| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.9539427243405263, | |
| "global_step": 68000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.043440014769605025, | |
| "learning_rate": 4.9275991891109183e-05, | |
| "loss": 3.7321220703125, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.08688002953921005, | |
| "learning_rate": 4.8551983782218364e-05, | |
| "loss": 3.381878173828125, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.13032004430881505, | |
| "learning_rate": 4.782797567332754e-05, | |
| "loss": 3.26605712890625, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.1737600590784201, | |
| "learning_rate": 4.7103967564436726e-05, | |
| "loss": 3.197267333984375, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.21720007384802512, | |
| "learning_rate": 4.63799594555459e-05, | |
| "loss": 3.14855224609375, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.21720007384802512, | |
| "eval_loss": 3.0360004901885986, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.2606400886176301, | |
| "learning_rate": 4.565595134665508e-05, | |
| "loss": 3.114377685546875, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.3040801033872352, | |
| "learning_rate": 4.493194323776426e-05, | |
| "loss": 3.08016552734375, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.3475201181568402, | |
| "learning_rate": 4.4207935128873444e-05, | |
| "loss": 3.05526904296875, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.3909601329264452, | |
| "learning_rate": 4.3483927019982625e-05, | |
| "loss": 3.033037353515625, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.43440014769605023, | |
| "learning_rate": 4.2759918911091806e-05, | |
| "loss": 3.012407958984375, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.43440014769605023, | |
| "eval_loss": 2.9199917316436768, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.47784016246565525, | |
| "learning_rate": 4.203591080220099e-05, | |
| "loss": 2.992706298828125, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.5212801772352602, | |
| "learning_rate": 4.131190269331017e-05, | |
| "loss": 2.975831298828125, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.5647201920048652, | |
| "learning_rate": 4.058789458441934e-05, | |
| "loss": 2.96051806640625, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.6081602067744704, | |
| "learning_rate": 3.986388647552853e-05, | |
| "loss": 2.9437431640625, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.6516002215440754, | |
| "learning_rate": 3.9139878366637704e-05, | |
| "loss": 2.9307431640625, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.6516002215440754, | |
| "eval_loss": 2.8523595333099365, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.6950402363136804, | |
| "learning_rate": 3.841587025774689e-05, | |
| "loss": 2.916943359375, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.7384802510832854, | |
| "learning_rate": 3.7691862148856066e-05, | |
| "loss": 2.905533447265625, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.7819202658528904, | |
| "learning_rate": 3.696785403996525e-05, | |
| "loss": 2.892618896484375, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.8253602806224954, | |
| "learning_rate": 3.624384593107443e-05, | |
| "loss": 2.885263671875, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.8688002953921005, | |
| "learning_rate": 3.551983782218361e-05, | |
| "loss": 2.873368896484375, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.8688002953921005, | |
| "eval_loss": 2.808135986328125, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.9122403101617055, | |
| "learning_rate": 3.479582971329279e-05, | |
| "loss": 2.860178955078125, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 0.9556803249313105, | |
| "learning_rate": 3.407182160440197e-05, | |
| "loss": 2.853407958984375, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 0.9991203397009155, | |
| "learning_rate": 3.334781349551115e-05, | |
| "loss": 2.846532470703125, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 1.042571214474213, | |
| "learning_rate": 3.262380538662033e-05, | |
| "loss": 2.816125244140625, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 1.086011229243818, | |
| "learning_rate": 3.189979727772951e-05, | |
| "loss": 2.8096533203125, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 1.086011229243818, | |
| "eval_loss": 2.7750799655914307, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 1.129451244013423, | |
| "learning_rate": 3.1175789168838695e-05, | |
| "loss": 2.803942626953125, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 1.172891258783028, | |
| "learning_rate": 3.045178105994787e-05, | |
| "loss": 2.798123046875, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 1.216331273552633, | |
| "learning_rate": 2.9727772951057054e-05, | |
| "loss": 2.79141748046875, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 1.259771288322238, | |
| "learning_rate": 2.900376484216623e-05, | |
| "loss": 2.784273193359375, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 1.303211303091843, | |
| "learning_rate": 2.8279756733275416e-05, | |
| "loss": 2.777781494140625, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 1.303211303091843, | |
| "eval_loss": 2.746640205383301, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 1.346651317861448, | |
| "learning_rate": 2.7555748624384593e-05, | |
| "loss": 2.774093994140625, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 1.3900913326310531, | |
| "learning_rate": 2.6831740515493774e-05, | |
| "loss": 2.765812744140625, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 1.4335313474006581, | |
| "learning_rate": 2.6107732406602952e-05, | |
| "loss": 2.763001708984375, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 1.4769713621702631, | |
| "learning_rate": 2.5383724297712136e-05, | |
| "loss": 2.756827880859375, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 1.5204113769398682, | |
