eval_steps_per_second": 4.822,
       "step": 500
     },
     {
-      "epoch": 0.4607329842931937,
-      "grad_norm": 3.0767672061920166,
-      "learning_rate": 2.7400000000000004e-06,
-      "loss": 0.6734,
+      "epoch": 0.46063651591289784,
+      "grad_norm": 14.466177940368652,
+      "learning_rate": 2.7350000000000005e-06,
+      "loss": 0.6738,
       "step": 550
     },
     {
-      "epoch": 0.5026178010471204,
-      "grad_norm": 8.150467872619629,
-      "learning_rate": 2.99e-06,
-      "loss": 0.6539,
+      "epoch": 0.5025125628140703,
+      "grad_norm": 3.282449722290039,
+      "learning_rate": 2.9850000000000002e-06,
+      "loss": 0.6517,
       "step": 600
     },
     {
-      "epoch": 0.5445026178010471,
-      "grad_norm": 2.2473220825195312,
-      "learning_rate": 3.2400000000000003e-06,
-      "loss": 0.657,
+      "epoch": 0.5443886097152428,
+      "grad_norm": 2.621645212173462,
+      "learning_rate": 3.2350000000000004e-06,
+      "loss": 0.6479,
       "step": 650
     },
     {
-      "epoch": 0.5863874345549738,
-      "grad_norm": 2.448143482208252,
-      "learning_rate": 3.49e-06,
-      "loss": 0.6258,
+      "epoch": 0.5862646566164154,
+      "grad_norm": 2.4493019580841064,
+      "learning_rate": 3.485e-06,
+      "loss": 0.6321,
       "step": 700
     },
     {
-      "epoch": 0.6282722513089005,
-      "grad_norm": 4.810636520385742,
-      "learning_rate": 3.74e-06,
-      "loss": 0.6155,
+      "epoch": 0.628140703517588,
+      "grad_norm": 1.919048547744751,
+      "learning_rate": 3.7350000000000002e-06,
+      "loss": 0.6407,
       "step": 750
     },
     {
-      "epoch": 0.6701570680628273,
-      "grad_norm": 2.5927581787109375,
-      "learning_rate": 3.990000000000001e-06,
-      "loss": 0.6073,
+      "epoch": 0.6700167504187605,
+      "grad_norm": 2.8260602951049805,
+      "learning_rate": 3.985000000000001e-06,
+      "loss": 0.6279,
       "step": 800
     },
     {
-      "epoch": 0.7120418848167539,
-      "grad_norm": 1.913765549659729,
-      "learning_rate": 4.24e-06,
-      "loss": 0.6118,
+      "epoch": 0.711892797319933,
+      "grad_norm": 3.333841323852539,
+      "learning_rate": 4.235e-06,
+      "loss": 0.6143,
       "step": 850
     },
     {
-      "epoch": 0.7539267015706806,
-      "grad_norm": 2.210376262664795,
-      "learning_rate": 4.49e-06,
-      "loss": 0.5807,
+      "epoch": 0.7537688442211056,
+      "grad_norm": 1.7883094549179077,
+      "learning_rate": 4.485e-06,
+      "loss": 0.5949,
       "step": 900
     },
     {
-      "epoch": 0.7958115183246073,
-      "grad_norm": 3.323096752166748,
-      "learning_rate": 4.74e-06,
-      "loss": 0.5915,
+      "epoch": 0.7956448911222781,
+      "grad_norm": 2.48634934425354,
+      "learning_rate": 4.735e-06,
+      "loss": 0.5875,
       "step": 950
     },
     {
-      "epoch": 0.837696335078534,
-      "grad_norm": 2.1455421447753906,
-      "learning_rate": 4.9900000000000005e-06,
-      "loss": 0.5606,
+      "epoch": 0.8375209380234506,
+      "grad_norm": 5.4277873039245605,
+      "learning_rate": 4.9850000000000006e-06,
+      "loss": 0.5916,
       "step": 1000
     },
     {
-      "epoch": 0.837696335078534,
-      "eval_loss": 0.49923568964004517,
-      "eval_runtime": 260.1008,
-      "eval_samples_per_second": 32.637,
-      "eval_steps_per_second": 4.083,
+      "epoch": 0.8375209380234506,
+      "eval_loss": 0.49684029817581177,
+      "eval_runtime": 219.2621,
+      "eval_samples_per_second": 38.716,
+      "eval_steps_per_second": 4.844,
       "step": 1000
     },
     {
-      "epoch": 0.8795811518324608,
-      "grad_norm": 2.7230141162872314,
-      "learning_rate": 5.240000000000001e-06,
-      "loss": 0.5731,
+      "epoch": 0.8793969849246231,
+      "grad_norm": 2.675995349884033,
+      "learning_rate": 5.235e-06,
+      "loss": 0.5676,
       "step": 1050
     },
     {
-      "epoch": 0.9214659685863874,
-      "grad_norm": 2.0273685455322266,
-      "learning_rate": 5.490000000000001e-06,
-      "loss": 0.5361,
+      "epoch": 0.9212730318257957,
+      "grad_norm": 2.507474422454834,
+      "learning_rate": 5.485e-06,
+      "loss": 0.5535,
       "step": 1100
     },
     {
-      "epoch": 0.9633507853403142,
-      "grad_norm": 3.9353837966918945,
-      "learning_rate": 5.74e-06,
-      "loss": 0.5476,
+      "epoch": 0.9631490787269682,
+      "grad_norm": 5.360454082489014,
+      "learning_rate": 5.735e-06,
+      "loss": 0.569,
       "step": 1150
     },
     {
-      "epoch": 1.0052356020942408,
-      "grad_norm": 2.7838985919952393,
-      "learning_rate": 5.99e-06,
-      "loss": 0.5356,
+      "epoch": 1.0050251256281406,
+      "grad_norm": 2.9445533752441406,
+      "learning_rate": 5.985000000000001e-06,
+      "loss": 0.5385,
       "step": 1200
     },
     {
-      "epoch": 1.0471204188481675,
-      "grad_norm": 2.8621716499328613,
-      "learning_rate": 6.24e-06,
-      "loss": 0.5413,
+      "epoch": 1.0469011725293131,
+      "grad_norm": 2.3764073848724365,
+      "learning_rate": 6.235000000000001e-06,
+      "loss": 0.5647,
       "step": 1250
     },
     {
-      "epoch": 1.0890052356020943,
-      "grad_norm": 2.490875005722046,
-      "learning_rate": 6.4900000000000005e-06,
-      "loss": 0.5375,
+      "epoch": 1.0887772194304857,
+      "grad_norm": 4.364608287811279,
+      "learning_rate": 6.485000000000001e-06,
+      "loss": 0.5554,
       "step": 1300
     },
     {
-      "epoch": 1.130890052356021,
-      "grad_norm": 3.052311897277832,
-      "learning_rate": 6.740000000000001e-06,
-      "loss": 0.5415,
+      "epoch": 1.1306532663316582,
+      "grad_norm": 4.688488006591797,
+      "learning_rate": 6.735000000000001e-06,
+      "loss": 0.5676,
       "step": 1350
     },
     {
-      "epoch": 1.1727748691099475,
-      "grad_norm": 2.5695457458496094,
-      "learning_rate": 6.99e-06,
-      "loss": 0.5299,
+      "epoch": 1.1725293132328307,
+      "grad_norm": 3.0410969257354736,
+      "learning_rate": 6.985000000000001e-06,
+      "loss": 0.5374,
       "step": 1400
     },
     {
-      "epoch": 1.2146596858638743,
-      "grad_norm": 2.5093400478363037,
-      "learning_rate": 7.24e-06,
-      "loss": 0.5175,
+      "epoch": 1.2144053601340032,
+      "grad_norm": 6.28264856338501,
+      "learning_rate": 7.235000000000001e-06,
+      "loss": 0.5275,
       "step": 1450
     },
     {
-      "epoch": 1.256544502617801,
-      "grad_norm": 3.005453586578369,
-      "learning_rate": 7.49e-06,
-      "loss": 0.5203,
+      "epoch": 1.2562814070351758,
+      "grad_norm": 3.7520318031311035,
+      "learning_rate": 7.485000000000001e-06,
+      "loss": 0.5206,
       "step": 1500
     },
     {
-      "epoch": 1.256544502617801,
-      "eval_loss": 0.45439228415489197,
-      "eval_runtime": 266.4154,
-      "eval_samples_per_second": 31.864,
-      "eval_steps_per_second": 3.986,
+      "epoch": 1.2562814070351758,
+      "eval_loss": 0.4592318534851074,
+      "eval_runtime": 220.2748,
+      "eval_samples_per_second": 38.538,
+      "eval_steps_per_second": 4.821,
       "step": 1500
     },
     {
-      "epoch": 1.2984293193717278,
-      "grad_norm": 3.572266101837158,
-      "learning_rate": 7.74e-06,
-      "loss": 0.5192,
+      "epoch": 1.2981574539363483,
+      "grad_norm": 2.9252817630767822,
+      "learning_rate": 7.735e-06,
+      "loss": 0.5175,
       "step": 1550
     },
     {
-      "epoch": 1.3403141361256545,
-      "grad_norm": 2.545226573944092,
-      "learning_rate": 7.990000000000001e-06,
-      "loss": 0.5128,
+      "epoch": 1.3400335008375208,
+      "grad_norm": 2.6018381118774414,
+      "learning_rate": 7.985e-06,
+      "loss": 0.5191,
       "step": 1600
     },
     {
-      "epoch": 1.3821989528795813,
-      "grad_norm": 2.4371402263641357,
-      "learning_rate": 8.24e-06,
-      "loss": 0.5118,
+      "epoch": 1.3819095477386933,
+      "grad_norm": 2.215420722961426,
+      "learning_rate": 8.235e-06,
+      "loss": 0.5274,
       "step": 1650
     },
     {
-      "epoch": 1.4240837696335078,
-      "grad_norm": 2.7607970237731934,
-      "learning_rate": 8.49e-06,
-      "loss": 0.5059,
+      "epoch": 1.4237855946398659,
+      "grad_norm": 2.787269353866577,
+      "learning_rate": 8.485000000000001e-06,
+      "loss": 0.5201,
       "step": 1700
     },
     {
-      "epoch": 1.4659685863874345,
-      "grad_norm": 2.4732346534729004,
-      "learning_rate": 8.740000000000001e-06,
-      "loss": 0.5129,
+      "epoch": 1.4656616415410384,
+      "grad_norm": 3.644702434539795,
+      "learning_rate": 8.735000000000002e-06,
+      "loss": 0.5157,
       "step": 1750
     },
     {
-      "epoch": 1.5078534031413613,
-      "grad_norm": 2.1999361515045166,
-      "learning_rate": 8.99e-06,
-      "loss": 0.4977,
+      "epoch": 1.507537688442211,
+      "grad_norm": 2.4051408767700195,
+      "learning_rate": 8.985000000000001e-06,
+      "loss": 0.5091,
       "step": 1800
     },
     {
-      "epoch": 1.5497382198952878,
-      "grad_norm": 2.2437193393707275,
-      "learning_rate": 9.240000000000001e-06,
-      "loss": 0.4964,
+      "epoch": 1.5494137353433834,
+      "grad_norm": 3.729388952255249,
+      "learning_rate": 9.235e-06,
+      "loss": 0.5134,
       "step": 1850
     },
     {
-      "epoch": 1.5916230366492146,
-      "grad_norm": 1.9161577224731445,
-      "learning_rate": 9.49e-06,
-      "loss": 0.4914,
+      "epoch": 1.591289782244556,
+      "grad_norm": 3.360603094100952,
+      "learning_rate": 9.485000000000002e-06,
+      "loss": 0.5041,
       "step": 1900
     },
     {
-      "epoch": 1.6335078534031413,
-      "grad_norm": 3.3452305793762207,
-      "learning_rate": 9.74e-06,
-      "loss": 0.4971,
+      "epoch": 1.6331658291457285,
+      "grad_norm": 3.6210439205169678,
+      "learning_rate": 9.735e-06,
+      "loss": 0.4993,
       "step": 1950
     },
     {
-      "epoch": 1.675392670157068,
-      "grad_norm": 1.818142294883728,
-      "learning_rate": 9.990000000000001e-06,
-      "loss": 0.4873,
+      "epoch": 1.675041876046901,
+      "grad_norm": 2.6245505809783936,
+      "learning_rate": 9.985000000000002e-06,
+      "loss": 0.4979,
       "step": 2000
     },
     {
-      "epoch": 1.675392670157068,
-      "eval_loss": 0.43226656317710876,
-      "eval_runtime": 264.8191,
-      "eval_samples_per_second": 32.056,
-      "eval_steps_per_second": 4.01,
+      "epoch": 1.675041876046901,
+      "eval_loss": 0.4388435482978821,
+      "eval_runtime": 220.5956,
+      "eval_samples_per_second": 38.482,
+      "eval_steps_per_second": 4.814,
       "step": 2000
     },
     {
-      "epoch": 1.7172774869109948,
-      "grad_norm": 2.1289865970611572,
-      "learning_rate": 9.920000000000002e-06,
-      "loss": 0.4809,
+      "epoch": 1.7169179229480735,
+      "grad_norm": 3.5152196884155273,
+      "learning_rate": 9.921666666666667e-06,
+      "loss": 0.4917,
       "step": 2050
     },
     {
-      "epoch": 1.7591623036649215,
-      "grad_norm": 3.333080768585205,
-      "learning_rate": 9.836666666666668e-06,
-      "loss": 0.4853,
+      "epoch": 1.758793969849246,
+      "grad_norm": 2.4508583545684814,
+      "learning_rate": 9.838333333333334e-06,
+      "loss": 0.5161,
       "step": 2100
     },
     {
-      "epoch": 1.8010471204188483,
-      "grad_norm": 3.106902599334717,
-      "learning_rate": 9.753333333333335e-06,
-      "loss": 0.4841,
+      "epoch": 1.8006700167504186,
+      "grad_norm": 3.44022536277771,
+      "learning_rate": 9.755e-06,
+      "loss": 0.4933,
       "step": 2150
     },
     {
-      "epoch": 1.8429319371727748,
-      "grad_norm": 3.021559238433838,
-      "learning_rate": 9.67e-06,
-      "loss": 0.4795,
+      "epoch": 1.8425460636515911,
+      "grad_norm": 2.480679512023926,
+      "learning_rate": 9.671666666666668e-06,
+      "loss": 0.4983,
       "step": 2200
     },
     {
-      "epoch": 1.8848167539267016,
-      "grad_norm": 2.128567695617676,
-      "learning_rate": 9.586666666666667e-06,
-      "loss": 0.4784,
+      "epoch": 1.8844221105527639,
+      "grad_norm": 2.595881462097168,
+      "learning_rate": 9.588333333333334e-06,
+      "loss": 0.4797,
       "step": 2250
     },
     {
-      "epoch": 1.9267015706806283,
-      "grad_norm": 2.3653769493103027,
-      "learning_rate": 9.503333333333334e-06,
-      "loss": 0.4792,
+      "epoch": 1.9262981574539364,
+      "grad_norm": 2.0328593254089355,
+      "learning_rate": 9.505000000000001e-06,
+      "loss": 0.483,
       "step": 2300
     },
     {
-      "epoch": 1.9685863874345548,
-      "grad_norm": 2.409151792526245,
-      "learning_rate": 9.42e-06,
-      "loss": 0.4727,
+      "epoch": 1.968174204355109,
+      "grad_norm": 3.7075388431549072,
+      "learning_rate": 9.421666666666668e-06,
+      "loss": 0.4922,
       "step": 2350
     },
     {
-      "epoch": 2.0104712041884816,
-      "grad_norm": 2.9087882041931152,
-      "learning_rate": 9.336666666666666e-06,
-      "loss": 0.4645,
+      "epoch": 2.0100502512562812,
+      "grad_norm": 2.3989317417144775,
+      "learning_rate": 9.338333333333333e-06,
+      "loss": 0.4846,
       "step": 2400
     },
     {
-      "epoch": 2.0523560209424083,
-      "grad_norm": 2.5785608291625977,
-      "learning_rate": 9.253333333333333e-06,
-      "loss": 0.4696,
+      "epoch": 2.0519262981574538,
+      "grad_norm": 2.803882360458374,
+      "learning_rate": 9.255e-06,
+      "loss": 0.4832,
       "step": 2450
     },
     {
-      "epoch": 2.094240837696335,
-      "grad_norm": 2.625706672668457,
-      "learning_rate": 9.17e-06,
-      "loss": 0.4688,
+      "epoch": 2.0938023450586263,
+      "grad_norm": 3.1467437744140625,
+      "learning_rate": 9.171666666666667e-06,
+      "loss": 0.4852,
       "step": 2500
     },
     {
-      "epoch": 2.094240837696335,
-      "eval_loss": 0.41558554768562317,
-      "eval_runtime": 267.455,
-      "eval_samples_per_second": 31.74,
-      "eval_steps_per_second": 3.971,
+      "epoch": 2.0938023450586263,
+      "eval_loss": 0.4210895597934723,
+      "eval_runtime": 222.2617,
+      "eval_samples_per_second": 38.194,
+      "eval_steps_per_second": 4.778,
       "step": 2500
     }
   ],
@@ -416,8 +416,8 @@
       "attributes": {}
     }
   },
-  "total_flos": 2.236933663709184e+16,
-  "train_batch_size": 16,
+  "total_flos": 2.424075092792179e+16,
+  "train_batch_size": 32,
   "trial_name": null,
   "trial_params": null
 }

checkpoint-2500/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6cbb0736de458ba42fdcabf5316a533e48ea0601626bceabd95443adeda6296a
+oid sha256:6248ade84a48a52ce5a07ff6e203bd335b96af70b5ce5ebd92ad797d93be969c
 size 5304