Training in progress, step 100000, checkpoint

checkpoint-100000/model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7559bb10e3cfdfc8f43d34e8dc868d0a445d9bcb2c76e5a04e7671e60ebab7ba
+oid sha256:dbd5a39fbd798a7604003ab05f0e5f0aaa03c3745e48864e76293c496062a15a
 size 51007160

checkpoint-100000/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5cf535e325f076010f92e6c74b81b43d8ef2b22a36c72ce367537b833a9171e9
+oid sha256:a0070cbc39ae4fd8f46c608408a850084667e1bcbec92bcf86a0adb946429a07
 size 102078202

checkpoint-100000/rng_state.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5bec24b2ced95a2472e0a7f9546979437812a7d4cd51b9ba45514a1fd57b6f73
+oid sha256:5a851830c471491452f666a1aa7b4cc675449267d566a3efd736c37d05d7e001
 size 14244

checkpoint-100000/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f69dc9ccdedaf3bfab458442f0bfb1377fc846c74b83d1cda9ba9247794ae3de
+oid sha256:a5006c3420b0fd8d364af0ca32eb23c67b0cd2ef9e5167273a11d19175a43511
 size 1000

checkpoint-100000/trainer_state.json

@@ -1,7 +1,7 @@
 {
-  "best_metric": 3.937915563583374,
-  "best_model_checkpoint": "/home/p318482/babyLM_controlled/models_trained/fr_clm/childes_42/checkpoint-24000",
-  "epoch": 157.23270440251574,
+  "best_metric": 4.579585075378418,
+  "best_model_checkpoint": "/home/p318482/babyLM_controlled/models_trained/de_clm/childes_42/checkpoint-32000",
+  "epoch": 104.82180293501048,
   "eval_steps": 2000,
   "global_step": 100000,
   "is_hyper_param_search": false,
@@ -9,585 +9,585 @@
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 3.1446540880503147,
-      "eval_loss": 6.666613578796387,
-      "eval_runtime": 0.6599,
-      "eval_samples_per_second": 1380.413,
-      "eval_steps_per_second": 86.37,
+      "epoch": 2.0964360587002098,
+      "eval_loss": 7.0908098220825195,
+      "eval_runtime": 0.9834,
+      "eval_samples_per_second": 1398.23,
+      "eval_steps_per_second": 87.453,
       "step": 2000
     },
     {
-      "epoch": 6.289308176100629,
-      "grad_norm": 1.431768536567688,
+      "epoch": 4.1928721174004195,
+      "grad_norm": 1.4699604511260986,
       "learning_rate": 1e-05,
-      "loss": 6.5315,
+      "loss": 6.9765,
       "step": 4000
     },
     {
-      "epoch": 6.289308176100629,
-      "eval_loss": 5.101934432983398,
-      "eval_runtime": 0.6406,
-      "eval_samples_per_second": 1422.145,
-      "eval_steps_per_second": 88.982,
+      "epoch": 4.1928721174004195,
+      "eval_loss": 5.874638557434082,
+      "eval_runtime": 0.973,
+      "eval_samples_per_second": 1413.214,
+      "eval_steps_per_second": 88.39,
       "step": 4000
     },
     {
-      "epoch": 9.433962264150944,
-      "eval_loss": 4.677049160003662,
-      "eval_runtime": 0.6439,
-      "eval_samples_per_second": 1414.724,
-      "eval_steps_per_second": 88.517,
+      "epoch": 6.289308176100629,
+      "eval_loss": 5.544075965881348,
+      "eval_runtime": 0.9898,
+      "eval_samples_per_second": 1389.228,
+      "eval_steps_per_second": 86.89,
       "step": 6000
     },
     {
-      "epoch": 12.578616352201259,
-      "grad_norm": 2.6387760639190674,
-      "learning_rate": 2e-05,
-      "loss": 4.3407,
+      "epoch": 8.385744234800839,
+      "grad_norm": 2.451873779296875,
+      "learning_rate": 1.9997500000000003e-05,
+      "loss": 5.2182,
       "step": 8000
     },
     {
-      "epoch": 12.578616352201259,
-      "eval_loss": 4.4404096603393555,
-      "eval_runtime": 0.6445,
-      "eval_samples_per_second": 1413.55,
-      "eval_steps_per_second": 88.444,
+      "epoch": 8.385744234800839,
+      "eval_loss": 5.278811454772949,
+      "eval_runtime": 0.9809,
+      "eval_samples_per_second": 1401.823,
+      "eval_steps_per_second": 87.678,
       "step": 8000
     },
     {
-      "epoch": 15.723270440251572,
-      "eval_loss": 4.286661624908447,
-      "eval_runtime": 0.6446,
-      "eval_samples_per_second": 1413.35,
-      "eval_steps_per_second": 88.431,
+      "epoch": 10.482180293501049,
+      "eval_loss": 5.099233150482178,
+      "eval_runtime": 0.9694,
+      "eval_samples_per_second": 1418.396,
+      "eval_steps_per_second": 88.714,
       "step": 10000
     },
     {
-      "epoch": 18.867924528301888,
-      "grad_norm": 2.381523609161377,
+      "epoch": 12.578616352201259,
+      "grad_norm": 2.490121841430664,
       "learning_rate": 2.9995e-05,
-      "loss": 3.8725,
+      "loss": 4.7379,
       "step": 12000
     },
     {
-      "epoch": 18.867924528301888,
-      "eval_loss": 4.171908855438232,
-      "eval_runtime": 0.6441,
-      "eval_samples_per_second": 1414.321,
-      "eval_steps_per_second": 88.492,
+      "epoch": 12.578616352201259,
+      "eval_loss": 4.970981597900391,
+      "eval_runtime": 0.9729,
+      "eval_samples_per_second": 1413.251,
+      "eval_steps_per_second": 88.392,
       "step": 12000
     },
     {
-      "epoch": 22.0125786163522,
-      "eval_loss": 4.096552848815918,
-      "eval_runtime": 0.6506,
-      "eval_samples_per_second": 1400.196,
-      "eval_steps_per_second": 87.608,
+      "epoch": 14.675052410901468,
+      "eval_loss": 4.875853061676025,
+      "eval_runtime": 0.9902,
+      "eval_samples_per_second": 1388.55,
+      "eval_steps_per_second": 86.848,
       "step": 14000
     },
     {
-      "epoch": 25.157232704402517,
-      "grad_norm": 2.4479799270629883,
-      "learning_rate": 3.999e-05,
-      "loss": 3.5953,
+      "epoch": 16.771488469601678,
+      "grad_norm": 2.485079765319824,
+      "learning_rate": 3.99925e-05,
+      "loss": 4.4249,
       "step": 16000
     },
     {
-      "epoch": 25.157232704402517,
-      "eval_loss": 4.033241271972656,
-      "eval_runtime": 0.6465,
-      "eval_samples_per_second": 1409.06,
-      "eval_steps_per_second": 88.163,
+      "epoch": 16.771488469601678,
+      "eval_loss": 4.8005266189575195,
+      "eval_runtime": 0.9709,
+      "eval_samples_per_second": 1416.144,
+      "eval_steps_per_second": 88.573,
       "step": 16000
     },
     {
-      "epoch": 28.30188679245283,
-      "eval_loss": 3.9956283569335938,
-      "eval_runtime": 0.6443,
-      "eval_samples_per_second": 1413.837,
-      "eval_steps_per_second": 88.462,
+      "epoch": 18.867924528301888,
+      "eval_loss": 4.743557453155518,
+      "eval_runtime": 0.9764,
+      "eval_samples_per_second": 1408.167,
+      "eval_steps_per_second": 88.074,
       "step": 18000
     },
     {
-      "epoch": 31.446540880503143,
-      "grad_norm": 2.266655921936035,
-      "learning_rate": 4.9985e-05,
-      "loss": 3.3822,
+      "epoch": 20.964360587002098,
+      "grad_norm": 2.311079978942871,
+      "learning_rate": 4.99875e-05,
+      "loss": 4.1842,
       "step": 20000
     },
     {
-      "epoch": 31.446540880503143,
-      "eval_loss": 3.970871686935425,
-      "eval_runtime": 0.6424,
-      "eval_samples_per_second": 1418.161,
-      "eval_steps_per_second": 88.732,
+      "epoch": 20.964360587002098,
+      "eval_loss": 4.692162990570068,
+      "eval_runtime": 0.9728,
+      "eval_samples_per_second": 1413.396,
+      "eval_steps_per_second": 88.401,
       "step": 20000
     },
     {
-      "epoch": 34.59119496855346,
-      "eval_loss": 3.951535224914551,
-      "eval_runtime": 0.6747,
-      "eval_samples_per_second": 1350.174,
-      "eval_steps_per_second": 84.479,
+      "epoch": 23.060796645702307,
+      "eval_loss": 4.648115634918213,
+      "eval_runtime": 0.9711,
+      "eval_samples_per_second": 1415.941,
+      "eval_steps_per_second": 88.561,
       "step": 22000
     },
     {
-      "epoch": 37.735849056603776,
-      "grad_norm": 2.3022546768188477,
-      "learning_rate": 5.9980000000000005e-05,
-      "loss": 3.2038,