| "learning_rate": 2.4659716188821317e-05, | |
| "loss": 2.753128173828125, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 1.5204113769398682, | |
| "eval_loss": 2.725342035293579, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 1.5638513917094732, | |
| "learning_rate": 2.3935708079930498e-05, | |
| "loss": 2.7459052734375, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 1.6072914064790782, | |
| "learning_rate": 2.3211699971039676e-05, | |
| "loss": 2.738887451171875, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 1.6507314212486832, | |
| "learning_rate": 2.2487691862148857e-05, | |
| "loss": 2.738287109375, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 1.6941714360182882, | |
| "learning_rate": 2.1763683753258038e-05, | |
| "loss": 2.731450439453125, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 1.7376114507878933, | |
| "learning_rate": 2.103967564436722e-05, | |
| "loss": 2.726150390625, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 1.7376114507878933, | |
| "eval_loss": 2.706486225128174, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 1.7810514655574983, | |
| "learning_rate": 2.03156675354764e-05, | |
| "loss": 2.72658447265625, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 1.8244914803271033, | |
| "learning_rate": 1.9591659426585578e-05, | |
| "loss": 2.723203125, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 1.8679314950967083, | |
| "learning_rate": 1.886765131769476e-05, | |
| "loss": 2.715607177734375, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 1.9113715098663133, | |
| "learning_rate": 1.814364320880394e-05, | |
| "loss": 2.7112373046875, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 1.9548115246359183, | |
| "learning_rate": 1.741963509991312e-05, | |
| "loss": 2.7094580078125, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 1.9548115246359183, | |
| "eval_loss": 2.690136432647705, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 1.9982515394055234, | |
| "learning_rate": 1.66956269910223e-05, | |
| "loss": 2.70530859375, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 2.041702414178821, | |
| "learning_rate": 1.5971618882131483e-05, | |
| "loss": 2.68336376953125, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 2.085142428948426, | |
| "learning_rate": 1.5247610773240662e-05, | |
| "loss": 2.68182470703125, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 2.128582443718031, | |
| "learning_rate": 1.4523602664349841e-05, | |
| "loss": 2.67953466796875, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 2.172022458487636, | |
| "learning_rate": 1.3799594555459022e-05, | |
| "loss": 2.67869482421875, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 2.172022458487636, | |
| "eval_loss": 2.679882287979126, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 2.215462473257241, | |
| "learning_rate": 1.3075586446568203e-05, | |
| "loss": 2.67307861328125, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 2.258902488026846, | |
| "learning_rate": 1.2351578337677383e-05, | |
| "loss": 2.6713193359375, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 2.302342502796451, | |
| "learning_rate": 1.1627570228786564e-05, | |
| "loss": 2.666735107421875, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 2.345782517566056, | |
| "learning_rate": 1.0903562119895745e-05, | |
| "loss": 2.665475341796875, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 2.389222532335661, | |
| "learning_rate": 1.0179554011004924e-05, | |
| "loss": 2.665841796875, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 2.389222532335661, | |
| "eval_loss": 2.66998553276062, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 2.432662547105266, | |
| "learning_rate": 9.455545902114105e-06, | |
| "loss": 2.66330908203125, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 2.476102561874871, | |
| "learning_rate": 8.731537793223284e-06, | |
| "loss": 2.66229638671875, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 2.519542576644476, | |
| "learning_rate": 8.007529684332465e-06, | |
| "loss": 2.65796484375, | |
| "step": 58000 | |
| }, | |
| { | |
| "epoch": 2.562982591414081, | |
| "learning_rate": 7.283521575441646e-06, | |
| "loss": 2.657759033203125, | |
| "step": 59000 | |
| }, | |
| { | |
| "epoch": 2.606422606183686, | |
| "learning_rate": 6.559513466550826e-06, | |
| "loss": 2.65596875, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 2.606422606183686, | |
| "eval_loss": 2.6612513065338135, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 2.649862620953291, | |
| "learning_rate": 5.835505357660006e-06, | |
| "loss": 2.656216064453125, | |
| "step": 61000 | |
| }, | |
| { | |
| "epoch": 2.693302635722896, | |
| "learning_rate": 5.111497248769187e-06, | |
| "loss": 2.654870361328125, | |
| "step": 62000 | |
| }, | |
| { | |
| "epoch": 2.736742650492501, | |
| "learning_rate": 4.387489139878367e-06, | |
| "loss": 2.653074951171875, | |
| "step": 63000 | |
| }, | |
| { | |
| "epoch": 2.7801826652621062, | |
| "learning_rate": 3.663481030987547e-06, | |
| "loss": 2.64907958984375, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 2.8236226800317112, | |
| "learning_rate": 2.9394729220967276e-06, | |
| "loss": 2.650592529296875, | |
| "step": 65000 | |
| }, | |
| { | |
| "epoch": 2.8236226800317112, | |
| "eval_loss": 2.655712127685547, | |
| "step": 65000 | |
| }, | |
| { | |
| "epoch": 2.8670626948013163, | |
| "learning_rate": 2.2154648132059077e-06, | |
| "loss": 2.651447998046875, | |
| "step": 66000 | |
| }, | |
| { | |
| "epoch": 2.9105027095709213, | |
| "learning_rate": 1.4914567043150885e-06, | |
| "loss": 2.647968994140625, | |
| "step": 67000 | |
| }, | |
| { | |
| "epoch": 2.9539427243405263, | |
| "learning_rate": 7.674485954242688e-07, | |
| "loss": 2.647386962890625, | |
| "step": 68000 | |
| } | |
| ], | |
| "max_steps": 69060, | |
| "num_train_epochs": 3, | |
| "total_flos": 4353523817019612480, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |