Training in progress, step 100000, checkpoint

checkpoint-100000/model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:fd72c9b44ae0d5cdae58ebd673076b11b94ffc9f03e54cf93d5ff95cd1f87073
+oid sha256:4ed9362952b4fd87e379a57f57f1be931144492d1af09c3d76139f60bb946e69
 size 51007160

checkpoint-100000/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2f67d291f29a655135c26d130495b93bb28a879980013ef9cf7867041b87ac36
+oid sha256:2bc76a1f89eb1f131e0d31c61897d2b3af6d5bbb476dcfc8f2d42bec7a809ebb
 size 102078202

checkpoint-100000/rng_state.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c3b9085b50c19f9ac60c90400a9c1e9d625894bc0cec5d57cba56af004f51e3e
+oid sha256:6e4af4adaf699c8fb88a64f00706b102e2f46f8cf32a083c3f6add1ead4b5e19
 size 14308

checkpoint-100000/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:06aecbd3f5ff905306913a3e00527d5fa2c1773d23de5143e3eb9e01b973eb0f
+oid sha256:a5006c3420b0fd8d364af0ca32eb23c67b0cd2ef9e5167273a11d19175a43511
 size 1000

checkpoint-100000/trainer_state.json

@@ -1,7 +1,7 @@
 {
-  "best_metric": 4.224654674530029,
-  "best_model_checkpoint": "/home/p318482/babyLM_controlled/models_trained/en_clm/wikipedia_30/checkpoint-72000",
-  "epoch": 52.59006047856955,
+  "best_metric": 3.5649592876434326,
+  "best_model_checkpoint": "/home/p318482/babyLM_controlled/models_trained/fr_clm/wikipedia_30/checkpoint-36000",
+  "epoch": 107.35373054213633,
   "eval_steps": 2000,
   "global_step": 100000,
   "is_hyper_param_search": false,
@@ -9,585 +9,585 @@
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 1.0518012095713911,
-      "eval_loss": 7.518920421600342,
-      "eval_runtime": 2.2399,
-      "eval_samples_per_second": 1422.808,
-      "eval_steps_per_second": 89.288,
+      "epoch": 2.1470746108427265,
+      "eval_loss": 7.138043403625488,
+      "eval_runtime": 0.8415,
+      "eval_samples_per_second": 1273.93,
+      "eval_steps_per_second": 79.621,
       "step": 2000
     },
     {
-      "epoch": 2.1036024191427822,
-      "grad_norm": 1.2947496175765991,
+      "epoch": 4.294149221685453,
+      "grad_norm": 1.455039381980896,
       "learning_rate": 1e-05,
-      "loss": 7.6141,
+      "loss": 7.213,
       "step": 4000
     },
     {
-      "epoch": 2.1036024191427822,
-      "eval_loss": 6.568221092224121,
-      "eval_runtime": 2.1236,
-      "eval_samples_per_second": 1500.783,
-      "eval_steps_per_second": 94.182,
+      "epoch": 4.294149221685453,
+      "eval_loss": 5.8541717529296875,
+      "eval_runtime": 0.7753,
+      "eval_samples_per_second": 1382.676,
+      "eval_steps_per_second": 86.417,
       "step": 4000
     },
     {
-      "epoch": 3.155403628714173,
-      "eval_loss": 6.16925573348999,
-      "eval_runtime": 2.1624,
-      "eval_samples_per_second": 1473.825,
-      "eval_steps_per_second": 92.49,
+      "epoch": 6.4412238325281805,
+      "eval_loss": 5.403579235076904,
+      "eval_runtime": 0.7542,
+      "eval_samples_per_second": 1421.437,
+      "eval_steps_per_second": 88.84,
       "step": 6000
     },
     {
-      "epoch": 4.2072048382855645,
-      "grad_norm": 2.330080270767212,
-      "learning_rate": 2e-05,
-      "loss": 6.2396,
+      "epoch": 8.588298443370906,
+      "grad_norm": 3.117489814758301,
+      "learning_rate": 1.9997500000000003e-05,
+      "loss": 5.4304,
       "step": 8000
     },
     {
-      "epoch": 4.2072048382855645,
-      "eval_loss": 5.901575088500977,
-      "eval_runtime": 2.1326,
-      "eval_samples_per_second": 1494.388,
-      "eval_steps_per_second": 93.78,
+      "epoch": 8.588298443370906,
+      "eval_loss": 5.049880504608154,
+      "eval_runtime": 0.7597,
+      "eval_samples_per_second": 1411.127,
+      "eval_steps_per_second": 88.195,
       "step": 8000
     },
     {
-      "epoch": 5.259006047856955,
-      "eval_loss": 5.67369270324707,
-      "eval_runtime": 2.1364,
-      "eval_samples_per_second": 1491.788,
-      "eval_steps_per_second": 93.617,
+      "epoch": 10.735373054213634,
+      "eval_loss": 4.760603427886963,
+      "eval_runtime": 0.7624,
+      "eval_samples_per_second": 1406.053,
+      "eval_steps_per_second": 87.878,
       "step": 10000
     },
     {
-      "epoch": 6.310807257428346,
-      "grad_norm": 2.565220594406128,
-      "learning_rate": 2.99925e-05,
-      "loss": 5.7217,
+      "epoch": 12.882447665056361,
+      "grad_norm": 2.9511141777038574,
+      "learning_rate": 2.9995e-05,
+      "loss": 4.771,
       "step": 12000
     },
     {
-      "epoch": 6.310807257428346,
-      "eval_loss": 5.460005760192871,
-      "eval_runtime": 2.1835,
-      "eval_samples_per_second": 1459.563,
-      "eval_steps_per_second": 91.595,
+      "epoch": 12.882447665056361,
+      "eval_loss": 4.517208576202393,
+      "eval_runtime": 0.764,
+      "eval_samples_per_second": 1403.097,
+      "eval_steps_per_second": 87.694,
       "step": 12000
     },
     {
-      "epoch": 7.362608466999737,
-      "eval_loss": 5.272489547729492,
-      "eval_runtime": 2.1833,
-      "eval_samples_per_second": 1459.691,
-      "eval_steps_per_second": 91.603,
+      "epoch": 15.029522275899087,
+      "eval_loss": 4.3206024169921875,
+      "eval_runtime": 0.7595,
+      "eval_samples_per_second": 1411.435,
+      "eval_steps_per_second": 88.215,
       "step": 14000
     },
     {
-      "epoch": 8.414409676571129,
-      "grad_norm": 2.7334115505218506,
-      "learning_rate": 3.99875e-05,
-      "loss": 5.3064,
+      "epoch": 17.176596886741816,
+      "grad_norm": 2.955515146255493,
+      "learning_rate": 3.999e-05,
+      "loss": 4.2888,
       "step": 16000
     },
     {
-      "epoch": 8.414409676571129,
-      "eval_loss": 5.1074652671813965,
-      "eval_runtime": 2.2402,
-      "eval_samples_per_second": 1422.652,
-      "eval_steps_per_second": 89.278,
+      "epoch": 17.176596886741816,
+      "eval_loss": 4.152973175048828,
+      "eval_runtime": 0.8247,
+      "eval_samples_per_second": 1299.91,
+      "eval_steps_per_second": 81.244,
       "step": 16000
     },
     {
-      "epoch": 9.46621088614252,
-      "eval_loss": 4.9716596603393555,
-      "eval_runtime": 2.1736,
-      "eval_samples_per_second": 1466.258,
-      "eval_steps_per_second": 92.015,
+      "epoch": 19.32367149758454,
+      "eval_loss": 4.015476703643799,
+      "eval_runtime": 0.7501,
+      "eval_samples_per_second": 1429.237,
+      "eval_steps_per_second": 89.327,
       "step": 18000
     },
     {
-      "epoch": 10.51801209571391,
-      "grad_norm": 2.6722376346588135,
-      "learning_rate": 4.9985e-05,
-      "loss": 4.9744,
+      "epoch": 21.470746108427267,
+      "grad_norm": 3.0531320571899414,
+      "learning_rate": 4.99875e-05,
+      "loss": 3.9141,
       "step": 20000
     },
     {
-      "epoch": 10.51801209571391,
-      "eval_loss": 4.864916801452637,
-      "eval_runtime": 2.2031,
-      "eval_samples_per_second": 1446.585,
-      "eval_steps_per_second": 90.78,
+      "epoch": 21.470746108427267,
+      "eval_loss": 3.896575927734375,
+      "eval_runtime": 0.768,
+      "eval_samples_per_second": 1395.843,
+      "eval_steps_per_second": 87.24,
       "step": 20000
     },
     {
-      "epoch": 11.569813305285301,
-      "eval_loss": 4.763542652130127,
-      "eval_runtime": 2.4382,
-      "eval_samples_per_second": 1307.131,
-      "eval_steps_per_second": 82.029,
+      "epoch": 23.617820719269993,
+      "eval_loss": 3.8046905994415283,
+      "eval_runtime": 0.7601,
+      "eval_samples_per_second": 1410.272,
+      "eval_steps_per_second": 88.142,
       "step": 22000
     },
     {
-      "epoch": 12.621614514856692,
-      "grad_norm": 2.4539408683776855,
