Training in progress, step 4500, checkpoint

checkpoint-4500/model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:28bd8c4540407750715005702e248f5a1bbcb47e12075aac89498f4e898f7897
+oid sha256:d0d776e6ccf57ee9d12d76d7d2c2701506fe7a487eb54c5cb504d483344b1a8c
 size 577789320

checkpoint-4500/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b1133bc2dd826fdbc1096ba73f7076059a8bab3ae057d17dc06a42e69b7ee279
+oid sha256:36e1a1a524b6e6626595eeca2a4f7e3b2c50faff2c6307eca727b8e4ca553d2f
 size 1155772233

checkpoint-4500/rng_state.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:55b41ee41df1b95daa7c247b046435c9be54cceaa08049c6c5013204de4ef1d1
+oid sha256:4477ab62e6f0d340919ab984f8b02c10833202f2ea466749260844bbb0d73968
 size 14244

checkpoint-4500/trainer_state.json

@@ -1,7 +1,7 @@
 {
-  "best_metric": 0.3396616578102112,
+  "best_metric": 0.3422691822052002,
   "best_model_checkpoint": "mikhail_panzo/zlm_b64_le4_s12000/checkpoint-4500",
-  "epoch": 3.770261780104712,
+  "epoch": 3.7688442211055277,
   "eval_steps": 500,
   "global_step": 4500,
   "is_hyper_param_search": false,
@@ -9,705 +9,705 @@
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.041884816753926704,
-      "grad_norm": 2.697540283203125,
-      "learning_rate": 2.4000000000000003e-06,
-      "loss": 1.0526,
+      "epoch": 0.04187604690117253,
+      "grad_norm": 3.0184223651885986,
+      "learning_rate": 2.35e-06,
+      "loss": 1.1206,
       "step": 50
     },
     {
-      "epoch": 0.08376963350785341,
-      "grad_norm": 4.0351996421813965,
-      "learning_rate": 4.9000000000000005e-06,
-      "loss": 0.8484,
+      "epoch": 0.08375209380234507,
+      "grad_norm": 3.4442005157470703,
+      "learning_rate": 4.85e-06,
+      "loss": 0.8185,
       "step": 100
     },
     {
-      "epoch": 0.1256544502617801,
-      "grad_norm": 4.858840465545654,
-      "learning_rate": 7.4e-06,
-      "loss": 0.7505,
+      "epoch": 0.12562814070351758,
+      "grad_norm": 2.6812493801116943,
+      "learning_rate": 7.35e-06,
+      "loss": 0.7576,
       "step": 150
     },
     {
-      "epoch": 0.16753926701570682,
-      "grad_norm": 2.396157741546631,
-      "learning_rate": 9.900000000000002e-06,
-      "loss": 0.6889,
+      "epoch": 0.16750418760469013,
+      "grad_norm": 4.518482685089111,
+      "learning_rate": 9.85e-06,
+      "loss": 0.6938,
       "step": 200
     },
     {
-      "epoch": 0.2094240837696335,
-      "grad_norm": 2.660891056060791,
-      "learning_rate": 1.24e-05,
-      "loss": 0.6363,
+      "epoch": 0.20938023450586266,
+      "grad_norm": 4.683146953582764,
+      "learning_rate": 1.235e-05,
+      "loss": 0.6386,
       "step": 250
     },
     {
-      "epoch": 0.2513089005235602,
-      "grad_norm": 2.848365068435669,
-      "learning_rate": 1.49e-05,
-      "loss": 0.6107,
+      "epoch": 0.25125628140703515,
+      "grad_norm": 2.7392578125,
+      "learning_rate": 1.485e-05,
+      "loss": 0.6226,
       "step": 300
     },
     {
-      "epoch": 0.2931937172774869,
-      "grad_norm": 2.0031681060791016,
-      "learning_rate": 1.74e-05,
-      "loss": 0.5673,
+      "epoch": 0.2931323283082077,
+      "grad_norm": 2.017056465148926,
+      "learning_rate": 1.7349999999999998e-05,
+      "loss": 0.6128,
       "step": 350
     },
     {
-      "epoch": 0.33507853403141363,
-      "grad_norm": 4.771584510803223,
-      "learning_rate": 1.9900000000000003e-05,
-      "loss": 0.5566,
+      "epoch": 0.33500837520938026,
+      "grad_norm": 3.6669230461120605,
+      "learning_rate": 1.985e-05,
+      "loss": 0.5675,
       "step": 400
     },
     {
-      "epoch": 0.3769633507853403,
-      "grad_norm": 9.876974105834961,
-      "learning_rate": 2.2400000000000002e-05,
-      "loss": 0.5487,
+      "epoch": 0.3768844221105528,
+      "grad_norm": 4.16896915435791,
+      "learning_rate": 2.235e-05,
+      "loss": 0.5655,
       "step": 450
     },
     {
-      "epoch": 0.418848167539267,
-      "grad_norm": 2.113546848297119,
-      "learning_rate": 2.4900000000000002e-05,
-      "loss": 0.5268,
+      "epoch": 0.4187604690117253,
