| { | |
| "best_metric": 28.696460806301626, | |
| "best_model_checkpoint": "./whisper-tiny-hi/checkpoint-45000", | |
| "epoch": 14.46952886316128, | |
| "eval_steps": 5000, | |
| "global_step": 45000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.06431902235086026, | |
| "grad_norm": 4.596249103546143, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 1.7191, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.12863804470172052, | |
| "grad_norm": 3.4318807125091553, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 0.9245, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.1929570670525808, | |
| "grad_norm": 3.58982253074646, | |
| "learning_rate": 6e-06, | |
| "loss": 0.6853, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.25727608940344104, | |
| "grad_norm": 3.402555227279663, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 0.5814, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.32159511175430133, | |
| "grad_norm": 3.178177833557129, | |
| "learning_rate": 1e-05, | |
| "loss": 0.5202, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.3859141341051616, | |
| "grad_norm": 3.04317569732666, | |
| "learning_rate": 9.96730957829356e-06, | |
| "loss": 0.4719, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.45023315645602185, | |
| "grad_norm": 3.090500831604004, | |
| "learning_rate": 9.93461915658712e-06, | |
| "loss": 0.4444, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.5145521788068821, | |
| "grad_norm": 3.2675209045410156, | |
| "learning_rate": 9.90192873488068e-06, | |
| "loss": 0.4275, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.5788712011577424, | |
| "grad_norm": 2.859715223312378, | |
| "learning_rate": 9.869238313174241e-06, | |
| "loss": 0.4076, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.6431902235086027, | |
| "grad_norm": 2.6303207874298096, | |
| "learning_rate": 9.836547891467801e-06, | |
| "loss": 0.3917, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.707509245859463, | |
| "grad_norm": 3.0497164726257324, | |
| "learning_rate": 9.804020921869893e-06, | |
| "loss": 0.381, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.7718282682103232, | |
| "grad_norm": 2.6169981956481934, | |
| "learning_rate": 9.771330500163453e-06, | |
| "loss": 0.3719, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.8361472905611834, | |
| "grad_norm": 2.708853006362915, | |
| "learning_rate": 9.738640078457014e-06, | |
| "loss": 0.3635, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.9004663129120437, | |
| "grad_norm": 2.8617489337921143, | |
| "learning_rate": 9.705949656750572e-06, | |
| "loss": 0.354, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.964785335262904, | |
| "grad_norm": 2.758882999420166, | |
| "learning_rate": 9.673259235044132e-06, | |
| "loss": 0.3474, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.0289435600578871, | |
| "grad_norm": 2.680941343307495, | |
| "learning_rate": 9.640568813337693e-06, | |
| "loss": 0.3395, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 1.0932625824087474, | |
| "grad_norm": 2.918921709060669, | |
| "learning_rate": 9.607878391631253e-06, | |
| "loss": 0.3299, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 1.1575816047596077, | |
| "grad_norm": 2.6285266876220703, | |
| "learning_rate": 9.575187969924813e-06, | |
| "loss": 0.3231, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 1.221900627110468, | |
| "grad_norm": 2.9504473209381104, | |
| "learning_rate": 9.542497548218372e-06, | |
| "loss": 0.3217, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 1.2862196494613283, | |
| "grad_norm": 2.8353495597839355, | |
| "learning_rate": 9.509807126511932e-06, | |
| "loss": 0.32, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.3505386718121883, | |
| "grad_norm": 2.4936184883117676, | |
| "learning_rate": 9.477280156914024e-06, | |
| "loss": 0.3141, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 1.4148576941630489, | |
| "grad_norm": 2.6884140968322754, | |
| "learning_rate": 9.444589735207586e-06, | |
| "loss": 0.3086, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 1.479176716513909, | |
| "grad_norm": 2.8961310386657715, | |
| "learning_rate": 9.411899313501145e-06, | |
| "loss": 0.308, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 1.5434957388647692, | |
| "grad_norm": 2.5737783908843994, | |
| "learning_rate": 9.379208891794705e-06, | |
| "loss": 0.3019, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 1.6078147612156295, | |
| "grad_norm": 2.5709033012390137, | |
| "learning_rate": 9.346518470088265e-06, | |
| "loss": 0.2974, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.6078147612156295, | |
| "eval_loss": 0.2771373987197876, | |
| "eval_runtime": 5695.1238, | |
| "eval_samples_per_second": 5.115, | |
| "eval_steps_per_second": 0.639, | |
| "eval_wer": 36.94923502637938, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.6721337835664898, | |
| "grad_norm": 2.3421154022216797, | |
| "learning_rate": 9.313828048381824e-06, | |
| "loss": 0.2959, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 1.73645280591735, | |
| "grad_norm": 2.662564277648926, | |
| "learning_rate": 9.281137626675386e-06, | |
| "loss": 0.2941, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 1.8007718282682104, | |
| "grad_norm": 2.4682416915893555, | |
| "learning_rate": 9.248447204968944e-06, | |
| "loss": 0.2932, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 1.8650908506190707, | |
| "grad_norm": 2.6210873126983643, | |
| "learning_rate": 9.215756783262505e-06, | |
| "loss": 0.2895, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 1.9294098729699307, | |
| "grad_norm": 2.6296238899230957, | |
| "learning_rate": 9.183066361556065e-06, | |
| "loss": 0.286, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 1.9937288953207912, | |
| "grad_norm": 2.486929178237915, | |
| "learning_rate": 9.150539391958157e-06, | |
| "loss": 0.2846, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 2.0578871201157742, | |
| "grad_norm": 2.202758550643921, | |
| "learning_rate": 9.117848970251717e-06, | |
| "loss": 0.2754, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 2.1222061424666343, | |
| "grad_norm": 2.4207451343536377, | |
| "learning_rate": 9.085158548545276e-06, | |
| "loss": 0.2745, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 2.186525164817495, | |
| "grad_norm": 2.1694822311401367, | |
| "learning_rate": 9.052468126838838e-06, | |
| "loss": 0.2732, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 2.250844187168355, | |
| "grad_norm": 2.7656798362731934, | |
| "learning_rate": 9.019777705132396e-06, | |
| "loss": 0.2709, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 2.3151632095192154, | |
| "grad_norm": 2.2421224117279053, | |
| "learning_rate": 8.987087283425957e-06, | |
| "loss": 0.2691, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 2.3794822318700755, | |
| "grad_norm": 2.469956874847412, | |
| "learning_rate": 8.954396861719517e-06, | |
| "loss": 0.268, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 2.443801254220936, | |
| "grad_norm": 2.4373421669006348, | |
| "learning_rate": 8.921706440013077e-06, | |
| "loss": 0.2656, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 2.508120276571796, | |
| "grad_norm": 2.5110926628112793, | |
| "learning_rate": 8.889016018306636e-06, | |
| "loss": 0.2668, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 2.5724392989226565, | |
| "grad_norm": 2.559202194213867, | |
| "learning_rate": 8.856325596600198e-06, | |
| "loss": 0.2623, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 2.6367583212735166, | |