+      "epoch": 25.157232704402517,
+      "grad_norm": 2.3314199447631836,
+      "learning_rate": 5.998250000000001e-05,
+      "loss": 3.9843,
       "step": 24000
     },
     {
-      "epoch": 37.735849056603776,
-      "eval_loss": 3.937915563583374,
-      "eval_runtime": 0.6427,
-      "eval_samples_per_second": 1417.462,
-      "eval_steps_per_second": 88.689,
+      "epoch": 25.157232704402517,
+      "eval_loss": 4.615537166595459,
+      "eval_runtime": 0.9745,
+      "eval_samples_per_second": 1410.994,
+      "eval_steps_per_second": 88.251,
       "step": 24000
     },
     {
-      "epoch": 40.880503144654085,
-      "eval_loss": 3.935504913330078,
-      "eval_runtime": 0.6463,
-      "eval_samples_per_second": 1409.641,
-      "eval_steps_per_second": 88.199,
+      "epoch": 27.253668763102727,
+      "eval_loss": 4.59817361831665,
+      "eval_runtime": 0.9725,
+      "eval_samples_per_second": 1413.828,
+      "eval_steps_per_second": 88.429,
       "step": 26000
     },
     {
-      "epoch": 44.0251572327044,
-      "grad_norm": 2.377114772796631,
-      "learning_rate": 6.997500000000001e-05,
-      "loss": 3.0443,
+      "epoch": 29.350104821802937,
+      "grad_norm": 2.2022252082824707,
+      "learning_rate": 6.99775e-05,
+      "loss": 3.8181,
       "step": 28000
     },
     {
-      "epoch": 44.0251572327044,
-      "eval_loss": 3.951632261276245,
-      "eval_runtime": 0.6452,
-      "eval_samples_per_second": 1412.017,
-      "eval_steps_per_second": 88.348,
+      "epoch": 29.350104821802937,
+      "eval_loss": 4.584521293640137,
+      "eval_runtime": 0.9704,
+      "eval_samples_per_second": 1416.917,
+      "eval_steps_per_second": 88.622,
       "step": 28000
     },
     {
-      "epoch": 47.16981132075472,
-      "eval_loss": 3.9560022354125977,
-      "eval_runtime": 0.6457,
-      "eval_samples_per_second": 1410.953,
-      "eval_steps_per_second": 88.281,
+      "epoch": 31.446540880503143,
+      "eval_loss": 4.581137657165527,
+      "eval_runtime": 0.9781,
+      "eval_samples_per_second": 1405.736,
+      "eval_steps_per_second": 87.922,
       "step": 30000
     },
     {
-      "epoch": 50.314465408805034,
-      "grad_norm": 2.4269351959228516,
-      "learning_rate": 7.997e-05,
-      "loss": 2.8937,
+      "epoch": 33.542976939203356,
+      "grad_norm": 2.109912872314453,
+      "learning_rate": 7.997250000000001e-05,
+      "loss": 3.6751,
       "step": 32000
     },
     {
-      "epoch": 50.314465408805034,
-      "eval_loss": 3.9868452548980713,
-      "eval_runtime": 0.642,
-      "eval_samples_per_second": 1418.921,
-      "eval_steps_per_second": 88.78,
+      "epoch": 33.542976939203356,
+      "eval_loss": 4.579585075378418,
+      "eval_runtime": 0.9725,
+      "eval_samples_per_second": 1413.852,
+      "eval_steps_per_second": 88.43,
       "step": 32000
     },
     {
-      "epoch": 53.459119496855344,
-      "eval_loss": 4.025730133056641,
-      "eval_runtime": 0.6451,
-      "eval_samples_per_second": 1412.122,
-      "eval_steps_per_second": 88.355,
+      "epoch": 35.63941299790356,
+      "eval_loss": 4.582756042480469,
+      "eval_runtime": 0.9724,
+      "eval_samples_per_second": 1413.996,
+      "eval_steps_per_second": 88.439,
       "step": 34000
     },
     {
-      "epoch": 56.60377358490566,
-      "grad_norm": 2.4868056774139404,
-      "learning_rate": 8.9965e-05,
-      "loss": 2.7503,
+      "epoch": 37.735849056603776,
+      "grad_norm": 2.1550486087799072,
+      "learning_rate": 8.996750000000001e-05,
+      "loss": 3.5484,
       "step": 36000
     },
     {
-      "epoch": 56.60377358490566,
-      "eval_loss": 4.050300598144531,
-      "eval_runtime": 0.6446,
-      "eval_samples_per_second": 1413.342,
-      "eval_steps_per_second": 88.431,
+      "epoch": 37.735849056603776,
+      "eval_loss": 4.586859226226807,
+      "eval_runtime": 0.979,
+      "eval_samples_per_second": 1404.481,
+      "eval_steps_per_second": 87.844,
       "step": 36000
     },
     {
-      "epoch": 59.74842767295598,
-      "eval_loss": 4.09123420715332,
-      "eval_runtime": 0.6444,
-      "eval_samples_per_second": 1413.818,
-      "eval_steps_per_second": 88.461,
+      "epoch": 39.83228511530398,
+      "eval_loss": 4.597586154937744,
+      "eval_runtime": 0.9841,
+      "eval_samples_per_second": 1397.167,
+      "eval_steps_per_second": 87.386,
       "step": 38000
     },
     {
-      "epoch": 62.893081761006286,
-      "grad_norm": 2.6151175498962402,
-      "learning_rate": 9.996000000000001e-05,
-      "loss": 2.6083,
+      "epoch": 41.928721174004195,
+      "grad_norm": 2.2574758529663086,
+      "learning_rate": 9.99625e-05,
+      "loss": 3.4328,
       "step": 40000
     },
     {
-      "epoch": 62.893081761006286,
-      "eval_loss": 4.143652439117432,
-      "eval_runtime": 0.6467,
-      "eval_samples_per_second": 1408.69,
-      "eval_steps_per_second": 88.14,
+      "epoch": 41.928721174004195,
+      "eval_loss": 4.6090407371521,
+      "eval_runtime": 0.9945,
+      "eval_samples_per_second": 1382.571,
+      "eval_steps_per_second": 86.474,
       "step": 40000
     },
     {
-      "epoch": 66.0377358490566,
-      "eval_loss": 4.220719814300537,
-      "eval_runtime": 0.6466,
-      "eval_samples_per_second": 1408.922,
-      "eval_steps_per_second": 88.154,
+      "epoch": 44.0251572327044,
+      "eval_loss": 4.629756450653076,
+      "eval_runtime": 0.9745,
+      "eval_samples_per_second": 1410.989,
+      "eval_steps_per_second": 88.251,
       "step": 42000
     },
     {
-      "epoch": 69.18238993710692,
-      "grad_norm": 2.823240280151367,
-      "learning_rate": 9.336333333333334e-05,
-      "loss": 2.4507,
+      "epoch": 46.121593291404615,
+      "grad_norm": 2.2658331394195557,
+      "learning_rate": 9.336166666666667e-05,
+      "loss": 3.31,
       "step": 44000
     },
     {
-      "epoch": 69.18238993710692,
-      "eval_loss": 4.291810989379883,
-      "eval_runtime": 0.6433,
-      "eval_samples_per_second": 1416.201,
-      "eval_steps_per_second": 88.61,
+      "epoch": 46.121593291404615,
+      "eval_loss": 4.659794330596924,
+      "eval_runtime": 0.9812,
+      "eval_samples_per_second": 1401.309,
+      "eval_steps_per_second": 87.645,
       "step": 44000
     },
     {
-      "epoch": 72.32704402515724,
-      "eval_loss": 4.3598246574401855,
-      "eval_runtime": 0.6463,
-      "eval_samples_per_second": 1409.638,
-      "eval_steps_per_second": 88.199,
+      "epoch": 48.21802935010482,
+      "eval_loss": 4.698253631591797,
+      "eval_runtime": 0.981,
+      "eval_samples_per_second": 1401.638,
+      "eval_steps_per_second": 87.666,
       "step": 46000
     },
     {
-      "epoch": 75.47169811320755,
-      "grad_norm": 3.11128830909729,
-      "learning_rate": 8.67e-05,
-      "loss": 2.2924,
+      "epoch": 50.314465408805034,
+      "grad_norm": 2.4032843112945557,
+      "learning_rate": 8.669833333333334e-05,
+      "loss": 3.1908,
       "step": 48000
     },
     {
-      "epoch": 75.47169811320755,
-      "eval_loss": 4.444736003875732,
-      "eval_runtime": 0.6454,
-      "eval_samples_per_second": 1411.606,
-      "eval_steps_per_second": 88.322,
+      "epoch": 50.314465408805034,
+      "eval_loss": 4.72628927230835,
+      "eval_runtime": 0.9912,
+      "eval_samples_per_second": 1387.274,
+      "eval_steps_per_second": 86.768,
       "step": 48000
     },
     {
-      "epoch": 78.61635220125787,
-      "eval_loss": 4.5188703536987305,
-      "eval_runtime": 0.6418,
-      "eval_samples_per_second": 1419.435,
-      "eval_steps_per_second": 88.812,
+      "epoch": 52.41090146750524,
+      "eval_loss": 4.762426376342773,
+      "eval_runtime": 0.9771,
+      "eval_samples_per_second": 1407.288,
+      "eval_steps_per_second": 88.019,
       "step": 50000
     },
     {
-      "epoch": 81.76100628930817,
-      "grad_norm": 3.3800017833709717,
-      "learning_rate": 8.003666666666667e-05,
-      "loss": 2.1545,
+      "epoch": 54.507337526205454,
+      "grad_norm": 2.641611337661743,
+      "learning_rate": 8.0035e-05,
+      "loss": 3.0864,
       "step": 52000
     },
     {
-      "epoch": 81.76100628930817,
-      "eval_loss": 4.572214126586914,
-      "eval_runtime": 0.6428,
-      "eval_samples_per_second": 1417.247,
-      "eval_steps_per_second": 88.675,
+      "epoch": 54.507337526205454,
+      "eval_loss": 4.79134464263916,
+      "eval_runtime": 0.9732,
+      "eval_samples_per_second": 1412.893,
+      "eval_steps_per_second": 88.37,
       "step": 52000
     },
     {
-      "epoch": 84.90566037735849,
-      "eval_loss": 4.64493465423584,
-      "eval_runtime": 0.6583,
-      "eval_samples_per_second": 1383.767,
-      "eval_steps_per_second": 86.58,
+      "epoch": 56.60377358490566,
+      "eval_loss": 4.826347827911377,
+      "eval_runtime": 0.9719,
+      "eval_samples_per_second": 1414.733,
+      "eval_steps_per_second": 88.485,
       "step": 54000
     },
     {
-      "epoch": 88.0503144654088,
-      "grad_norm": 3.550813674926758,
-      "learning_rate": 7.337333333333334e-05,
-      "loss": 2.0329,
+      "epoch": 58.700209643605874,
+      "grad_norm": 2.6470491886138916,
+      "learning_rate": 7.337166666666667e-05,
+      "loss": 2.993,
       "step": 56000
     },
     {
-      "epoch": 88.0503144654088,
-      "eval_loss": 4.751666069030762,
-      "eval_runtime": 0.6576,
-      "eval_samples_per_second": 1385.337,
-      "eval_steps_per_second": 86.679,
+      "epoch": 58.700209643605874,
+      "eval_loss": 4.853786945343018,
+      "eval_runtime": 0.975,
+      "eval_samples_per_second": 1410.209,
+      "eval_steps_per_second": 88.202,
       "step": 56000
     },
     {
-      "epoch": 91.19496855345912,
-      "eval_loss": 4.808305263519287,
-      "eval_runtime": 0.6473,
-      "eval_samples_per_second": 1407.397,
-      "eval_steps_per_second": 88.059,
+      "epoch": 60.79664570230608,
+      "eval_loss": 4.8770365715026855,
+      "eval_runtime": 0.9725,
+      "eval_samples_per_second": 1413.926,
+      "eval_steps_per_second": 88.435,
       "step": 58000
     },
     {
-      "epoch": 94.33962264150944,
-      "grad_norm": 3.9259555339813232,
-      "learning_rate": 6.671e-05,
-      "loss": 1.9237,
+      "epoch": 62.893081761006286,
+      "grad_norm": 2.761204957962036,
+      "learning_rate": 6.670833333333333e-05,
+      "loss": 2.9108,
       "step": 60000
     },
     {
-      "epoch": 94.33962264150944,
-      "eval_loss": 4.880630016326904,
-      "eval_runtime": 0.6462,
-      "eval_samples_per_second": 1409.79,
-      "eval_steps_per_second": 88.209,
+      "epoch": 62.893081761006286,
+      "eval_loss": 4.909708499908447,
+      "eval_runtime": 0.9753,
+      "eval_samples_per_second": 1409.824,
+      "eval_steps_per_second": 88.178,
       "step": 60000
     },
     {
-      "epoch": 97.48427672955975,
-      "eval_loss": 4.933507919311523,
-      "eval_runtime": 0.6628,
-      "eval_samples_per_second": 1374.396,
-      "eval_steps_per_second": 85.994,
+      "epoch": 64.98951781970649,
+      "eval_loss": 4.94859504699707,
+      "eval_runtime": 0.9735,
+      "eval_samples_per_second": 1412.386,
+      "eval_steps_per_second": 88.338,
       "step": 62000
     },
     {
-      "epoch": 100.62893081761007,
-      "grad_norm": 3.90450382232666,
-      "learning_rate": 6.0046666666666676e-05,
-      "loss": 1.833,
+      "epoch": 67.08595387840671,
+      "grad_norm": 3.005192279815674,
+      "learning_rate": 6.0045000000000005e-05,
+      "loss": 2.8352,
       "step": 64000
     },
     {
-      "epoch": 100.62893081761007,
-      "eval_loss": 4.989780902862549,
-      "eval_runtime": 0.644,
-      "eval_samples_per_second": 1414.675,
-      "eval_steps_per_second": 88.514,
+      "epoch": 67.08595387840671,
+      "eval_loss": 4.992880821228027,
+      "eval_runtime": 0.9755,
+      "eval_samples_per_second": 1409.563,
+      "eval_steps_per_second": 88.162,
       "step": 64000
     },
     {
-      "epoch": 103.77358490566037,
-      "eval_loss": 5.051305770874023,
-      "eval_runtime": 0.645,
-      "eval_samples_per_second": 1412.427,
-      "eval_steps_per_second": 88.374,
+      "epoch": 69.18238993710692,
+      "eval_loss": 5.033893585205078,
+      "eval_runtime": 0.9733,
+      "eval_samples_per_second": 1412.778,
+      "eval_steps_per_second": 88.363,
       "step": 66000
     },
     {
-      "epoch": 106.91823899371069,
-      "grad_norm": 4.007218837738037,
-      "learning_rate": 5.338333333333334e-05,
-      "loss": 1.7526,
+      "epoch": 71.27882599580713,
+      "grad_norm": 2.9778764247894287,
+      "learning_rate": 5.338166666666668e-05,
+      "loss": 2.7677,
       "step": 68000
     },
     {
-      "epoch": 106.91823899371069,
-      "eval_loss": 5.101851940155029,
-      "eval_runtime": 0.6464,
-      "eval_samples_per_second": 1409.425,
-      "eval_steps_per_second": 88.186,
+      "epoch": 71.27882599580713,
+      "eval_loss": 5.051624774932861,
+      "eval_runtime": 0.9743,
+      "eval_samples_per_second": 1411.21,
+      "eval_steps_per_second": 88.265,
       "step": 68000
     },
     {
-      "epoch": 110.062893081761,
-      "eval_loss": 5.1573052406311035,
-      "eval_runtime": 0.6554,
-      "eval_samples_per_second": 1389.962,
-      "eval_steps_per_second": 86.968,
+      "epoch": 73.37526205450733,
+      "eval_loss": 5.0868730545043945,
+      "eval_runtime": 0.9734,
+      "eval_samples_per_second": 1412.585,
+      "eval_steps_per_second": 88.351,
       "step": 70000
     },
     {
-      "epoch": 113.20754716981132,
-      "grad_norm": 3.952075958251953,
-      "learning_rate": 4.671666666666667e-05,
-      "loss": 1.6779,
+      "epoch": 75.47169811320755,
+      "grad_norm": 3.199427366256714,
+      "learning_rate": 4.6718333333333336e-05,
+      "loss": 2.708,
       "step": 72000
     },
     {
-      "epoch": 113.20754716981132,
-      "eval_loss": 5.2098236083984375,
-      "eval_runtime": 0.6442,
-      "eval_samples_per_second": 1414.099,
-      "eval_steps_per_second": 88.478,
+      "epoch": 75.47169811320755,
+      "eval_loss": 5.107797622680664,
+      "eval_runtime": 0.9734,
+      "eval_samples_per_second": 1412.585,
+      "eval_steps_per_second": 88.351,
       "step": 72000
     },
     {
-      "epoch": 116.35220125786164,
-      "eval_loss": 5.253559112548828,
-      "eval_runtime": 0.6424,
-      "eval_samples_per_second": 1418.023,
-      "eval_steps_per_second": 88.724,
+      "epoch": 77.56813417190776,
+      "eval_loss": 5.131660461425781,
+      "eval_runtime": 0.9755,
+      "eval_samples_per_second": 1409.548,
+      "eval_steps_per_second": 88.161,
       "step": 74000
     },
     {
-      "epoch": 119.49685534591195,
-      "grad_norm": 4.343034744262695,
-      "learning_rate": 4.0050000000000004e-05,
-      "loss": 1.6143,
+      "epoch": 79.66457023060796,
+      "grad_norm": 3.2438466548919678,
+      "learning_rate": 4.0055e-05,
+      "loss": 2.6552,
       "step": 76000
     },
     {
-      "epoch": 119.49685534591195,
-      "eval_loss": 5.306056022644043,
-      "eval_runtime": 0.6406,
-      "eval_samples_per_second": 1422.045,
-      "eval_steps_per_second": 88.975,
+      "epoch": 79.66457023060796,
+      "eval_loss": 5.159754276275635,
+      "eval_runtime": 0.9741,
+      "eval_samples_per_second": 1411.574,
+      "eval_steps_per_second": 88.288,
       "step": 76000
     },
     {
-      "epoch": 122.64150943396227,
-      "eval_loss": 5.35061502456665,
-      "eval_runtime": 0.6403,
-      "eval_samples_per_second": 1422.818,
-      "eval_steps_per_second": 89.024,
+      "epoch": 81.76100628930817,
+      "eval_loss": 5.17739200592041,
+      "eval_runtime": 0.9699,
+      "eval_samples_per_second": 1417.607,
+      "eval_steps_per_second": 88.665,
       "step": 78000
     },
     {
-      "epoch": 125.78616352201257,
-      "grad_norm": 4.502026557922363,
-      "learning_rate": 3.338666666666666e-05,
-      "loss": 1.5572,
+      "epoch": 83.85744234800839,
+      "grad_norm": 3.3774938583374023,
+      "learning_rate": 3.339166666666667e-05,
+      "loss": 2.6082,
       "step": 80000
     },
     {
-      "epoch": 125.78616352201257,
-      "eval_loss": 5.388923645019531,
-      "eval_runtime": 0.6446,
-      "eval_samples_per_second": 1413.291,
-      "eval_steps_per_second": 88.428,
+      "epoch": 83.85744234800839,
+      "eval_loss": 5.192777156829834,
+      "eval_runtime": 0.9709,
+      "eval_samples_per_second": 1416.242,
+      "eval_steps_per_second": 88.579,
       "step": 80000
     },
     {
-      "epoch": 128.9308176100629,
-      "eval_loss": 5.426227569580078,
-      "eval_runtime": 0.641,
-      "eval_samples_per_second": 1421.183,
-      "eval_steps_per_second": 88.921,
+      "epoch": 85.9538784067086,
+      "eval_loss": 5.227346897125244,
+      "eval_runtime": 0.9893,
+      "eval_samples_per_second": 1389.87,
+      "eval_steps_per_second": 86.93,
       "step": 82000
     },
     {
-      "epoch": 132.0754716981132,
-      "grad_norm": 4.323946475982666,
-      "learning_rate": 2.6723333333333332e-05,
-      "loss": 1.5054,
+      "epoch": 88.0503144654088,
+      "grad_norm": 3.3834476470947266,
+      "learning_rate": 2.6728333333333333e-05,
+      "loss": 2.5633,
       "step": 84000
     },
     {
-      "epoch": 132.0754716981132,
-      "eval_loss": 5.468140602111816,
-      "eval_runtime": 0.6422,
-      "eval_samples_per_second": 1418.578,
-      "eval_steps_per_second": 88.758,
+      "epoch": 88.0503144654088,
+      "eval_loss": 5.249727725982666,
+      "eval_runtime": 0.9723,
+      "eval_samples_per_second": 1414.208,
+      "eval_steps_per_second": 88.452,
       "step": 84000
     },
     {
-      "epoch": 135.22012578616352,
-      "eval_loss": 5.506562232971191,
-      "eval_runtime": 0.643,
-      "eval_samples_per_second": 1416.831,
-      "eval_steps_per_second": 88.649,
+      "epoch": 90.14675052410901,
+      "eval_loss": 5.264400959014893,
+      "eval_runtime": 0.9715,
+      "eval_samples_per_second": 1415.317,
+      "eval_steps_per_second": 88.522,
       "step": 86000
     },
     {
-      "epoch": 138.36477987421384,
-      "grad_norm": 4.422414302825928,
-      "learning_rate": 2.006e-05,
-      "loss": 1.4587,
+      "epoch": 92.24318658280923,
+      "grad_norm": 3.5035271644592285,
+      "learning_rate": 2.0065000000000002e-05,
+      "loss": 2.5227,
       "step": 88000
     },
     {
-      "epoch": 138.36477987421384,
-      "eval_loss": 5.5312581062316895,
-      "eval_runtime": 0.6438,
-      "eval_samples_per_second": 1415.14,
-      "eval_steps_per_second": 88.543,
+      "epoch": 92.24318658280923,
+      "eval_loss": 5.28403902053833,
+      "eval_runtime": 0.977,
+      "eval_samples_per_second": 1407.338,
+      "eval_steps_per_second": 88.023,
       "step": 88000
     },
     {
-      "epoch": 141.50943396226415,
-      "eval_loss": 5.551713466644287,
-      "eval_runtime": 0.6408,
-      "eval_samples_per_second": 1421.582,
-      "eval_steps_per_second": 88.946,
+      "epoch": 94.33962264150944,
+      "eval_loss": 5.292140483856201,
+      "eval_runtime": 0.9839,
+      "eval_samples_per_second": 1397.466,
+      "eval_steps_per_second": 87.405,
       "step": 90000
     },
     {
-      "epoch": 144.65408805031447,
-      "grad_norm": 4.622466564178467,
-      "learning_rate": 1.3396666666666668e-05,
-      "loss": 1.4189,
+      "epoch": 96.43605870020964,
+      "grad_norm": 3.6662752628326416,
+      "learning_rate": 1.3400000000000002e-05,
+      "loss": 2.4873,
       "step": 92000
     },
     {
-      "epoch": 144.65408805031447,
-      "eval_loss": 5.576663017272949,
-      "eval_runtime": 0.6402,
-      "eval_samples_per_second": 1423.05,
-      "eval_steps_per_second": 89.038,
+      "epoch": 96.43605870020964,
+      "eval_loss": 5.311823844909668,
+      "eval_runtime": 0.9742,
+      "eval_samples_per_second": 1411.403,
+      "eval_steps_per_second": 88.277,
       "step": 92000
     },
     {
-      "epoch": 147.7987421383648,
-      "eval_loss": 5.598996162414551,
-      "eval_runtime": 0.646,
-      "eval_samples_per_second": 1410.272,
-      "eval_steps_per_second": 88.239,
+      "epoch": 98.53249475890985,
+      "eval_loss": 5.320519924163818,
+      "eval_runtime": 0.9772,
+      "eval_samples_per_second": 1407.037,
+      "eval_steps_per_second": 88.004,
       "step": 94000
     },
     {
-      "epoch": 150.9433962264151,
-      "grad_norm": 4.559875011444092,
-      "learning_rate": 6.733333333333333e-06,
-      "loss": 1.3839,
+      "epoch": 100.62893081761007,
+      "grad_norm": 3.796828031539917,
+      "learning_rate": 6.736666666666668e-06,
+      "loss": 2.458,
       "step": 96000
     },
     {
-      "epoch": 150.9433962264151,
-      "eval_loss": 5.617443561553955,
-      "eval_runtime": 0.6446,
-      "eval_samples_per_second": 1413.201,
-      "eval_steps_per_second": 88.422,
+      "epoch": 100.62893081761007,
+      "eval_loss": 5.330761909484863,
+      "eval_runtime": 0.9782,
+      "eval_samples_per_second": 1405.577,
+      "eval_steps_per_second": 87.912,
       "step": 96000
     },
     {
-      "epoch": 154.08805031446542,
-      "eval_loss": 5.624011993408203,
-      "eval_runtime": 0.6442,
-      "eval_samples_per_second": 1414.112,
-      "eval_steps_per_second": 88.479,
+      "epoch": 102.72536687631028,
+      "eval_loss": 5.336517810821533,
+      "eval_runtime": 0.9705,
+      "eval_samples_per_second": 1416.858,
+      "eval_steps_per_second": 88.618,
       "step": 98000
     },
     {
-      "epoch": 157.23270440251574,
-      "grad_norm": 4.707168102264404,
-      "learning_rate": 7e-08,
-      "loss": 1.3559,
+      "epoch": 104.82180293501048,
+      "grad_norm": 3.6499862670898438,
+      "learning_rate": 7.333333333333334e-08,
+      "loss": 2.4331,
       "step": 100000
     },
     {
-      "epoch": 157.23270440251574,
-      "eval_loss": 5.62923526763916,
-      "eval_runtime": 0.6428,
-      "eval_samples_per_second": 1417.264,
-      "eval_steps_per_second": 88.676,
+      "epoch": 104.82180293501048,
+      "eval_loss": 5.338971138000488,
+      "eval_runtime": 0.9811,
+      "eval_samples_per_second": 1401.545,
+      "eval_steps_per_second": 87.66,
       "step": 100000
     }
   ],
   "logging_steps": 4000,
   "max_steps": 100000,
   "num_input_tokens_seen": 0,
-  "num_train_epochs": 158,
+  "num_train_epochs": 105,
   "save_steps": 4000,
   "stateful_callbacks": {
     "TrainerControl": {
@@ -601,7 +601,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 2.5854087677362176e+16,
+  "total_flos": 2.5856623848751104e+16,
   "train_batch_size": 16,
   "trial_name": null,
   "trial_params": null

checkpoint-100000/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:aeaea898ceb075df76fa5bb83c2bfb29fe7b45a806b68020f1e85d260c39893a
+oid sha256:7e7557b0dc8ef07c9db255e1167d98819525b043deb4934f580e21d716914111
 size 5368