-      "learning_rate": 5.9985e-05,
-      "loss": 4.7273,
+      "epoch": 25.764895330112722,
+      "grad_norm": 3.089989185333252,
+      "learning_rate": 5.998250000000001e-05,
+      "loss": 3.6154,
       "step": 24000
     },
     {
-      "epoch": 12.621614514856692,
-      "eval_loss": 4.683297634124756,
-      "eval_runtime": 2.452,
-      "eval_samples_per_second": 1299.733,
-      "eval_steps_per_second": 81.565,
+      "epoch": 25.764895330112722,
+      "eval_loss": 3.7359092235565186,
+      "eval_runtime": 0.7778,
+      "eval_samples_per_second": 1378.238,
+      "eval_steps_per_second": 86.14,
       "step": 24000
     },
     {
-      "epoch": 13.673415724428082,
-      "eval_loss": 4.621574401855469,
-      "eval_runtime": 2.1825,
-      "eval_samples_per_second": 1460.226,
-      "eval_steps_per_second": 91.636,
+      "epoch": 27.911969940955448,
+      "eval_loss": 3.6784231662750244,
+      "eval_runtime": 0.7541,
+      "eval_samples_per_second": 1421.491,
+      "eval_steps_per_second": 88.843,
       "step": 26000
     },
     {
-      "epoch": 14.725216933999475,
-      "grad_norm": 2.4913346767425537,
-      "learning_rate": 6.998e-05,
-      "loss": 4.5397,
+      "epoch": 30.059044551798173,
+      "grad_norm": 3.3495922088623047,
+      "learning_rate": 6.997500000000001e-05,
+      "loss": 3.3661,
       "step": 28000
     },
     {
-      "epoch": 14.725216933999475,
-      "eval_loss": 4.562623977661133,
-      "eval_runtime": 2.201,
-      "eval_samples_per_second": 1447.965,
-      "eval_steps_per_second": 90.867,
+      "epoch": 30.059044551798173,
+      "eval_loss": 3.6360318660736084,
+      "eval_runtime": 0.7603,
+      "eval_samples_per_second": 1409.913,
+      "eval_steps_per_second": 88.12,
       "step": 28000
     },
     {
-      "epoch": 15.777018143570865,
-      "eval_loss": 4.506168842315674,
-      "eval_runtime": 2.2491,
-      "eval_samples_per_second": 1416.988,
-      "eval_steps_per_second": 88.923,
+      "epoch": 32.2061191626409,
+      "eval_loss": 3.6019041538238525,
+      "eval_runtime": 0.7638,
+      "eval_samples_per_second": 1403.547,
+      "eval_steps_per_second": 87.722,
       "step": 30000
     },
     {
-      "epoch": 16.828819353142258,
-      "grad_norm": 2.2661523818969727,
-      "learning_rate": 7.99775e-05,
-      "loss": 4.3839,
+      "epoch": 34.35319377348363,
+      "grad_norm": 3.1256051063537598,
+      "learning_rate": 7.997250000000001e-05,
+      "loss": 3.1473,
       "step": 32000
     },
     {
-      "epoch": 16.828819353142258,
-      "eval_loss": 4.464449882507324,
-      "eval_runtime": 2.253,
-      "eval_samples_per_second": 1414.55,
-      "eval_steps_per_second": 88.77,
+      "epoch": 34.35319377348363,
+      "eval_loss": 3.5816054344177246,
+      "eval_runtime": 0.7628,
+      "eval_samples_per_second": 1405.391,
+      "eval_steps_per_second": 87.837,
       "step": 32000
     },
     {
-      "epoch": 17.88062056271365,
-      "eval_loss": 4.426318168640137,
-      "eval_runtime": 2.2688,
-      "eval_samples_per_second": 1404.709,
-      "eval_steps_per_second": 88.152,
+      "epoch": 36.500268384326354,
+      "eval_loss": 3.5698702335357666,
+      "eval_runtime": 0.7617,
+      "eval_samples_per_second": 1407.381,
+      "eval_steps_per_second": 87.961,
       "step": 34000
     },
     {
-      "epoch": 18.93242177228504,
-      "grad_norm": 2.226025104522705,
-      "learning_rate": 8.9975e-05,
-      "loss": 4.2529,
+      "epoch": 38.64734299516908,
+      "grad_norm": 2.921318769454956,
+      "learning_rate": 8.997000000000001e-05,
+      "loss": 2.9533,
       "step": 36000
     },
     {
-      "epoch": 18.93242177228504,
-      "eval_loss": 4.3944478034973145,
-      "eval_runtime": 2.294,
-      "eval_samples_per_second": 1389.273,
-      "eval_steps_per_second": 87.184,
+      "epoch": 38.64734299516908,
+      "eval_loss": 3.5649592876434326,
+      "eval_runtime": 0.7607,
+      "eval_samples_per_second": 1409.149,
+      "eval_steps_per_second": 88.072,
       "step": 36000
     },
     {
-      "epoch": 19.98422298185643,
-      "eval_loss": 4.365135669708252,
-      "eval_runtime": 2.2176,
-      "eval_samples_per_second": 1437.165,
-      "eval_steps_per_second": 90.189,
+      "epoch": 40.794417606011805,
+      "eval_loss": 3.5666539669036865,
+      "eval_runtime": 0.7558,
+      "eval_samples_per_second": 1418.297,
+      "eval_steps_per_second": 88.644,
       "step": 38000
     },
     {
-      "epoch": 21.03602419142782,
-      "grad_norm": 2.1669206619262695,
-      "learning_rate": 9.997e-05,
-      "loss": 4.1409,
+      "epoch": 42.941492216854535,
+      "grad_norm": 3.202237129211426,
+      "learning_rate": 9.996500000000001e-05,
+      "loss": 2.777,
       "step": 40000
     },
     {
-      "epoch": 21.03602419142782,
-      "eval_loss": 4.351284503936768,
-      "eval_runtime": 2.1934,
-      "eval_samples_per_second": 1453.028,
-      "eval_steps_per_second": 91.185,
+      "epoch": 42.941492216854535,
+      "eval_loss": 3.5746841430664062,
+      "eval_runtime": 0.7722,
+      "eval_samples_per_second": 1388.315,
+      "eval_steps_per_second": 86.77,
       "step": 40000
     },
     {
-      "epoch": 22.08782540099921,
-      "eval_loss": 4.321634292602539,
-      "eval_runtime": 2.2273,
-      "eval_samples_per_second": 1430.895,
-      "eval_steps_per_second": 89.796,
+      "epoch": 45.088566827697264,
+      "eval_loss": 3.5878281593322754,
+      "eval_runtime": 0.7639,
+      "eval_samples_per_second": 1403.298,
+      "eval_steps_per_second": 87.706,
       "step": 42000
     },
     {
-      "epoch": 23.139626610570602,
-      "grad_norm": 2.0360896587371826,
-      "learning_rate": 9.335333333333333e-05,
-      "loss": 4.0334,
+      "epoch": 47.235641438539986,
+      "grad_norm": 2.9902379512786865,
+      "learning_rate": 9.336e-05,
+      "loss": 2.6015,
       "step": 44000
     },
     {
-      "epoch": 23.139626610570602,
-      "eval_loss": 4.300807476043701,
-      "eval_runtime": 2.3226,
-      "eval_samples_per_second": 1372.142,
-      "eval_steps_per_second": 86.109,
+      "epoch": 47.235641438539986,
+      "eval_loss": 3.6107263565063477,
+      "eval_runtime": 0.7601,
+      "eval_samples_per_second": 1410.262,
+      "eval_steps_per_second": 88.141,
       "step": 44000
     },
     {
-      "epoch": 24.191427820141993,
-      "eval_loss": 4.279116153717041,
-      "eval_runtime": 2.2177,
-      "eval_samples_per_second": 1437.082,
-      "eval_steps_per_second": 90.184,
+      "epoch": 49.382716049382715,
+      "eval_loss": 3.6260793209075928,
+      "eval_runtime": 0.7605,
+      "eval_samples_per_second": 1409.681,
+      "eval_steps_per_second": 88.105,
       "step": 46000
     },
     {
-      "epoch": 25.243229029713383,
-      "grad_norm": 2.137535572052002,
-      "learning_rate": 8.668833333333334e-05,
-      "loss": 3.9378,
+      "epoch": 51.529790660225444,
+      "grad_norm": 3.1626901626586914,
+      "learning_rate": 8.669833333333334e-05,
+      "loss": 2.4429,
       "step": 48000
     },
     {
-      "epoch": 25.243229029713383,
-      "eval_loss": 4.2718939781188965,
-      "eval_runtime": 2.2019,
-      "eval_samples_per_second": 1447.39,
-      "eval_steps_per_second": 90.831,
+      "epoch": 51.529790660225444,
+      "eval_loss": 3.6414248943328857,
+      "eval_runtime": 0.7596,
+      "eval_samples_per_second": 1411.254,
+      "eval_steps_per_second": 88.203,
       "step": 48000
     },
     {
-      "epoch": 26.295030239284774,
-      "eval_loss": 4.26237678527832,
-      "eval_runtime": 2.2417,
-      "eval_samples_per_second": 1421.698,
