| { |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 9.979036827195468, |
| "eval_steps": 500, |
| "global_step": 4410, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.11331444759206799, |
| "grad_norm": 0.2005129361523877, |
| "learning_rate": 1.977324263038549e-05, |
| "loss": 0.4547, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.22662889518413598, |
| "grad_norm": 4.374481882033845, |
| "learning_rate": 1.9546485260770977e-05, |
| "loss": 0.225, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.33994334277620397, |
| "grad_norm": 0.1016014971343205, |
| "learning_rate": 1.9319727891156463e-05, |
| "loss": 0.2215, |
| "step": 150 |
| }, |
| { |
| "epoch": 0.45325779036827196, |
| "grad_norm": 0.06295163563225963, |
| "learning_rate": 1.9092970521541953e-05, |
| "loss": 0.2203, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.56657223796034, |
| "grad_norm": 0.06293734713169016, |
| "learning_rate": 1.886621315192744e-05, |
| "loss": 0.2195, |
| "step": 250 |
| }, |
| { |
| "epoch": 0.6798866855524079, |
| "grad_norm": 0.060515101336106644, |
| "learning_rate": 1.863945578231293e-05, |
| "loss": 0.22, |
| "step": 300 |
| }, |
| { |
| "epoch": 0.7932011331444759, |
| "grad_norm": 0.07540683199497876, |
| "learning_rate": 1.8412698412698415e-05, |
| "loss": 0.2196, |
| "step": 350 |
| }, |
| { |
| "epoch": 0.9065155807365439, |
| "grad_norm": 0.08983383050228771, |
| "learning_rate": 1.81859410430839e-05, |
| "loss": 0.2193, |
| "step": 400 |
| }, |
| { |
| "epoch": 1.0181303116147309, |
| "grad_norm": 0.0675672563417117, |
| "learning_rate": 1.795918367346939e-05, |
| "loss": 0.2156, |
| "step": 450 |
| }, |
| { |
| "epoch": 1.1314447592067989, |
| "grad_norm": 0.07024561390451645, |
| "learning_rate": 1.7732426303854877e-05, |
| "loss": 0.2186, |
| "step": 500 |
| }, |
| { |
| "epoch": 1.2447592067988669, |
| "grad_norm": 0.07966954244619774, |
| "learning_rate": 1.7505668934240366e-05, |
| "loss": 0.2184, |
| "step": 550 |
| }, |
| { |
| "epoch": 1.3580736543909349, |
| "grad_norm": 0.04818417686134014, |
| "learning_rate": 1.7278911564625852e-05, |
| "loss": 0.2182, |
| "step": 600 |
| }, |
| { |
| "epoch": 1.4713881019830028, |
| "grad_norm": 0.05242535702602152, |
| "learning_rate": 1.705215419501134e-05, |
| "loss": 0.218, |
| "step": 650 |
| }, |
| { |
| "epoch": 1.5847025495750708, |
| "grad_norm": 0.06651711822208371, |
| "learning_rate": 1.6825396825396828e-05, |
| "loss": 0.2179, |
| "step": 700 |
| }, |
| { |
| "epoch": 1.6980169971671388, |
| "grad_norm": 0.19794356005762634, |
| "learning_rate": 1.6598639455782314e-05, |
| "loss": 0.2178, |
| "step": 750 |
| }, |
| { |
| "epoch": 1.8113314447592068, |
| "grad_norm": 0.054498240100240465, |
| "learning_rate": 1.63718820861678e-05, |
| "loss": 0.2182, |
| "step": 800 |
| }, |
| { |
| "epoch": 1.9246458923512748, |
| "grad_norm": 0.053181868189447694, |
| "learning_rate": 1.614512471655329e-05, |
| "loss": 0.2177, |
| "step": 850 |
| }, |
| { |
| "epoch": 2.0362606232294618, |