| "grad_norm": 2.6701242923736572, | |
| "learning_rate": 8.82379862700229e-06, | |
| "loss": 0.2642, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 2.7010773436243767, | |
| "grad_norm": 2.3247900009155273, | |
| "learning_rate": 8.791108205295848e-06, | |
| "loss": 0.2621, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 2.765396365975237, | |
| "grad_norm": 2.2806427478790283, | |
| "learning_rate": 8.75841778358941e-06, | |
| "loss": 0.2601, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 2.8297153883260977, | |
| "grad_norm": 2.487159013748169, | |
| "learning_rate": 8.725727361882969e-06, | |
| "loss": 0.2596, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 2.8940344106769578, | |
| "grad_norm": 2.2094194889068604, | |
| "learning_rate": 8.693036940176529e-06, | |
| "loss": 0.2587, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 2.958353433027818, | |
| "grad_norm": 2.349698781967163, | |
| "learning_rate": 8.660346518470088e-06, | |
| "loss": 0.2569, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 3.0225116578228013, | |
| "grad_norm": 2.1951253414154053, | |
| "learning_rate": 8.62765609676365e-06, | |
| "loss": 0.2507, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 3.0868306801736614, | |
| "grad_norm": 2.2560884952545166, | |
| "learning_rate": 8.594965675057208e-06, | |
| "loss": 0.2469, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 3.1511497025245214, | |
| "grad_norm": 2.3260977268218994, | |
| "learning_rate": 8.562275253350769e-06, | |
| "loss": 0.2457, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 3.215468724875382, | |
| "grad_norm": 2.318385601043701, | |
| "learning_rate": 8.529584831644329e-06, | |
| "loss": 0.2463, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 3.215468724875382, | |
| "eval_loss": 0.23996804654598236, | |
| "eval_runtime": 5706.3154, | |
| "eval_samples_per_second": 5.105, | |
| "eval_steps_per_second": 0.638, | |
| "eval_wer": 33.16859977100777, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 3.279787747226242, | |
| "grad_norm": 2.278331756591797, | |
| "learning_rate": 8.496894409937889e-06, | |
| "loss": 0.2471, | |
| "step": 10200 | |
| }, | |
| { | |
| "epoch": 3.3441067695771025, | |
| "grad_norm": 2.1796886920928955, | |
| "learning_rate": 8.46436744033998e-06, | |
| "loss": 0.244, | |
| "step": 10400 | |
| }, | |
| { | |
| "epoch": 3.4084257919279626, | |
| "grad_norm": 2.3623690605163574, | |
| "learning_rate": 8.43167701863354e-06, | |
| "loss": 0.2453, | |
| "step": 10600 | |
| }, | |
| { | |
| "epoch": 3.472744814278823, | |
| "grad_norm": 2.242501974105835, | |
| "learning_rate": 8.398986596927102e-06, | |
| "loss": 0.2436, | |
| "step": 10800 | |
| }, | |
| { | |
| "epoch": 3.537063836629683, | |
| "grad_norm": 2.4463303089141846, | |
| "learning_rate": 8.366296175220661e-06, | |
| "loss": 0.244, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 3.6013828589805437, | |
| "grad_norm": 2.387897253036499, | |
| "learning_rate": 8.333605753514221e-06, | |
| "loss": 0.2436, | |
| "step": 11200 | |
| }, | |
| { | |
| "epoch": 3.6657018813314037, | |
| "grad_norm": 2.386038303375244, | |
| "learning_rate": 8.30091533180778e-06, | |
| "loss": 0.2431, | |
| "step": 11400 | |
| }, | |
| { | |
| "epoch": 3.730020903682264, | |
| "grad_norm": 2.2631173133850098, | |
| "learning_rate": 8.26822491010134e-06, | |
| "loss": 0.2399, | |
| "step": 11600 | |
| }, | |
| { | |
| "epoch": 3.7943399260331243, | |
| "grad_norm": 2.458087921142578, | |
| "learning_rate": 8.2355344883949e-06, | |
| "loss": 0.2409, | |
| "step": 11800 | |
| }, | |
| { | |
| "epoch": 3.858658948383985, | |
| "grad_norm": 2.400782585144043, | |
| "learning_rate": 8.202844066688462e-06, | |
| "loss": 0.2396, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 3.922977970734845, | |
| "grad_norm": 2.1437551975250244, | |
| "learning_rate": 8.170153644982021e-06, | |
| "loss": 0.2375, | |
| "step": 12200 | |
| }, | |
| { | |
| "epoch": 3.987296993085705, | |
| "grad_norm": 2.8162059783935547, | |
| "learning_rate": 8.137626675384113e-06, | |
| "loss": 0.2385, | |
| "step": 12400 | |
| }, | |
| { | |
| "epoch": 4.051455217880688, | |
| "grad_norm": 2.505800485610962, | |
| "learning_rate": 8.104936253677673e-06, | |
| "loss": 0.2317, | |
| "step": 12600 | |
| }, | |
| { | |
| "epoch": 4.1157742402315485, | |
| "grad_norm": 2.2435059547424316, | |
| "learning_rate": 8.072245831971233e-06, | |
| "loss": 0.2301, | |
| "step": 12800 | |
| }, | |
| { | |
| "epoch": 4.1800932625824085, | |
| "grad_norm": 2.208909034729004, | |
| "learning_rate": 8.039555410264793e-06, | |
| "loss": 0.2305, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 4.244412284933269, | |
| "grad_norm": 2.0333073139190674, | |
| "learning_rate": 8.006864988558352e-06, | |
| "loss": 0.2283, | |
| "step": 13200 | |
| }, | |
| { | |
| "epoch": 4.30873130728413, | |
| "grad_norm": 2.437255620956421, | |
| "learning_rate": 7.974174566851914e-06, | |
| "loss": 0.2285, | |
| "step": 13400 | |
| }, | |
| { | |
| "epoch": 4.37305032963499, | |
| "grad_norm": 2.1313934326171875, | |
| "learning_rate": 7.941484145145473e-06, | |
| "loss": 0.2268, | |
| "step": 13600 | |
| }, | |
| { | |
| "epoch": 4.43736935198585, | |
| "grad_norm": 2.3695502281188965, | |
| "learning_rate": 7.908793723439033e-06, | |
| "loss": 0.2281, | |
| "step": 13800 | |
| }, | |
| { | |
| "epoch": 4.50168837433671, | |
| "grad_norm": 2.3714799880981445, | |
| "learning_rate": 7.876103301732593e-06, | |
| "loss": 0.2285, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 4.566007396687571, | |
| "grad_norm": 2.148329019546509, | |
| "learning_rate": 7.843412880026154e-06, | |
| "loss": 0.2285, | |
| "step": 14200 | |
| }, | |
| { | |
| "epoch": 4.630326419038431, | |
| "grad_norm": 2.6452717781066895, | |
| "learning_rate": 7.810885910428245e-06, | |
| "loss": 0.2286, | |
| "step": 14400 | |
| }, | |
| { | |
| "epoch": 4.694645441389291, | |
| "grad_norm": 2.203331470489502, | |
| "learning_rate": 7.778195488721804e-06, | |
| "loss": 0.2262, | |
| "step": 14600 | |
| }, | |
| { | |
| "epoch": 4.758964463740151, | |
| "grad_norm": 2.239518642425537, | |
| "learning_rate": 7.745505067015366e-06, | |
| "loss": 0.2255, | |
| "step": 14800 | |
| }, | |
| { | |
| "epoch": 4.823283486091011, | |
| "grad_norm": 2.2322895526885986, | |
| "learning_rate": 7.712814645308925e-06, | |
| "loss": 0.2284, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 4.823283486091011, | |
| "eval_loss": 0.22340841591358185, | |
| "eval_runtime": 6283.1507, | |
| "eval_samples_per_second": 4.636, | |
| "eval_steps_per_second": 0.58, | |
| "eval_wer": 31.49793737232363, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 4.887602508441872, | |
| "grad_norm": 2.171865224838257, | |
| "learning_rate": 7.680124223602485e-06, | |
| "loss": 0.2286, | |
| "step": 15200 | |
| }, | |
| { | |
| "epoch": 4.951921530792732, | |
| "grad_norm": 2.2217211723327637, | |
| "learning_rate": 7.647433801896045e-06, | |
| "loss": 0.2237, | |
| "step": 15400 | |
| }, | |
| { | |
| "epoch": 5.0160797555877155, | |
| "grad_norm": 2.2957966327667236, | |
| "learning_rate": 7.614743380189605e-06, | |
| "loss": 0.2247, | |
| "step": 15600 | |
| }, | |
| { | |
| "epoch": 5.0803987779385755, | |
| "grad_norm": 2.2841222286224365, | |
| "learning_rate": 7.582052958483165e-06, | |
| "loss": 0.2184, | |
| "step": 15800 | |
| }, | |
| { | |
| "epoch": 5.144717800289436, | |
| "grad_norm": 2.9461894035339355, | |
| "learning_rate": 7.5493625367767255e-06, | |
| "loss": 0.2176, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 5.209036822640296, | |
| "grad_norm": 2.2747585773468018, | |
| "learning_rate": 7.516672115070285e-06, | |
| "loss": 0.2171, | |
| "step": 16200 | |
| }, | |
| { | |
| "epoch": 5.273355844991156, | |
| "grad_norm": 2.220026969909668, | |
| "learning_rate": 7.4841451454723765e-06, | |
| "loss": 0.217, | |
| "step": 16400 | |
| }, | |
| { | |
| "epoch": 5.337674867342017, | |
| "grad_norm": 2.335299253463745, | |
| "learning_rate": 7.451454723765937e-06, | |
| "loss": 0.218, | |
| "step": 16600 | |
| }, | |
| { | |
| "epoch": 5.401993889692877, | |
| "grad_norm": 2.2069528102874756, | |
| "learning_rate": 7.418764302059497e-06, | |
| "loss": 0.2172, | |
| "step": 16800 | |
| }, | |
| { | |
| "epoch": 5.466312912043737, | |
| "grad_norm": 2.500458002090454, | |
| "learning_rate": 7.386073880353057e-06, | |
| "loss": 0.2174, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 5.530631934394597, | |
| "grad_norm": 2.2738373279571533, | |
| "learning_rate": 7.353383458646617e-06, | |
| "loss": 0.2186, | |
| "step": 17200 | |
| }, | |
| { | |
| "epoch": 5.594950956745458, | |
| "grad_norm": 2.0394363403320312, | |
| "learning_rate": 7.3206930369401776e-06, | |
| "loss": 0.2168, | |
| "step": 17400 | |
| }, | |
| { | |
| "epoch": 5.659269979096318, | |
| "grad_norm": 2.0548949241638184, | |
| "learning_rate": 7.288002615233737e-06, | |
| "loss": 0.2167, | |
| "step": 17600 | |
| }, | |
| { | |
| "epoch": 5.723589001447178, | |
| "grad_norm": 2.0610129833221436, | |
| "learning_rate": 7.255312193527298e-06, | |
| "loss": 0.2154, | |
| "step": 17800 | |
| }, | |
| { | |
| "epoch": 5.787908023798038, | |
| "grad_norm": 2.0888965129852295, | |
| "learning_rate": 7.222621771820857e-06, | |
| "loss": 0.2157, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 5.852227046148899, | |
| "grad_norm": 2.3282668590545654, | |
| "learning_rate": 7.189931350114417e-06, | |
| "loss": 0.2162, | |
| "step": 18200 | |
| }, | |
| { | |
| "epoch": 5.916546068499759, | |
| "grad_norm": 2.2571957111358643, | |
| "learning_rate": 7.157404380516509e-06, | |
| "loss": 0.2174, | |
| "step": 18400 | |
| }, | |
| { | |
| "epoch": 5.980865090850619, | |
| "grad_norm": 2.201342821121216, | |
| "learning_rate": 7.124713958810069e-06, | |
| "loss": 0.2137, | |
| "step": 18600 | |
| }, | |
| { | |
| "epoch": 6.045023315645603, | |
| "grad_norm": 2.3466155529022217, | |
| "learning_rate": 7.09202353710363e-06, | |
| "loss": 0.2098, | |
| "step": 18800 | |
| }, | |
| { | |
| "epoch": 6.109342337996463, | |
| "grad_norm": 2.0573270320892334, | |
| "learning_rate": 7.059333115397189e-06, | |
| "loss": 0.2089, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 6.173661360347323, | |
| "grad_norm": 2.026381492614746, | |
| "learning_rate": 7.02664269369075e-06, | |
| "loss": 0.2085, | |
| "step": 19200 | |
| }, | |
| { | |
| "epoch": 6.237980382698183, | |
| "grad_norm": 2.2646751403808594, | |
| "learning_rate": 6.993952271984309e-06, | |
| "loss": 0.2074, | |
| "step": 19400 | |
| }, | |
| { | |
| "epoch": 6.302299405049043, | |
| "grad_norm": 2.0940823554992676, | |
| "learning_rate": 6.961261850277869e-06, | |
| "loss": 0.2086, | |
| "step": 19600 | |
| }, | |
| { | |
| "epoch": 6.366618427399904, | |
| "grad_norm": 2.2303521633148193, | |
| "learning_rate": 6.928571428571429e-06, | |
| "loss": 0.2096, | |
| "step": 19800 | |
| }, | |
| { | |
| "epoch": 6.430937449750764, | |
| "grad_norm": 2.2108261585235596, | |
| "learning_rate": 6.8958810068649895e-06, | |
| "loss": 0.2089, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 6.430937449750764, | |
| "eval_loss": 0.2148449569940567, | |
| "eval_runtime": 6212.1239, | |
| "eval_samples_per_second": 4.689, | |
| "eval_steps_per_second": 0.586, | |
| "eval_wer": 30.543488717322866, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 6.495256472101624, | |
| "grad_norm": 2.3403568267822266, | |
| "learning_rate": 6.863190585158549e-06, | |
| "loss": 0.2096, | |
| "step": 20200 | |
| }, | |
| { | |
| "epoch": 6.559575494452484, | |
| "grad_norm": 2.071211576461792, | |
| "learning_rate": 6.830663615560641e-06, | |
| "loss": 0.2076, | |
| "step": 20400 | |
| }, | |
| { | |
| "epoch": 6.623894516803345, | |
| "grad_norm": 2.031785488128662, | |
| "learning_rate": 6.797973193854202e-06, | |
| "loss": 0.2075, | |
| "step": 20600 | |
| }, | |
| { | |
| "epoch": 6.688213539154205, | |
| "grad_norm": 2.1697540283203125, | |
| "learning_rate": 6.765282772147761e-06, | |
| "loss": 0.2086, | |
| "step": 20800 | |
| }, | |
| { | |
| "epoch": 6.752532561505065, | |
| "grad_norm": 2.1642816066741943, | |
| "learning_rate": 6.732592350441321e-06, | |
| "loss": 0.2094, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 6.816851583855925, | |
| "grad_norm": 2.093594789505005, | |
| "learning_rate": 6.699901928734881e-06, | |
| "loss": 0.2065, | |
| "step": 21200 | |
| }, | |
| { | |
| "epoch": 6.881170606206785, | |
| "grad_norm": 2.0487377643585205, | |
| "learning_rate": 6.6672115070284415e-06, | |
| "loss": 0.2074, | |
| "step": 21400 | |
| }, | |
| { | |
| "epoch": 6.945489628557646, | |
| "grad_norm": 2.0511105060577393, | |
| "learning_rate": 6.634521085322001e-06, | |
| "loss": 0.2072, | |
| "step": 21600 | |
| }, | |
| { | |
| "epoch": 7.009647853352629, | |
| "grad_norm": 2.1460208892822266, | |
| "learning_rate": 6.601830663615562e-06, | |
| "loss": 0.204, | |
| "step": 21800 | |
| }, | |
| { | |
| "epoch": 7.07396687570349, | |
| "grad_norm": 2.093196392059326, | |
| "learning_rate": 6.569140241909121e-06, | |
| "loss": 0.2013, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 7.13828589805435, | |
| "grad_norm": 2.342313766479492, | |
| "learning_rate": 6.536449820202682e-06, | |
| "loss": 0.1994, | |
| "step": 22200 | |
| }, | |
| { | |
| "epoch": 7.20260492040521, | |
| "grad_norm": 2.05419921875, | |
| "learning_rate": 6.503759398496241e-06, | |
| "loss": 0.2023, | |
| "step": 22400 | |
| }, | |
| { | |
| "epoch": 7.26692394275607, | |
| "grad_norm": 2.248352289199829, | |
| "learning_rate": 6.471232428898333e-06, | |
| "loss": 0.2024, | |
| "step": 22600 | |
| }, | |
| { | |
| "epoch": 7.33124296510693, | |
| "grad_norm": 2.308692216873169, | |
| "learning_rate": 6.4385420071918935e-06, | |
| "loss": 0.2007, | |
| "step": 22800 | |
| }, | |
| { | |
| "epoch": 7.395561987457791, | |
| "grad_norm": 2.0985047817230225, | |
| "learning_rate": 6.405851585485453e-06, | |
| "loss": 0.1989, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 7.459881009808651, | |
| "grad_norm": 2.1786727905273438, | |
| "learning_rate": 6.373161163779014e-06, | |
| "loss": 0.2009, | |
| "step": 23200 | |
| }, | |
| { | |
| "epoch": 7.524200032159511, | |
| "grad_norm": 2.358017921447754, | |
| "learning_rate": 6.3404707420725734e-06, | |
| "loss": 0.2011, | |
| "step": 23400 | |
| }, | |
| { | |
| "epoch": 7.588519054510371, | |
| "grad_norm": 2.221282482147217, | |
| "learning_rate": 6.307780320366134e-06, | |
| "loss": 0.2, | |
| "step": 23600 | |
| }, | |
| { | |
| "epoch": 7.652838076861232, | |
| "grad_norm": 2.1816442012786865, | |
| "learning_rate": 6.275089898659693e-06, | |
| "loss": 0.1995, | |
| "step": 23800 | |
| }, | |
| { | |
| "epoch": 7.717157099212092, | |
| "grad_norm": 2.1074419021606445, | |
| "learning_rate": 6.242399476953253e-06, | |
| "loss": 0.2008, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 7.781476121562952, | |
| "grad_norm": 2.12418532371521, | |
| "learning_rate": 6.209709055246813e-06, | |
| "loss": 0.2029, | |
| "step": 24200 | |
| }, | |
| { | |
| "epoch": 7.845795143913812, | |
| "grad_norm": 2.0285849571228027, | |
| "learning_rate": 6.177018633540374e-06, | |
| "loss": 0.2005, | |
| "step": 24400 | |
| }, | |
| { | |
| "epoch": 7.910114166264673, | |
| "grad_norm": 2.1275014877319336, | |
| "learning_rate": 6.144491663942466e-06, | |
| "loss": 0.2014, | |
| "step": 24600 | |
| }, | |
| { | |
| "epoch": 7.974433188615533, | |
| "grad_norm": 2.047292709350586, | |
| "learning_rate": 6.1118012422360254e-06, | |
| "loss": 0.1999, | |
| "step": 24800 | |
| }, | |
| { | |
| "epoch": 8.038591413410517, | |
| "grad_norm": 2.2886574268341064, | |
| "learning_rate": 6.079110820529586e-06, | |
| "loss": 0.197, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 8.038591413410517, | |
| "eval_loss": 0.20965221524238586, | |
| "eval_runtime": 6307.1939, | |
| "eval_samples_per_second": 4.619, | |
| "eval_steps_per_second": 0.577, | |
| "eval_wer": 29.90140507668599, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 8.102910435761377, | |
| "grad_norm": 2.158328056335449, | |
| "learning_rate": 6.046420398823145e-06, | |
| "loss": 0.1942, | |
| "step": 25200 | |
| }, | |
| { | |
| "epoch": 8.167229458112237, | |
| "grad_norm": 2.0762622356414795, | |
| "learning_rate": 6.013729977116705e-06, | |
| "loss": 0.1941, | |
| "step": 25400 | |
| }, | |
| { | |
| "epoch": 8.231548480463097, | |
| "grad_norm": 2.2431387901306152, | |
| "learning_rate": 5.981039555410265e-06, | |
| "loss": 0.1951, | |
| "step": 25600 | |
| }, | |
| { | |
| "epoch": 8.295867502813957, | |
| "grad_norm": 1.9715090990066528, | |
| "learning_rate": 5.948349133703826e-06, | |
| "loss": 0.1946, | |
| "step": 25800 | |
| }, | |
| { | |
| "epoch": 8.360186525164817, | |
| "grad_norm": 2.1715543270111084, | |
| "learning_rate": 5.915658711997385e-06, | |
| "loss": 0.1949, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 8.424505547515677, | |
| "grad_norm": 2.1786696910858154, | |
| "learning_rate": 5.882968290290946e-06, | |
| "loss": 0.1954, | |
| "step": 26200 | |
| }, | |
| { | |
| "epoch": 8.488824569866537, | |
| "grad_norm": 2.2478559017181396, | |
| "learning_rate": 5.8502778685845056e-06, | |
| "loss": 0.1943, | |
| "step": 26400 | |
| }, | |
| { | |
| "epoch": 8.553143592217399, | |
| "grad_norm": 2.0193216800689697, | |
| "learning_rate": 5.817750898986597e-06, | |
| "loss": 0.1931, | |
| "step": 26600 | |
| }, | |
| { | |
| "epoch": 8.61746261456826, | |
| "grad_norm": 2.024120330810547, | |
| "learning_rate": 5.785060477280157e-06, | |
| "loss": 0.1944, | |
| "step": 26800 | |
| }, | |
| { | |
| "epoch": 8.68178163691912, | |
| "grad_norm": 1.9868948459625244, | |
| "learning_rate": 5.752370055573717e-06, | |
| "loss": 0.1967, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 8.74610065926998, | |
| "grad_norm": 2.0132243633270264, | |
| "learning_rate": 5.719679633867278e-06, | |
| "loss": 0.194, | |
| "step": 27200 | |
| }, | |
| { | |
| "epoch": 8.81041968162084, | |
| "grad_norm": 2.1960537433624268, | |
| "learning_rate": 5.686989212160837e-06, | |
| "loss": 0.1945, | |
| "step": 27400 | |
| }, | |
| { | |
| "epoch": 8.8747387039717, | |
| "grad_norm": 2.118748903274536, | |
| "learning_rate": 5.654298790454398e-06, | |
| "loss": 0.1943, | |
| "step": 27600 | |
| }, | |
| { | |
| "epoch": 8.93905772632256, | |
| "grad_norm": 2.0620296001434326, | |
| "learning_rate": 5.621608368747958e-06, | |
| "loss": 0.1948, | |
| "step": 27800 | |
| }, | |
| { | |
| "epoch": 9.003215951117543, | |
| "grad_norm": 2.0099258422851562, | |
| "learning_rate": 5.588917947041518e-06, | |
| "loss": 0.1953, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 9.067534973468403, | |
| "grad_norm": 2.1706490516662598, | |
| "learning_rate": 5.556227525335077e-06, | |
| "loss": 0.187, | |
| "step": 28200 | |
| }, | |
| { | |
| "epoch": 9.131853995819263, | |
| "grad_norm": 1.989235758781433, | |
| "learning_rate": 5.523537103628637e-06, | |