-      "eval_steps_per_second": 89.219,
+      "epoch": 53.676865271068166,
+      "eval_loss": 3.6636786460876465,
+      "eval_runtime": 0.7637,
+      "eval_samples_per_second": 1403.705,
+      "eval_steps_per_second": 87.732,
       "step": 50000
     },
     {
-      "epoch": 27.346831448856165,
-      "grad_norm": 2.1131036281585693,
-      "learning_rate": 8.002333333333334e-05,
-      "loss": 3.8595,
+      "epoch": 55.823939881910896,
+      "grad_norm": 3.162830114364624,
+      "learning_rate": 8.003333333333333e-05,
+      "loss": 2.3125,
       "step": 52000
     },
     {
-      "epoch": 27.346831448856165,
-      "eval_loss": 4.253220558166504,
-      "eval_runtime": 2.2045,
-      "eval_samples_per_second": 1445.692,
-      "eval_steps_per_second": 90.724,
+      "epoch": 55.823939881910896,
+      "eval_loss": 3.6777544021606445,
+      "eval_runtime": 0.7597,
+      "eval_samples_per_second": 1411.137,
+      "eval_steps_per_second": 88.196,
       "step": 52000
     },
     {
-      "epoch": 28.39863265842756,
-      "eval_loss": 4.243188381195068,
-      "eval_runtime": 2.1654,
-      "eval_samples_per_second": 1471.751,
-      "eval_steps_per_second": 92.36,
+      "epoch": 57.971014492753625,
+      "eval_loss": 3.703284502029419,
+      "eval_runtime": 0.757,
+      "eval_samples_per_second": 1416.089,
+      "eval_steps_per_second": 88.506,
       "step": 54000
     },
     {
-      "epoch": 29.45043386799895,
-      "grad_norm": 2.1626858711242676,
-      "learning_rate": 7.335833333333335e-05,
-      "loss": 3.7927,
+      "epoch": 60.11808910359635,
+      "grad_norm": 3.4666566848754883,
+      "learning_rate": 7.337000000000001e-05,
+      "loss": 2.1989,
       "step": 56000
     },
     {
-      "epoch": 29.45043386799895,
-      "eval_loss": 4.237761974334717,
-      "eval_runtime": 2.3384,
-      "eval_samples_per_second": 1362.919,
-      "eval_steps_per_second": 85.53,
+      "epoch": 60.11808910359635,
+      "eval_loss": 3.7409818172454834,
+      "eval_runtime": 0.7476,
+      "eval_samples_per_second": 1433.993,
+      "eval_steps_per_second": 89.625,
       "step": 56000
     },
     {
-      "epoch": 30.50223507757034,
-      "eval_loss": 4.237727165222168,
-      "eval_runtime": 2.2527,
-      "eval_samples_per_second": 1414.746,
-      "eval_steps_per_second": 88.782,
+      "epoch": 62.265163714439076,
+      "eval_loss": 3.7754786014556885,
+      "eval_runtime": 0.7506,
+      "eval_samples_per_second": 1428.237,
+      "eval_steps_per_second": 89.265,
       "step": 58000
     },
     {
-      "epoch": 31.55403628714173,
-      "grad_norm": 2.1217854022979736,
-      "learning_rate": 6.669333333333334e-05,
-      "loss": 3.7367,
+      "epoch": 64.4122383252818,
+      "grad_norm": 3.3885388374328613,
+      "learning_rate": 6.670666666666668e-05,
+      "loss": 2.1044,
       "step": 60000
     },
     {
-      "epoch": 31.55403628714173,
-      "eval_loss": 4.233393669128418,
-      "eval_runtime": 2.2874,
-      "eval_samples_per_second": 1393.302,
-      "eval_steps_per_second": 87.437,
+      "epoch": 64.4122383252818,
+      "eval_loss": 3.7875912189483643,
+      "eval_runtime": 0.7543,
+      "eval_samples_per_second": 1421.139,
+      "eval_steps_per_second": 88.821,
       "step": 60000
     },
     {
-      "epoch": 32.60583749671312,
-      "eval_loss": 4.229984283447266,
-      "eval_runtime": 2.3158,
-      "eval_samples_per_second": 1376.171,
-      "eval_steps_per_second": 86.362,
+      "epoch": 66.55931293612453,
+      "eval_loss": 3.8081138134002686,
+      "eval_runtime": 0.7633,
+      "eval_samples_per_second": 1404.466,
+      "eval_steps_per_second": 87.779,
       "step": 62000
     },
     {
-      "epoch": 33.657638706284516,
-      "grad_norm": 2.3173110485076904,
-      "learning_rate": 6.003e-05,
-      "loss": 3.6869,
+      "epoch": 68.70638754696726,
+      "grad_norm": 3.763136148452759,
+      "learning_rate": 6.004333333333334e-05,
+      "loss": 2.0257,
       "step": 64000
     },
     {
-      "epoch": 33.657638706284516,
-      "eval_loss": 4.226880073547363,
-      "eval_runtime": 2.3034,
-      "eval_samples_per_second": 1383.631,
-      "eval_steps_per_second": 86.83,
+      "epoch": 68.70638754696726,
+      "eval_loss": 3.822213649749756,
+      "eval_runtime": 0.7598,
+      "eval_samples_per_second": 1410.912,
+      "eval_steps_per_second": 88.182,
       "step": 64000
     },
     {
-      "epoch": 34.7094399158559,
-      "eval_loss": 4.224526405334473,
-      "eval_runtime": 2.2795,
-      "eval_samples_per_second": 1398.138,
-      "eval_steps_per_second": 87.74,
+      "epoch": 70.85346215780999,
+      "eval_loss": 3.8411173820495605,
+      "eval_runtime": 0.7613,
+      "eval_samples_per_second": 1408.036,
+      "eval_steps_per_second": 88.002,
       "step": 66000
     },
     {
-      "epoch": 35.7612411254273,
-      "grad_norm": 2.3296120166778564,
-      "learning_rate": 5.336333333333333e-05,
-      "loss": 3.6416,
+      "epoch": 73.00053676865271,
+      "grad_norm": 4.504343509674072,
+      "learning_rate": 5.338000000000001e-05,
+      "loss": 1.9563,
       "step": 68000
     },
     {
-      "epoch": 35.7612411254273,
-      "eval_loss": 4.225712299346924,
-      "eval_runtime": 2.2719,
-      "eval_samples_per_second": 1402.762,
-      "eval_steps_per_second": 88.03,
+      "epoch": 73.00053676865271,
+      "eval_loss": 3.848768949508667,
+      "eval_runtime": 0.7688,
+      "eval_samples_per_second": 1394.369,
+      "eval_steps_per_second": 87.148,
       "step": 68000
     },
     {
-      "epoch": 36.813042334998684,
-      "eval_loss": 4.220585346221924,
-      "eval_runtime": 2.2631,
-      "eval_samples_per_second": 1408.243,
-      "eval_steps_per_second": 88.374,
+      "epoch": 75.14761137949544,
+      "eval_loss": 3.8914690017700195,
+      "eval_runtime": 0.7542,
+      "eval_samples_per_second": 1421.304,
+      "eval_steps_per_second": 88.831,
       "step": 70000
     },
     {
-      "epoch": 37.86484354457008,
-      "grad_norm": 2.309408187866211,
-      "learning_rate": 4.669833333333333e-05,
-      "loss": 3.6032,
+      "epoch": 77.29468599033817,
+      "grad_norm": 3.6598613262176514,
+      "learning_rate": 4.671666666666667e-05,
+      "loss": 1.8905,
       "step": 72000
     },
     {
-      "epoch": 37.86484354457008,
-      "eval_loss": 4.224654674530029,
-      "eval_runtime": 2.272,
-      "eval_samples_per_second": 1402.745,
-      "eval_steps_per_second": 88.029,
+      "epoch": 77.29468599033817,
+      "eval_loss": 3.9079344272613525,
+      "eval_runtime": 0.7509,
+      "eval_samples_per_second": 1427.665,
+      "eval_steps_per_second": 89.229,
       "step": 72000
     },
     {
-      "epoch": 38.916644754141466,
-      "eval_loss": 4.227168083190918,
-      "eval_runtime": 2.2483,
-      "eval_samples_per_second": 1417.498,
-      "eval_steps_per_second": 88.955,
+      "epoch": 79.44176060118089,
+      "eval_loss": 3.9169275760650635,
+      "eval_runtime": 0.7546,
+      "eval_samples_per_second": 1420.536,
+      "eval_steps_per_second": 88.783,
       "step": 74000
     },
     {
-      "epoch": 39.96844596371286,
-      "grad_norm": 2.425579309463501,
-      "learning_rate": 4.0033333333333335e-05,
-      "loss": 3.5669,
+      "epoch": 81.58883521202361,
+      "grad_norm": 3.5242605209350586,
+      "learning_rate": 4.005333333333334e-05,
+      "loss": 1.836,
       "step": 76000
     },
     {
-      "epoch": 39.96844596371286,
-      "eval_loss": 4.226228713989258,
-      "eval_runtime": 2.2145,
-      "eval_samples_per_second": 1439.153,
-      "eval_steps_per_second": 90.314,
+      "epoch": 81.58883521202361,
+      "eval_loss": 3.9381563663482666,
+      "eval_runtime": 0.7535,
+      "eval_samples_per_second": 1422.783,
+      "eval_steps_per_second": 88.924,
       "step": 76000
     },
     {
-      "epoch": 41.02024717328425,
-      "eval_loss": 4.230894565582275,
-      "eval_runtime": 2.3349,
-      "eval_samples_per_second": 1364.932,
-      "eval_steps_per_second": 85.656,
+      "epoch": 83.73590982286635,
+      "eval_loss": 3.9429502487182617,
+      "eval_runtime": 0.7517,
+      "eval_samples_per_second": 1426.084,
+      "eval_steps_per_second": 89.13,
       "step": 78000
     },
     {
-      "epoch": 42.07204838285564,
-      "grad_norm": 2.4452333450317383,
-      "learning_rate": 3.337e-05,
-      "loss": 3.533,
+      "epoch": 85.88298443370907,
+      "grad_norm": 3.842487096786499,
+      "learning_rate": 3.339e-05,
+      "loss": 1.7885,
       "step": 80000
     },
     {
-      "epoch": 42.07204838285564,
-      "eval_loss": 4.230439186096191,
-      "eval_runtime": 2.2462,
-      "eval_samples_per_second": 1418.848,
-      "eval_steps_per_second": 89.04,
+      "epoch": 85.88298443370907,
+      "eval_loss": 3.9470765590667725,
+      "eval_runtime": 0.7546,
+      "eval_samples_per_second": 1420.557,
+      "eval_steps_per_second": 88.785,
       "step": 80000
     },
     {
-      "epoch": 43.12384959242703,
-      "eval_loss": 4.2289557456970215,
-      "eval_runtime": 2.404,
-      "eval_samples_per_second": 1325.692,
-      "eval_steps_per_second": 83.194,
+      "epoch": 88.03005904455179,
+      "eval_loss": 3.9667794704437256,
+      "eval_runtime": 0.7733,
+      "eval_samples_per_second": 1386.292,
+      "eval_steps_per_second": 86.643,
       "step": 82000
     },
     {
-      "epoch": 44.17565080199842,
-      "grad_norm": 2.6046276092529297,
-      "learning_rate": 2.6705000000000003e-05,
-      "loss": 3.5033,
+      "epoch": 90.17713365539453,
+      "grad_norm": 3.9361636638641357,
+      "learning_rate": 2.6726666666666666e-05,
+      "loss": 1.7431,
       "step": 84000
     },
     {
-      "epoch": 44.17565080199842,
-      "eval_loss": 4.233131408691406,
-      "eval_runtime": 2.3346,
-      "eval_samples_per_second": 1365.098,
-      "eval_steps_per_second": 85.667,
+      "epoch": 90.17713365539453,
+      "eval_loss": 3.986008882522583,
+      "eval_runtime": 0.7467,
+      "eval_samples_per_second": 1435.654,
+      "eval_steps_per_second": 89.728,
       "step": 84000
     },
     {
-      "epoch": 45.22745201156982,
-      "eval_loss": 4.232224941253662,
-      "eval_runtime": 2.2625,
-      "eval_samples_per_second": 1408.62,
-      "eval_steps_per_second": 88.398,
+      "epoch": 92.32420826623725,
+      "eval_loss": 4.008753299713135,
+      "eval_runtime": 0.7576,
+      "eval_samples_per_second": 1415.003,
+      "eval_steps_per_second": 88.438,
       "step": 86000
     },
     {
-      "epoch": 46.279253221141204,
-      "grad_norm": 2.675853967666626,
-      "learning_rate": 2.004e-05,
-      "loss": 3.4759,
+      "epoch": 94.47128287707997,
+      "grad_norm": 4.01881742477417,
+      "learning_rate": 2.0063333333333335e-05,
+      "loss": 1.7024,
       "step": 88000
     },
     {
-      "epoch": 46.279253221141204,
-      "eval_loss": 4.231053352355957,
-      "eval_runtime": 2.273,
-      "eval_samples_per_second": 1402.13,
-      "eval_steps_per_second": 87.991,
+      "epoch": 94.47128287707997,
+      "eval_loss": 4.013205528259277,
+      "eval_runtime": 0.7488,
+      "eval_samples_per_second": 1431.63,
+      "eval_steps_per_second": 89.477,
       "step": 88000
     },
     {
-      "epoch": 47.3310544307126,
-      "eval_loss": 4.234147071838379,
-      "eval_runtime": 2.4703,
-      "eval_samples_per_second": 1290.135,