| "grad_norm": 0.0562999012232746, |
| "learning_rate": 1.5918367346938776e-05, |
| "loss": 0.2142, |
| "step": 900 |
| }, |
| { |
| "epoch": 2.1495750708215295, |
| "grad_norm": 0.06574460863314742, |
| "learning_rate": 1.5691609977324265e-05, |
| "loss": 0.2174, |
| "step": 950 |
| }, |
| { |
| "epoch": 2.2628895184135978, |
| "grad_norm": 0.07473982858505401, |
| "learning_rate": 1.546485260770975e-05, |
| "loss": 0.2174, |
| "step": 1000 |
| }, |
| { |
| "epoch": 2.376203966005666, |
| "grad_norm": 0.04914272343711847, |
| "learning_rate": 1.523809523809524e-05, |
| "loss": 0.2172, |
| "step": 1050 |
| }, |
| { |
| "epoch": 2.4895184135977337, |
| "grad_norm": 0.060028789070772234, |
| "learning_rate": 1.5011337868480727e-05, |
| "loss": 0.2172, |
| "step": 1100 |
| }, |
| { |
| "epoch": 2.6028328611898015, |
| "grad_norm": 0.04603015974224043, |
| "learning_rate": 1.4784580498866215e-05, |
| "loss": 0.2172, |
| "step": 1150 |
| }, |
| { |
| "epoch": 2.7161473087818697, |
| "grad_norm": 0.05801340941717548, |
| "learning_rate": 1.4557823129251703e-05, |
| "loss": 0.217, |
| "step": 1200 |
| }, |
| { |
| "epoch": 2.829461756373938, |
| "grad_norm": 0.03757088207951635, |
| "learning_rate": 1.433106575963719e-05, |
| "loss": 0.2169, |
| "step": 1250 |
| }, |
| { |
| "epoch": 2.9427762039660057, |
| "grad_norm": 0.0392731960266638, |
| "learning_rate": 1.4104308390022677e-05, |
| "loss": 0.2169, |
| "step": 1300 |
| }, |
| { |
| "epoch": 3.0543909348441924, |
| "grad_norm": 0.06241785187409858, |
| "learning_rate": 1.3877551020408165e-05, |
| "loss": 0.2136, |
| "step": 1350 |
| }, |
| { |
| "epoch": 3.1677053824362607, |
| "grad_norm": 0.06071070650139286, |
| "learning_rate": 1.3650793650793652e-05, |
| "loss": 0.2186, |
| "step": 1400 |
| }, |
| { |
| "epoch": 3.2810198300283284, |
| "grad_norm": 0.0408584442690185, |
| "learning_rate": 1.342403628117914e-05, |
| "loss": 0.2171, |
| "step": 1450 |
| }, |
| { |
| "epoch": 3.3943342776203966, |
| "grad_norm": 0.04286108941316469, |
| "learning_rate": 1.3197278911564626e-05, |
| "loss": 0.217, |
| "step": 1500 |
| }, |
| { |
| "epoch": 3.507648725212465, |
| "grad_norm": 0.050377281789735655, |
| "learning_rate": 1.2970521541950114e-05, |
| "loss": 0.2195, |
| "step": 1550 |
| }, |
| { |
| "epoch": 3.6209631728045326, |
| "grad_norm": 0.05622603657613179, |
| "learning_rate": 1.2743764172335602e-05, |
| "loss": 0.2174, |
| "step": 1600 |
| }, |
| { |
| "epoch": 3.7342776203966004, |
| "grad_norm": 0.048485290627651875, |
| "learning_rate": 1.251700680272109e-05, |
| "loss": 0.217, |
| "step": 1650 |
| }, |
| { |
| "epoch": 3.8475920679886686, |
| "grad_norm": 0.049896749884447776, |
| "learning_rate": 1.2290249433106578e-05, |
| "loss": 0.2169, |
| "step": 1700 |
| }, |
| { |
| "epoch": 3.960906515580737, |
| "grad_norm": 0.042955138530708765, |
| "learning_rate": 1.2063492063492064e-05, |
| "loss": 0.2168, |
| "step": 1750 |
| }, |
| { |
| "epoch": 4.0725212464589235, |
| "grad_norm": 0.04972531513985689, |
| "learning_rate": 1.1836734693877552e-05, |
| "loss": 0.2133, |
| "step": 1800 |
| }, |
| { |
| "epoch": 4.185835694050992, |
| "grad_norm": 0.03493608076161152, |
| "learning_rate": 1.160997732426304e-05, |
| "loss": 0.2165, |
| "step": 1850 |
| }, |
| { |
| "epoch": 4.299150141643059, |
| "grad_norm": 0.05936933051159733, |
| "learning_rate": 1.1383219954648527e-05, |
| "loss": 0.2166, |
| "step": 1900 |
| }, |
| { |
| "epoch": 4.412464589235127, |
| "grad_norm": 0.044813050936368184, |
| "learning_rate": 1.1156462585034013e-05, |
| "loss": 0.2165, |
| "step": 1950 |
| }, |
| { |
| "epoch": 4.5257790368271955, |
| "grad_norm": 0.039668826323639036, |
| "learning_rate": 1.0929705215419501e-05, |
| "loss": 0.2165, |
| "step": 2000 |
| }, |
| { |
| "epoch": 4.639093484419264, |
| "grad_norm": 0.04978591341331415, |
| "learning_rate": 1.0702947845804989e-05, |
| "loss": 0.2164, |
| "step": 2050 |
| }, |
| { |
| "epoch": 4.752407932011332, |
| "grad_norm": 0.04572653204373289, |
| "learning_rate": 1.0476190476190477e-05, |
| "loss": 0.2162, |
| "step": 2100 |
| }, |
| { |
| "epoch": 4.865722379603399, |
| "grad_norm": 0.0414740460984779, |
| "learning_rate": 1.0249433106575966e-05, |
| "loss": 0.216, |
| "step": 2150 |
| }, |
| { |
| "epoch": 4.9790368271954675, |
| "grad_norm": 0.044261316481047364, |
| "learning_rate": 1.0022675736961451e-05, |
| "loss": 0.216, |
| "step": 2200 |
| }, |
| { |
| "epoch": 5.090651558073654, |
| "grad_norm": 0.039290118974741946, |
| "learning_rate": 9.795918367346939e-06, |
| "loss": 0.2126, |
| "step": 2250 |
| }, |
| { |
| "epoch": 5.203966005665722, |
| "grad_norm": 0.0558689696926692, |
| "learning_rate": 9.569160997732427e-06, |
| "loss": 0.2158, |
| "step": 2300 |
| }, |
| { |
| "epoch": 5.317280453257791, |
| "grad_norm": 0.04013222575202666, |
| "learning_rate": 9.342403628117914e-06, |
| "loss": 0.2158, |
| "step": 2350 |
| }, |
| { |
| "epoch": 5.430594900849858, |
| "grad_norm": 0.03940518538256405, |
| "learning_rate": 9.115646258503402e-06, |
| "loss": 0.2156, |
| "step": 2400 |
| }, |
| { |
| "epoch": 5.543909348441926, |
| "grad_norm": 0.036248333651368, |
| "learning_rate": 8.888888888888888e-06, |
| "loss": 0.2156, |
| "step": 2450 |
| }, |
| { |
| "epoch": 5.657223796033994, |
| "grad_norm": 0.030178531163303088, |
| "learning_rate": 8.662131519274378e-06, |
| "loss": 0.2155, |
| "step": 2500 |
| }, |
| { |
| "epoch": 5.770538243626063, |
| "grad_norm": 0.040097656925639, |
| "learning_rate": 8.435374149659866e-06, |
| "loss": 0.2154, |
| "step": 2550 |
| }, |
| { |
| "epoch": 5.88385269121813, |
| "grad_norm": 0.05225991360322306, |
| "learning_rate": 8.208616780045352e-06, |
| "loss": 0.2154, |
| "step": 2600 |
| }, |
| { |
| "epoch": 5.997167138810198, |
| "grad_norm": 0.03410100715408048, |
| "learning_rate": 7.98185941043084e-06, |
| "loss": 0.2154, |
| "step": 2650 |
| }, |
| { |
| "epoch": 6.108781869688385, |
| "grad_norm": 0.04267307174763642, |
| "learning_rate": 7.755102040816327e-06, |
| "loss": 0.2119, |
| "step": 2700 |
| }, |
| { |
| "epoch": 6.222096317280453, |
| "grad_norm": 0.04309446113662568, |
| "learning_rate": 7.528344671201815e-06, |
| "loss": 0.2152, |
| "step": 2750 |
| }, |
| { |
| "epoch": 6.335410764872521, |
| "grad_norm": 0.041102870476639485, |
| "learning_rate": 7.301587301587301e-06, |
| "loss": 0.2151, |
| "step": 2800 |
| }, |
| { |
| "epoch": 6.4487252124645895, |
| "grad_norm": 0.04551287987357368, |
| "learning_rate": 7.07482993197279e-06, |
| "loss": 0.2151, |
| "step": 2850 |
| }, |
| { |
| "epoch": 6.562039660056657, |
| "grad_norm": 0.04323820835646204, |
| "learning_rate": 6.848072562358277e-06, |
| "loss": 0.2149, |
| "step": 2900 |
| }, |
| { |
| "epoch": 6.675354107648725, |
| "grad_norm": 0.05251190173310129, |
| "learning_rate": 6.621315192743765e-06, |
| "loss": 0.2149, |
| "step": 2950 |
| }, |
| { |
| "epoch": 6.788668555240793, |
| "grad_norm": 0.07880028642512263, |
| "learning_rate": 6.394557823129253e-06, |
| "loss": 0.2149, |
| "step": 3000 |
| }, |
| { |
| "epoch": 6.9019830028328615, |
| "grad_norm": 0.03479910796542844, |
| "learning_rate": 6.16780045351474e-06, |
| "loss": 0.2148, |
| "step": 3050 |
| }, |
| { |
| "epoch": 7.013597733711048, |
| "grad_norm": 0.03458517402957525, |
| "learning_rate": 5.9410430839002275e-06, |
| "loss": 0.2115, |
| "step": 3100 |
| }, |
| { |
| "epoch": 7.126912181303116, |
| "grad_norm": 0.034419006405346834, |
| "learning_rate": 5.7142857142857145e-06, |
| "loss": 0.2145, |
| "step": 3150 |
| }, |
| { |
| "epoch": 7.240226628895184, |
| "grad_norm": 0.03629297170824396, |
| "learning_rate": 5.487528344671202e-06, |
| "loss": 0.2144, |
| "step": 3200 |
| }, |
| { |
| "epoch": 7.353541076487252, |
| "grad_norm": 0.03858771713025695, |
| "learning_rate": 5.260770975056689e-06, |
| "loss": 0.2143, |
| "step": 3250 |
| }, |
| { |
| "epoch": 7.46685552407932, |
| "grad_norm": 0.0350919228123147, |
| "learning_rate": 5.034013605442177e-06, |
| "loss": 0.2143, |
| "step": 3300 |
| }, |
| { |
| "epoch": 7.580169971671388, |
| "grad_norm": 0.048002057370294546, |
| "learning_rate": 4.807256235827665e-06, |
| "loss": 0.2142, |
| "step": 3350 |
| }, |
| { |
| "epoch": 7.693484419263456, |
| "grad_norm": 0.03299980225000233, |
| "learning_rate": 4.580498866213152e-06, |
| "loss": 0.2142, |
| "step": 3400 |
| }, |
| { |
| "epoch": 7.806798866855524, |
| "grad_norm": 0.0366304286363175, |
| "learning_rate": 4.35374149659864e-06, |
| "loss": 0.2141, |
| "step": 3450 |
| }, |
| { |
| "epoch": 7.920113314447592, |
| "grad_norm": 0.039492638736222725, |
| "learning_rate": 4.126984126984127e-06, |
| "loss": 0.2141, |
| "step": 3500 |
| }, |
| { |
| "epoch": 8.03172804532578, |
| "grad_norm": 0.052547554783668696, |
| "learning_rate": 3.9002267573696154e-06, |
| "loss": 0.2108, |
| "step": 3550 |
| }, |
| { |
| "epoch": 8.145042492917847, |
| "grad_norm": 0.052357108980609274, |
| "learning_rate": 3.6734693877551024e-06, |
| "loss": 0.2138, |
| "step": 3600 |
| }, |
| { |
| "epoch": 8.258356940509914, |
| "grad_norm": 0.04203804316106579, |
| "learning_rate": 3.44671201814059e-06, |
| "loss": 0.2137, |
| "step": 3650 |
| }, |
| { |
| "epoch": 8.371671388101984, |
| "grad_norm": 0.03800793699581388, |
| "learning_rate": 3.2199546485260772e-06, |
| "loss": 0.2138, |
| "step": 3700 |
| }, |
| { |
| "epoch": 8.48498583569405, |
| "grad_norm": 0.03338231105442374, |
| "learning_rate": 2.993197278911565e-06, |
| "loss": 0.2138, |
| "step": 3750 |
| }, |
| { |
| "epoch": 8.598300283286118, |
| "grad_norm": 0.03462642479987778, |
| "learning_rate": 2.7664399092970525e-06, |
| "loss": 0.2136, |
| "step": 3800 |
| }, |
| { |
| "epoch": 8.711614730878187, |
| "grad_norm": 0.03345477511212533, |
| "learning_rate": 2.53968253968254e-06, |
| "loss": 0.2136, |
| "step": 3850 |
| }, |
| { |
| "epoch": 8.824929178470255, |
| "grad_norm": 0.04065897379519029, |
| "learning_rate": 2.3129251700680273e-06, |
| "loss": 0.2136, |
| "step": 3900 |
| }, |
| { |
| "epoch": 8.938243626062324, |
| "grad_norm": 0.04230058689139044, |
| "learning_rate": 2.086167800453515e-06, |
| "loss": 0.2135, |
| "step": 3950 |
| }, |
| { |
| "epoch": 9.04985835694051, |
| "grad_norm": 0.038128150813345994, |
| "learning_rate": 1.8594104308390023e-06, |
| "loss": 0.2102, |
| "step": 4000 |
| }, |
| { |
| "epoch": 9.163172804532579, |
| "grad_norm": 0.03384083547412833, |
| "learning_rate": 1.6326530612244897e-06, |
| "loss": 0.2132, |
| "step": 4050 |
| }, |
| { |
| "epoch": 9.276487252124646, |
| "grad_norm": 0.03967958263100997, |
| "learning_rate": 1.4058956916099775e-06, |
| "loss": 0.2131, |
| "step": 4100 |
| }, |
| { |
| "epoch": 9.389801699716713, |
| "grad_norm": 0.04693574868490394, |
| "learning_rate": 1.179138321995465e-06, |
| "loss": 0.2131, |
| "step": 4150 |
| }, |
| { |
| "epoch": 9.503116147308782, |
| "grad_norm": 0.03686909642390895, |
| "learning_rate": 9.523809523809525e-07, |
| "loss": 0.213, |
| "step": 4200 |
| }, |
| { |
| "epoch": 9.61643059490085, |
| "grad_norm": 0.04021478097064685, |
| "learning_rate": 7.2562358276644e-07, |
| "loss": 0.213, |
| "step": 4250 |
| }, |
| { |
| "epoch": 9.729745042492917, |
| "grad_norm": 0.034565423196844705, |
| "learning_rate": 4.988662131519275e-07, |
| "loss": 0.2129, |
| "step": 4300 |
| }, |
| { |
| "epoch": 9.843059490084986, |
| "grad_norm": 0.04106503466855132, |
| "learning_rate": 2.72108843537415e-07, |
| "loss": 0.2128, |
| "step": 4350 |
| }, |
| { |
| "epoch": 9.956373937677053, |
| "grad_norm": 0.03631468837696881, |
| "learning_rate": 4.53514739229025e-08, |
| "loss": 0.2127, |
| "step": 4400 |
| } |
| ], |
| "logging_steps": 50, |
| "max_steps": 4410, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 10, |
| "save_steps": 1000, |
| "stateful_callbacks": { |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": true, |
| "should_training_stop": true |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 1.1136582509658112e+16, |
| "train_batch_size": 8, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