+      "grad_norm": 3.691459894180298,
+      "learning_rate": 2.485e-05,
+      "loss": 0.5487,
       "step": 500
     },
     {
-      "epoch": 0.418848167539267,
-      "eval_loss": 0.46557939052581787,
-      "eval_runtime": 297.9193,
-      "eval_samples_per_second": 28.494,
-      "eval_steps_per_second": 3.565,
+      "epoch": 0.4187604690117253,
+      "eval_loss": 0.47458764910697937,
+      "eval_runtime": 214.7196,
+      "eval_samples_per_second": 39.535,
+      "eval_steps_per_second": 4.946,
       "step": 500
     },
     {
-      "epoch": 0.4607329842931937,
-      "grad_norm": 2.7003655433654785,
-      "learning_rate": 2.7400000000000002e-05,
-      "loss": 0.5212,
+      "epoch": 0.46063651591289784,
+      "grad_norm": 3.3792996406555176,
+      "learning_rate": 2.7350000000000004e-05,
+      "loss": 0.5232,
       "step": 550
     },
     {
-      "epoch": 0.5026178010471204,
-      "grad_norm": 5.649191379547119,
-      "learning_rate": 2.9900000000000002e-05,
-      "loss": 0.5138,
+      "epoch": 0.5025125628140703,
+      "grad_norm": 3.6263363361358643,
+      "learning_rate": 2.985e-05,
+      "loss": 0.5132,
       "step": 600
     },
     {
-      "epoch": 0.5445026178010471,
-      "grad_norm": 2.272181749343872,
-      "learning_rate": 3.24e-05,
-      "loss": 0.5121,
+      "epoch": 0.5443886097152428,
+      "grad_norm": 2.652538776397705,
+      "learning_rate": 3.235e-05,
+      "loss": 0.5067,
       "step": 650
     },
     {
-      "epoch": 0.5863874345549738,
-      "grad_norm": 2.957671642303467,
-      "learning_rate": 3.49e-05,
-      "loss": 0.496,
+      "epoch": 0.5862646566164154,
+      "grad_norm": 3.5355172157287598,
+      "learning_rate": 3.485e-05,
+      "loss": 0.4981,
       "step": 700
     },
     {
-      "epoch": 0.6282722513089005,
-      "grad_norm": 3.577693462371826,
-      "learning_rate": 3.74e-05,
-      "loss": 0.4886,
+      "epoch": 0.628140703517588,
+      "grad_norm": 2.5035619735717773,
+      "learning_rate": 3.735e-05,
+      "loss": 0.5094,
       "step": 750
     },
     {
-      "epoch": 0.6701570680628273,
-      "grad_norm": 2.466113328933716,
-      "learning_rate": 3.99e-05,
-      "loss": 0.4879,
+      "epoch": 0.6700167504187605,
+      "grad_norm": 2.9585518836975098,
+      "learning_rate": 3.9850000000000006e-05,
+      "loss": 0.5053,
       "step": 800
     },
     {
-      "epoch": 0.7120418848167539,
-      "grad_norm": 2.941208600997925,
-      "learning_rate": 4.24e-05,
-      "loss": 0.4865,
+      "epoch": 0.711892797319933,
+      "grad_norm": 9.997116088867188,
+      "learning_rate": 4.235e-05,
+      "loss": 0.502,
       "step": 850
     },
     {
-      "epoch": 0.7539267015706806,
-      "grad_norm": 2.3843142986297607,
-      "learning_rate": 4.49e-05,
-      "loss": 0.4689,
+      "epoch": 0.7537688442211056,
+      "grad_norm": 1.8630377054214478,
+      "learning_rate": 4.4850000000000006e-05,
+      "loss": 0.483,
       "step": 900
     },
     {
-      "epoch": 0.7958115183246073,
-      "grad_norm": 2.232208728790283,
-      "learning_rate": 4.74e-05,
-      "loss": 0.4768,
+      "epoch": 0.7956448911222781,
+      "grad_norm": 1.5534099340438843,
+      "learning_rate": 4.735e-05,
+      "loss": 0.4743,
       "step": 950
     },
     {
-      "epoch": 0.837696335078534,
-      "grad_norm": 2.1702969074249268,
-      "learning_rate": 4.99e-05,
-      "loss": 0.4583,
+      "epoch": 0.8375209380234506,
+      "grad_norm": 2.8155503273010254,
+      "learning_rate": 4.9850000000000006e-05,
+      "loss": 0.483,
       "step": 1000
     },
     {
-      "epoch": 0.837696335078534,
-      "eval_loss": 0.4044812321662903,
-      "eval_runtime": 296.3081,
-      "eval_samples_per_second": 28.649,
-      "eval_steps_per_second": 3.584,
+      "epoch": 0.8375209380234506,
+      "eval_loss": 0.4226582944393158,
+      "eval_runtime": 213.2339,
+      "eval_samples_per_second": 39.811,
+      "eval_steps_per_second": 4.98,
       "step": 1000
     },
     {
-      "epoch": 0.8795811518324608,
-      "grad_norm": 2.7613942623138428,
-      "learning_rate": 5.2400000000000007e-05,
-      "loss": 0.4664,
+      "epoch": 0.8793969849246231,
+      "grad_norm": 4.2074055671691895,
+      "learning_rate": 5.235e-05,
+      "loss": 0.4625,
       "step": 1050
     },
     {
-      "epoch": 0.9214659685863874,
-      "grad_norm": 2.4752321243286133,
-      "learning_rate": 5.4900000000000006e-05,
-      "loss": 0.442,
+      "epoch": 0.9212730318257957,
+      "grad_norm": 2.215416431427002,
+      "learning_rate": 5.485e-05,
+      "loss": 0.461,
       "step": 1100
     },
     {
-      "epoch": 0.9633507853403142,
-      "grad_norm": 2.27407169342041,
-      "learning_rate": 5.74e-05,
-      "loss": 0.4498,
+      "epoch": 0.9631490787269682,
+      "grad_norm": 3.3321831226348877,
+      "learning_rate": 5.7350000000000005e-05,
+      "loss": 0.4807,
       "step": 1150
     },
     {
-      "epoch": 1.0052356020942408,
-      "grad_norm": 1.8998862504959106,
-      "learning_rate": 5.99e-05,
-      "loss": 0.4433,
+      "epoch": 1.0050251256281406,
+      "grad_norm": 3.5796420574188232,
+      "learning_rate": 5.9850000000000005e-05,
+      "loss": 0.4456,
       "step": 1200
     },
     {
-      "epoch": 1.0471204188481675,
-      "grad_norm": 2.8479738235473633,
-      "learning_rate": 6.24e-05,
-      "loss": 0.4417,
+      "epoch": 1.0469011725293131,
+      "grad_norm": 2.3565518856048584,
+      "learning_rate": 6.235000000000001e-05,
+      "loss": 0.4664,
       "step": 1250
     },
     {
-      "epoch": 1.0890052356020943,
-      "grad_norm": 1.9095953702926636,
-      "learning_rate": 6.49e-05,
-      "loss": 0.444,
+      "epoch": 1.0887772194304857,
+      "grad_norm": 3.147794485092163,
+      "learning_rate": 6.485e-05,
+      "loss": 0.4681,
       "step": 1300
     },
     {
-      "epoch": 1.130890052356021,
-      "grad_norm": 1.368077278137207,
-      "learning_rate": 6.740000000000001e-05,
-      "loss": 0.4519,
+      "epoch": 1.1306532663316582,
+      "grad_norm": 3.028409957885742,
+      "learning_rate": 6.735e-05,
+      "loss": 0.4663,
       "step": 1350
     },
     {
-      "epoch": 1.1727748691099475,
-      "grad_norm": 2.0456721782684326,
-      "learning_rate": 6.99e-05,
-      "loss": 0.4411,
+      "epoch": 1.1725293132328307,
+      "grad_norm": 4.820915222167969,
+      "learning_rate": 6.985e-05,
+      "loss": 0.4493,
       "step": 1400
     },
     {
-      "epoch": 1.2146596858638743,
-      "grad_norm": 1.6217619180679321,
-      "learning_rate": 7.24e-05,
-      "loss": 0.4324,
+      "epoch": 1.2144053601340032,
+      "grad_norm": 1.93254816532135,
+      "learning_rate": 7.235000000000001e-05,
+      "loss": 0.4413,
       "step": 1450
     },
     {
-      "epoch": 1.256544502617801,
-      "grad_norm": 2.961646795272827,
-      "learning_rate": 7.49e-05,
-      "loss": 0.4321,
+      "epoch": 1.2562814070351758,
+      "grad_norm": 2.237015724182129,
+      "learning_rate": 7.485e-05,
+      "loss": 0.432,
       "step": 1500
     },
     {
-      "epoch": 1.256544502617801,
-      "eval_loss": 0.3930450975894928,
-      "eval_runtime": 301.1428,
-      "eval_samples_per_second": 28.189,
-      "eval_steps_per_second": 3.527,
+      "epoch": 1.2562814070351758,
+      "eval_loss": 0.39829039573669434,
+      "eval_runtime": 217.3782,
+      "eval_samples_per_second": 39.052,
+      "eval_steps_per_second": 4.885,
       "step": 1500
     },
     {
-      "epoch": 1.2984293193717278,
-      "grad_norm": 2.543201446533203,
-      "learning_rate": 7.740000000000001e-05,
-      "loss": 0.4364,
+      "epoch": 1.2981574539363483,
+      "grad_norm": 1.709120273590088,
+      "learning_rate": 7.735e-05,
+      "loss": 0.4361,
       "step": 1550
     },
     {
-      "epoch": 1.3403141361256545,
-      "grad_norm": 1.5807560682296753,
-      "learning_rate": 7.99e-05,
-      "loss": 0.433,
+      "epoch": 1.3400335008375208,
+      "grad_norm": 3.733328104019165,
+      "learning_rate": 7.985e-05,
+      "loss": 0.4357,
       "step": 1600
     },
     {
-      "epoch": 1.3821989528795813,
-      "grad_norm": 1.587385892868042,
-      "learning_rate": 8.24e-05,
-      "loss": 0.4285,
+      "epoch": 1.3819095477386933,
+      "grad_norm": 4.131324768066406,
+      "learning_rate": 8.235000000000001e-05,
+      "loss": 0.4445,
       "step": 1650
     },
     {
-      "epoch": 1.4240837696335078,
-      "grad_norm": 2.1177687644958496,
-      "learning_rate": 8.49e-05,
-      "loss": 0.4209,
+      "epoch": 1.4237855946398659,
+      "grad_norm": 2.2893927097320557,
+      "learning_rate": 8.485e-05,
+      "loss": 0.4379,
       "step": 1700
     },
     {
-      "epoch": 1.4659685863874345,
-      "grad_norm": 1.6078702211380005,
-      "learning_rate": 8.740000000000001e-05,
-      "loss": 0.426,
+      "epoch": 1.4656616415410384,
+      "grad_norm": 2.1464734077453613,
+      "learning_rate": 8.735000000000001e-05,
+      "loss": 0.4393,
       "step": 1750
     },
     {
-      "epoch": 1.5078534031413613,
-      "grad_norm": 2.10492205619812,
-      "learning_rate": 8.99e-05,
-      "loss": 0.4191,
+      "epoch": 1.507537688442211,
+      "grad_norm": 2.01698637008667,
+      "learning_rate": 8.985e-05,
+      "loss": 0.4341,
       "step": 1800
     },
     {
-      "epoch": 1.5497382198952878,
-      "grad_norm": 2.169309616088867,
-      "learning_rate": 9.240000000000001e-05,
-      "loss": 0.4178,
+      "epoch": 1.5494137353433834,
+      "grad_norm": 2.259115219116211,
+      "learning_rate": 9.235000000000001e-05,
+      "loss": 0.4324,
       "step": 1850
     },
     {
-      "epoch": 1.5916230366492146,
-      "grad_norm": 1.559006929397583,
-      "learning_rate": 9.49e-05,
-      "loss": 0.4133,
+      "epoch": 1.591289782244556,
+      "grad_norm": 2.0072858333587646,
+      "learning_rate": 9.485e-05,
+      "loss": 0.4327,
       "step": 1900
     },
     {
-      "epoch": 1.6335078534031413,
-      "grad_norm": 2.0436346530914307,
-      "learning_rate": 9.74e-05,
-      "loss": 0.4234,
+      "epoch": 1.6331658291457285,
+      "grad_norm": 2.060303211212158,
+      "learning_rate": 9.735000000000001e-05,
+      "loss": 0.4279,
       "step": 1950
     },
     {
-      "epoch": 1.675392670157068,
-      "grad_norm": 2.089486837387085,
-      "learning_rate": 9.99e-05,
-      "loss": 0.4092,
+      "epoch": 1.675041876046901,
+      "grad_norm": 1.2846741676330566,
+      "learning_rate": 9.985000000000001e-05,
+      "loss": 0.429,
       "step": 2000
     },
     {
-      "epoch": 1.675392670157068,
-      "eval_loss": 0.37686464190483093,
-      "eval_runtime": 310.3113,
-      "eval_samples_per_second": 27.356,
-      "eval_steps_per_second": 3.422,
+      "epoch": 1.675041876046901,
+      "eval_loss": 0.3953319191932678,
+      "eval_runtime": 217.8108,
+      "eval_samples_per_second": 38.974,
+      "eval_steps_per_second": 4.876,
       "step": 2000
     },
     {
-      "epoch": 1.7179057591623037,
-      "grad_norm": 2.684563398361206,
-      "learning_rate": 9.952e-05,
-      "loss": 0.4099,
+      "epoch": 1.7169179229480735,
+      "grad_norm": 2.253840923309326,
+      "learning_rate": 9.953e-05,
+      "loss": 0.4264,
       "step": 2050
     },
     {
-      "epoch": 1.7597905759162304,
-      "grad_norm": 1.7815254926681519,
-      "learning_rate": 9.902e-05,
-      "loss": 0.4093,
+      "epoch": 1.758793969849246,
+      "grad_norm": 1.3114299774169922,
+      "learning_rate": 9.903e-05,
+      "loss": 0.4355,
       "step": 2100
     },
     {
-      "epoch": 1.8016753926701572,
-      "grad_norm": 1.9900892972946167,
-      "learning_rate": 9.852e-05,
-      "loss": 0.4116,
+      "epoch": 1.8006700167504186,
+      "grad_norm": 1.8178610801696777,
+      "learning_rate": 9.853e-05,
+      "loss": 0.4256,
       "step": 2150
     },
     {
-      "epoch": 1.8435602094240837,
-      "grad_norm": 1.8637460470199585,
-      "learning_rate": 9.802e-05,
-      "loss": 0.4089,
+      "epoch": 1.8425460636515911,
+      "grad_norm": 2.1193735599517822,
+      "learning_rate": 9.803e-05,
+      "loss": 0.4305,
       "step": 2200
     },
     {
-      "epoch": 1.8854450261780105,
-      "grad_norm": 1.58171546459198,
-      "learning_rate": 9.752e-05,
-      "loss": 0.4053,
+      "epoch": 1.8844221105527639,
+      "grad_norm": 1.6871263980865479,
+      "learning_rate": 9.753e-05,
+      "loss": 0.4106,
       "step": 2250
     },
     {
-      "epoch": 1.9273298429319372,
-      "grad_norm": 1.4995630979537964,
-      "learning_rate": 9.702e-05,
-      "loss": 0.4049,
+      "epoch": 1.9262981574539364,
+      "grad_norm": 1.2699921131134033,
+      "learning_rate": 9.703000000000001e-05,
+      "loss": 0.4156,
       "step": 2300
     },
     {
-      "epoch": 1.9692146596858637,
-      "grad_norm": 2.447049379348755,
-      "learning_rate": 9.652e-05,
-      "loss": 0.4029,
+      "epoch": 1.968174204355109,
+      "grad_norm": 1.7966769933700562,
+      "learning_rate": 9.653000000000001e-05,
+      "loss": 0.4257,
       "step": 2350
     },
     {
-      "epoch": 2.0110994764397905,
-      "grad_norm": 1.8776074647903442,
-      "learning_rate": 9.602e-05,
-      "loss": 0.3988,
+      "epoch": 2.0100502512562812,
+      "grad_norm": 1.7838683128356934,
+      "learning_rate": 9.603000000000001e-05,
+      "loss": 0.4156,
       "step": 2400
     },
     {
-      "epoch": 2.0529842931937172,
-      "grad_norm": 3.1785056591033936,
-      "learning_rate": 9.552000000000001e-05,
-      "loss": 0.4023,
+      "epoch": 2.0519262981574538,
+      "grad_norm": 1.4161527156829834,
+      "learning_rate": 9.553e-05,
+      "loss": 0.4096,
       "step": 2450
     },
     {
-      "epoch": 2.094869109947644,
-      "grad_norm": 1.4416707754135132,
-      "learning_rate": 9.502000000000001e-05,
-      "loss": 0.4008,
+      "epoch": 2.0938023450586263,
+      "grad_norm": 1.5864896774291992,
+      "learning_rate": 9.503e-05,
+      "loss": 0.4168,
       "step": 2500
     },
     {
-      "epoch": 2.094869109947644,
-      "eval_loss": 0.35975897312164307,
-      "eval_runtime": 275.2855,
-      "eval_samples_per_second": 30.837,
-      "eval_steps_per_second": 3.858,
+      "epoch": 2.0938023450586263,
+      "eval_loss": 0.3701198697090149,
+      "eval_runtime": 220.4399,
+      "eval_samples_per_second": 38.509,
+      "eval_steps_per_second": 4.818,
       "step": 2500
     },
     {
-      "epoch": 2.1367539267015707,
-      "grad_norm": 1.3332245349884033,
-      "learning_rate": 9.452000000000001e-05,
-      "loss": 0.3985,
+      "epoch": 2.135678391959799,
+      "grad_norm": 1.2460856437683105,
+      "learning_rate": 9.453e-05,
+      "loss": 0.415,
       "step": 2550
     },
     {
-      "epoch": 2.1786387434554975,
-      "grad_norm": 1.291338324546814,
-      "learning_rate": 9.402000000000001e-05,
-      "loss": 0.4021,
+      "epoch": 2.1775544388609713,
+      "grad_norm": 1.994315266609192,
+      "learning_rate": 9.403e-05,
+      "loss": 0.4075,
       "step": 2600
     },
     {
-      "epoch": 2.220523560209424,
-      "grad_norm": 1.6172423362731934,
-      "learning_rate": 9.352000000000001e-05,
-      "loss": 0.4006,
+      "epoch": 2.219430485762144,
+      "grad_norm": 2.148324966430664,
+      "learning_rate": 9.353000000000001e-05,
+      "loss": 0.4077,
       "step": 2650
     },
     {
-      "epoch": 2.262408376963351,
-      "grad_norm": 1.8729867935180664,
-      "learning_rate": 9.302e-05,
-      "loss": 0.3966,
+      "epoch": 2.2613065326633164,
+      "grad_norm": 1.7242035865783691,
+      "learning_rate": 9.303000000000001e-05,
+      "loss": 0.4044,
       "step": 2700
     },
     {
-      "epoch": 2.3042931937172773,
-      "grad_norm": 1.623901605606079,
-      "learning_rate": 9.252e-05,
-      "loss": 0.3951,
+      "epoch": 2.3031825795644894,
+      "grad_norm": 1.912110447883606,
+      "learning_rate": 9.253000000000001e-05,
+      "loss": 0.4075,
       "step": 2750
     },
     {
-      "epoch": 2.346178010471204,
-      "grad_norm": 1.032320261001587,
-      "learning_rate": 9.202e-05,
-      "loss": 0.3904,
+      "epoch": 2.3450586264656614,
+      "grad_norm": 1.3661189079284668,
+      "learning_rate": 9.203000000000001e-05,
+      "loss": 0.4041,