-      "eval_steps_per_second": 80.962,
+      "epoch": 96.61835748792271,
+      "eval_loss": 4.0260210037231445,
+      "eval_runtime": 0.7603,
+      "eval_samples_per_second": 1410.015,
+      "eval_steps_per_second": 88.126,
       "step": 90000
     },
     {
-      "epoch": 48.382855640283985,
-      "grad_norm": 2.581390142440796,
-      "learning_rate": 1.3375000000000002e-05,
-      "loss": 3.4528,
+      "epoch": 98.76543209876543,
+      "grad_norm": 3.9451382160186768,
+      "learning_rate": 1.3400000000000002e-05,
+      "loss": 1.6687,
       "step": 92000
     },
     {
-      "epoch": 48.382855640283985,
-      "eval_loss": 4.23477840423584,
-      "eval_runtime": 2.2586,
-      "eval_samples_per_second": 1411.035,
-      "eval_steps_per_second": 88.549,
+      "epoch": 98.76543209876543,
+      "eval_loss": 4.035834789276123,
+      "eval_runtime": 0.7619,
+      "eval_samples_per_second": 1406.935,
+      "eval_steps_per_second": 87.933,
       "step": 92000
     },
     {
-      "epoch": 49.43465684985538,
-      "eval_loss": 4.232587814331055,
-      "eval_runtime": 2.2225,
-      "eval_samples_per_second": 1433.971,
-      "eval_steps_per_second": 89.989,
+      "epoch": 100.91250670960815,
+      "eval_loss": 4.029040813446045,
+      "eval_runtime": 0.7506,
+      "eval_samples_per_second": 1428.162,
+      "eval_steps_per_second": 89.26,
       "step": 94000
     },
     {
-      "epoch": 50.48645805942677,
-      "grad_norm": 2.7233216762542725,
-      "learning_rate": 6.710000000000001e-06,
-      "loss": 3.4319,
+      "epoch": 103.05958132045089,
+      "grad_norm": 4.011170387268066,
+      "learning_rate": 6.736666666666668e-06,
+      "loss": 1.6369,
       "step": 96000
     },
     {
-      "epoch": 50.48645805942677,
-      "eval_loss": 4.2330780029296875,
-      "eval_runtime": 2.2298,
-      "eval_samples_per_second": 1429.262,
-      "eval_steps_per_second": 89.693,
+      "epoch": 103.05958132045089,
+      "eval_loss": 4.04220724105835,
+      "eval_runtime": 0.7565,
+      "eval_samples_per_second": 1417.059,
+      "eval_steps_per_second": 88.566,
       "step": 96000
     },
     {
-      "epoch": 51.53825926899816,
-      "eval_loss": 4.232585906982422,
-      "eval_runtime": 2.9048,
-      "eval_samples_per_second": 1097.167,
-      "eval_steps_per_second": 68.853,
+      "epoch": 105.20665593129361,
+      "eval_loss": 4.044515609741211,
+      "eval_runtime": 0.7518,
+      "eval_samples_per_second": 1425.874,
+      "eval_steps_per_second": 89.117,
       "step": 98000
     },
     {
-      "epoch": 52.59006047856955,
-      "grad_norm": 2.644059181213379,
-      "learning_rate": 4.5e-08,
-      "loss": 3.4136,
+      "epoch": 107.35373054213633,
+      "grad_norm": 3.8686037063598633,
+      "learning_rate": 7.333333333333334e-08,
+      "loss": 1.6129,
       "step": 100000
     },
     {
-      "epoch": 52.59006047856955,
-      "eval_loss": 4.232517242431641,
-      "eval_runtime": 2.2133,
-      "eval_samples_per_second": 1439.961,
-      "eval_steps_per_second": 90.365,
+      "epoch": 107.35373054213633,
+      "eval_loss": 4.045902252197266,
+      "eval_runtime": 0.7523,
+      "eval_samples_per_second": 1424.957,
+      "eval_steps_per_second": 89.06,
       "step": 100000
     }
   ],
   "logging_steps": 4000,
   "max_steps": 100000,
   "num_input_tokens_seen": 0,
-  "num_train_epochs": 53,
+  "num_train_epochs": 108,
   "save_steps": 4000,
   "stateful_callbacks": {
     "TrainerControl": {
@@ -601,7 +601,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 2.5850315930440704e+16,
+  "total_flos": 2.5849709221380096e+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:ab68434ec7156d3a63e4783f824871a48d62c7c2fdcc831fdd5471c5f4aec7fe
+oid sha256:26102d0ac750ce531ebda9d7577ccea6c8ecbc93b04bdff23226e376dd5609a2
 size 5368