       "step": 2800
     },
     {
-      "epoch": 2.3880628272251307,
-      "grad_norm": 2.0027778148651123,
-      "learning_rate": 9.152e-05,
-      "loss": 0.3933,
+      "epoch": 2.3869346733668344,
+      "grad_norm": 1.7083531618118286,
+      "learning_rate": 9.153e-05,
+      "loss": 0.396,
       "step": 2850
     },
     {
-      "epoch": 2.4299476439790575,
-      "grad_norm": 1.5681087970733643,
-      "learning_rate": 9.102e-05,
-      "loss": 0.4024,
+      "epoch": 2.4288107202680065,
+      "grad_norm": 1.7829670906066895,
+      "learning_rate": 9.103e-05,
+      "loss": 0.4042,
       "step": 2900
     },
     {
-      "epoch": 2.4718324607329842,
-      "grad_norm": 1.2643622159957886,
-      "learning_rate": 9.052000000000001e-05,
-      "loss": 0.3939,
+      "epoch": 2.4706867671691795,
+      "grad_norm": 2.188753366470337,
+      "learning_rate": 9.053e-05,
+      "loss": 0.3963,
       "step": 2950
     },
     {
-      "epoch": 2.513717277486911,
-      "grad_norm": 1.4029465913772583,
-      "learning_rate": 9.002000000000001e-05,
-      "loss": 0.3941,
+      "epoch": 2.5125628140703515,
+      "grad_norm": 1.2608036994934082,
+      "learning_rate": 9.003e-05,
+      "loss": 0.4021,
       "step": 3000
     },
     {
-      "epoch": 2.513717277486911,
-      "eval_loss": 0.3533318042755127,
-      "eval_runtime": 271.9646,
-      "eval_samples_per_second": 31.214,
-      "eval_steps_per_second": 3.905,
+      "epoch": 2.5125628140703515,
+      "eval_loss": 0.3613213300704956,
+      "eval_runtime": 214.7226,
+      "eval_samples_per_second": 39.535,
+      "eval_steps_per_second": 4.946,
       "step": 3000
     },
     {
-      "epoch": 2.5556020942408377,
-      "grad_norm": 1.6669938564300537,
-      "learning_rate": 8.952000000000001e-05,
-      "loss": 0.3908,
+      "epoch": 2.5544388609715245,
+      "grad_norm": 1.501607060432434,
+      "learning_rate": 8.953e-05,
+      "loss": 0.3897,
       "step": 3050
     },
     {
-      "epoch": 2.5974869109947645,
-      "grad_norm": 1.6020203828811646,
-      "learning_rate": 8.902e-05,
-      "loss": 0.3905,
+      "epoch": 2.5963149078726966,
+      "grad_norm": 1.321588397026062,
+      "learning_rate": 8.903e-05,
+      "loss": 0.3948,
       "step": 3100
     },
     {
-      "epoch": 2.639371727748691,
-      "grad_norm": 2.2321176528930664,
-      "learning_rate": 8.852e-05,
-      "loss": 0.389,
+      "epoch": 2.6381909547738696,
+      "grad_norm": 1.5581566095352173,
+      "learning_rate": 8.853000000000001e-05,
+      "loss": 0.4005,
       "step": 3150
     },
     {
-      "epoch": 2.681256544502618,
-      "grad_norm": 1.0864609479904175,
-      "learning_rate": 8.802e-05,
-      "loss": 0.3906,
+      "epoch": 2.6800670016750416,
+      "grad_norm": 1.3824856281280518,
+      "learning_rate": 8.803e-05,
+      "loss": 0.3981,
       "step": 3200
     },
     {
-      "epoch": 2.7231413612565447,
-      "grad_norm": 2.0679004192352295,
-      "learning_rate": 8.752e-05,
-      "loss": 0.3869,
+      "epoch": 2.7219430485762146,
+      "grad_norm": 1.7730051279067993,
+      "learning_rate": 8.753e-05,
+      "loss": 0.3982,
       "step": 3250
     },
     {
-      "epoch": 2.765026178010471,
-      "grad_norm": 2.6636288166046143,
-      "learning_rate": 8.702e-05,
-      "loss": 0.3841,
+      "epoch": 2.7638190954773867,
+      "grad_norm": 1.516622543334961,
+      "learning_rate": 8.703e-05,
+      "loss": 0.3906,
       "step": 3300
     },
     {
-      "epoch": 2.8069109947643978,
-      "grad_norm": 1.435549259185791,
-      "learning_rate": 8.652e-05,
-      "loss": 0.3841,
+      "epoch": 2.8056951423785597,
+      "grad_norm": 1.2814849615097046,
+      "learning_rate": 8.653e-05,
+      "loss": 0.3859,
       "step": 3350
     },
     {
-      "epoch": 2.8487958115183245,
-      "grad_norm": 2.208448648452759,
-      "learning_rate": 8.602e-05,
-      "loss": 0.3828,
+      "epoch": 2.8475711892797317,
+      "grad_norm": 1.239619255065918,
+      "learning_rate": 8.603e-05,
+      "loss": 0.3942,
       "step": 3400
     },
     {
-      "epoch": 2.8906806282722513,
-      "grad_norm": 1.731292963027954,
-      "learning_rate": 8.552e-05,
-      "loss": 0.3842,
+      "epoch": 2.8894472361809047,
+      "grad_norm": 2.24064040184021,
+      "learning_rate": 8.553e-05,
+      "loss": 0.3938,
       "step": 3450
     },
     {
-      "epoch": 2.932565445026178,
-      "grad_norm": 1.7065523862838745,
-      "learning_rate": 8.502e-05,
-      "loss": 0.3834,
+      "epoch": 2.931323283082077,
+      "grad_norm": 1.1459012031555176,
+      "learning_rate": 8.503e-05,
+      "loss": 0.3925,
       "step": 3500
     },
     {
-      "epoch": 2.932565445026178,
-      "eval_loss": 0.3439213037490845,
-      "eval_runtime": 274.3567,
-      "eval_samples_per_second": 30.941,
-      "eval_steps_per_second": 3.871,
+      "epoch": 2.931323283082077,
+      "eval_loss": 0.35087427496910095,
+      "eval_runtime": 217.1018,
+      "eval_samples_per_second": 39.101,
+      "eval_steps_per_second": 4.892,
       "step": 3500
     },
     {
-      "epoch": 2.9744502617801047,
-      "grad_norm": 1.5389653444290161,
-      "learning_rate": 8.452e-05,
-      "loss": 0.3915,
+      "epoch": 2.9731993299832498,
+      "grad_norm": 1.8844960927963257,
+      "learning_rate": 8.453e-05,
+      "loss": 0.3803,
       "step": 3550
     },
     {
-      "epoch": 3.0163350785340315,
-      "grad_norm": 1.3524985313415527,
-      "learning_rate": 8.402e-05,
-      "loss": 0.3843,
+      "epoch": 3.0150753768844223,
+      "grad_norm": 1.5738004446029663,
+      "learning_rate": 8.403000000000001e-05,
+      "loss": 0.3895,
       "step": 3600
     },
     {
-      "epoch": 3.0582198952879582,
-      "grad_norm": 1.5389949083328247,
-      "learning_rate": 8.352000000000001e-05,
-      "loss": 0.3822,
+      "epoch": 3.056951423785595,
+      "grad_norm": 1.19874906539917,
+      "learning_rate": 8.353000000000001e-05,
+      "loss": 0.3902,
       "step": 3650
     },
     {
-      "epoch": 3.100104712041885,
-      "grad_norm": 1.6620662212371826,
-      "learning_rate": 8.302000000000001e-05,
-      "loss": 0.38,
+      "epoch": 3.0988274706867673,
+      "grad_norm": 2.4810609817504883,
+      "learning_rate": 8.303000000000001e-05,
+      "loss": 0.3863,
       "step": 3700
     },
     {
-      "epoch": 3.1419895287958117,
-      "grad_norm": 1.187556505203247,
-      "learning_rate": 8.252e-05,
-      "loss": 0.3768,
+      "epoch": 3.14070351758794,
+      "grad_norm": 1.4734585285186768,
+      "learning_rate": 8.253000000000001e-05,
+      "loss": 0.391,
       "step": 3750
     },
     {
-      "epoch": 3.183874345549738,
-      "grad_norm": 2.6059765815734863,
-      "learning_rate": 8.202e-05,
-      "loss": 0.3799,
+      "epoch": 3.1825795644891124,
+      "grad_norm": 1.6449118852615356,
+      "learning_rate": 8.203000000000001e-05,
+      "loss": 0.3888,
       "step": 3800
     },
     {
-      "epoch": 3.2257591623036648,
-      "grad_norm": 1.3912620544433594,
-      "learning_rate": 8.152e-05,
-      "loss": 0.3776,
+      "epoch": 3.224455611390285,
+      "grad_norm": 2.5607736110687256,
+      "learning_rate": 8.153e-05,
+      "loss": 0.3893,
       "step": 3850
     },
     {
-      "epoch": 3.2676439790575915,
-      "grad_norm": 1.4691671133041382,
-      "learning_rate": 8.102000000000001e-05,
-      "loss": 0.3755,
+      "epoch": 3.2663316582914574,
+      "grad_norm": 1.0670469999313354,
+      "learning_rate": 8.103e-05,
+      "loss": 0.3853,
       "step": 3900
     },
     {
-      "epoch": 3.3095287958115183,
-      "grad_norm": 1.452143669128418,
-      "learning_rate": 8.052000000000001e-05,
-      "loss": 0.3821,
+      "epoch": 3.30820770519263,
+      "grad_norm": 2.191486120223999,
+      "learning_rate": 8.053e-05,
+      "loss": 0.3832,
       "step": 3950
     },
     {
-      "epoch": 3.351413612565445,
-      "grad_norm": 0.9386736750602722,
-      "learning_rate": 8.002000000000001e-05,
-      "loss": 0.3699,
+      "epoch": 3.3500837520938025,
+      "grad_norm": 1.0259532928466797,
+      "learning_rate": 8.003e-05,
+      "loss": 0.3839,
       "step": 4000
     },
     {
-      "epoch": 3.351413612565445,
-      "eval_loss": 0.342171847820282,
-      "eval_runtime": 268.7667,
-      "eval_samples_per_second": 31.585,
-      "eval_steps_per_second": 3.951,
+      "epoch": 3.3500837520938025,
+      "eval_loss": 0.35058465600013733,
+      "eval_runtime": 214.7925,
+      "eval_samples_per_second": 39.522,
+      "eval_steps_per_second": 4.944,
       "step": 4000
     },
     {
-      "epoch": 3.3932984293193718,
-      "grad_norm": 1.6316275596618652,
-      "learning_rate": 7.952000000000001e-05,
-      "loss": 0.3741,
+      "epoch": 3.391959798994975,
+      "grad_norm": 2.1106534004211426,
+      "learning_rate": 7.953e-05,
+      "loss": 0.4088,
       "step": 4050
     },
     {
-      "epoch": 3.4351832460732985,
-      "grad_norm": 1.2369744777679443,
-      "learning_rate": 7.902e-05,
-      "loss": 0.3728,
+      "epoch": 3.4338358458961475,
+      "grad_norm": 1.7350833415985107,
+      "learning_rate": 7.903000000000001e-05,
+      "loss": 0.3923,
       "step": 4100
     },
     {
-      "epoch": 3.4770680628272252,
-      "grad_norm": 1.48318350315094,
+      "epoch": 3.47571189279732,
+      "grad_norm": 1.1461760997772217,
       "learning_rate": 7.853000000000001e-05,
-      "loss": 0.3772,
+      "loss": 0.3924,
       "step": 4150
     },
     {
-      "epoch": 3.518952879581152,
-      "grad_norm": 1.4171572923660278,
+      "epoch": 3.5175879396984926,
+      "grad_norm": 1.3420217037200928,
       "learning_rate": 7.803e-05,
-      "loss": 0.3717,
+      "loss": 0.3833,
       "step": 4200
     },
     {
-      "epoch": 3.5608376963350787,
-      "grad_norm": 1.5374139547348022,
+      "epoch": 3.559463986599665,
+      "grad_norm": 1.1494131088256836,
       "learning_rate": 7.753e-05,
-      "loss": 0.3774,
+      "loss": 0.3821,
       "step": 4250
     },
     {
-      "epoch": 3.602722513089005,
-      "grad_norm": 1.6316169500350952,
+      "epoch": 3.6013400335008376,
+      "grad_norm": 2.058499813079834,
       "learning_rate": 7.703e-05,
-      "loss": 0.3731,
+      "loss": 0.3784,
       "step": 4300
     },
     {
-      "epoch": 3.644607329842932,
-      "grad_norm": 1.749531865119934,
+      "epoch": 3.64321608040201,
+      "grad_norm": 1.8158782720565796,
       "learning_rate": 7.653e-05,
-      "loss": 0.3729,
+      "loss": 0.3833,
       "step": 4350
     },
     {
-      "epoch": 3.6864921465968585,
-      "grad_norm": 1.4812896251678467,
+      "epoch": 3.6850921273031827,
+      "grad_norm": 1.8433794975280762,
       "learning_rate": 7.603e-05,
-      "loss": 0.375,
+      "loss": 0.3768,
       "step": 4400
     },
     {
-      "epoch": 3.7283769633507853,
-      "grad_norm": 1.9387775659561157,
+      "epoch": 3.726968174204355,
+      "grad_norm": 1.6182619333267212,
       "learning_rate": 7.553e-05,
-      "loss": 0.375,
+      "loss": 0.3772,
       "step": 4450
     },
     {
-      "epoch": 3.770261780104712,
-      "grad_norm": 1.4758845567703247,
+      "epoch": 3.7688442211055277,
+      "grad_norm": 1.5222620964050293,
       "learning_rate": 7.503e-05,
-      "loss": 0.3774,
+      "loss": 0.3798,
       "step": 4500
     },
     {
-      "epoch": 3.770261780104712,
-      "eval_loss": 0.3396616578102112,
-      "eval_runtime": 265.2963,
-      "eval_samples_per_second": 31.998,
-      "eval_steps_per_second": 4.003,
+      "epoch": 3.7688442211055277,
+      "eval_loss": 0.3422691822052002,
+      "eval_runtime": 214.805,
+      "eval_samples_per_second": 39.52,
+      "eval_steps_per_second": 4.944,
       "step": 4500
     }
   ],
@@ -728,8 +728,8 @@
       "attributes": {}
     }
   },
-  "total_flos": 4.040100442105517e+16,
-  "train_batch_size": 16,
+  "total_flos": 4.376940325501075e+16,
+  "train_batch_size": 32,
   "trial_name": null,
   "trial_params": null
 }

checkpoint-4500/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:fa8f41edfc5db87ff9f21c8561c4e1b9e9e17c2b5ee34213317807a5e0e41371
+oid sha256:fb8203a5324b57ae1a64f032a5185016d65cbc7738fd0978239bc5b2518d051b
 size 5304