| "loss": 0.1896, | |
| "step": 28400 | |
| }, | |
| { | |
| "epoch": 9.196173018170123, | |
| "grad_norm": 2.4199020862579346, | |
| "learning_rate": 5.49101013403073e-06, | |
| "loss": 0.1889, | |
| "step": 28600 | |
| }, | |
| { | |
| "epoch": 9.260492040520983, | |
| "grad_norm": 2.318398952484131, | |
| "learning_rate": 5.458319712324289e-06, | |
| "loss": 0.1899, | |
| "step": 28800 | |
| }, | |
| { | |
| "epoch": 9.324811062871845, | |
| "grad_norm": 2.014559745788574, | |
| "learning_rate": 5.42562929061785e-06, | |
| "loss": 0.1899, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 9.389130085222705, | |
| "grad_norm": 2.0770111083984375, | |
| "learning_rate": 5.39293886891141e-06, | |
| "loss": 0.1904, | |
| "step": 29200 | |
| }, | |
| { | |
| "epoch": 9.453449107573565, | |
| "grad_norm": 2.1191565990448, | |
| "learning_rate": 5.36024844720497e-06, | |
| "loss": 0.1903, | |
| "step": 29400 | |
| }, | |
| { | |
| "epoch": 9.517768129924425, | |
| "grad_norm": 1.8838344812393188, | |
| "learning_rate": 5.327558025498529e-06, | |
| "loss": 0.191, | |
| "step": 29600 | |
| }, | |
| { | |
| "epoch": 9.582087152275285, | |
| "grad_norm": 2.064694404602051, | |
| "learning_rate": 5.2948676037920895e-06, | |
| "loss": 0.189, | |
| "step": 29800 | |
| }, | |
| { | |
| "epoch": 9.646406174626145, | |
| "grad_norm": 2.2259907722473145, | |
| "learning_rate": 5.262177182085649e-06, | |
| "loss": 0.1896, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 9.646406174626145, | |
| "eval_loss": 0.20603837072849274, | |
| "eval_runtime": 6469.2735, | |
| "eval_samples_per_second": 4.503, | |
| "eval_steps_per_second": 0.563, | |
| "eval_wer": 29.357595694795236, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 9.710725196977005, | |
| "grad_norm": 2.174774169921875, | |
| "learning_rate": 5.229486760379209e-06, | |
| "loss": 0.1898, | |
| "step": 30200 | |
| }, | |
| { | |
| "epoch": 9.775044219327865, | |
| "grad_norm": 2.170811414718628, | |
| "learning_rate": 5.1967963386727695e-06, | |
| "loss": 0.1885, | |
| "step": 30400 | |
| }, | |
| { | |
| "epoch": 9.839363241678726, | |
| "grad_norm": 2.454228401184082, | |
| "learning_rate": 5.164105916966329e-06, | |
| "loss": 0.1889, | |
| "step": 30600 | |
| }, | |
| { | |
| "epoch": 9.903682264029587, | |
| "grad_norm": 2.278226613998413, | |
| "learning_rate": 5.131578947368422e-06, | |
| "loss": 0.1891, | |
| "step": 30800 | |
| }, | |
| { | |
| "epoch": 9.968001286380447, | |
| "grad_norm": 2.151634931564331, | |
| "learning_rate": 5.098888525661981e-06, | |
| "loss": 0.1881, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 10.032159511175431, | |
| "grad_norm": 2.1410274505615234, | |
| "learning_rate": 5.0661981039555416e-06, | |
| "loss": 0.1861, | |
| "step": 31200 | |
| }, | |
| { | |
| "epoch": 10.096478533526291, | |
| "grad_norm": 2.1889536380767822, | |
| "learning_rate": 5.033507682249101e-06, | |
| "loss": 0.184, | |
| "step": 31400 | |
| }, | |
| { | |
| "epoch": 10.160797555877151, | |
| "grad_norm": 2.2102479934692383, | |
| "learning_rate": 5.000817260542662e-06, | |
| "loss": 0.1849, | |
| "step": 31600 | |
| }, | |
| { | |
| "epoch": 10.225116578228011, | |
| "grad_norm": 2.0018393993377686, | |
| "learning_rate": 4.9681268388362215e-06, | |
| "loss": 0.1833, | |
| "step": 31800 | |
| }, | |
| { | |
| "epoch": 10.289435600578871, | |
| "grad_norm": 2.114179849624634, | |
| "learning_rate": 4.935436417129781e-06, | |
| "loss": 0.1831, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 10.353754622929731, | |
| "grad_norm": 1.9474581480026245, | |
| "learning_rate": 4.902745995423342e-06, | |
| "loss": 0.1842, | |
| "step": 32200 | |
| }, | |
| { | |
| "epoch": 10.418073645280591, | |
| "grad_norm": 2.0356032848358154, | |
| "learning_rate": 4.8700555737169014e-06, | |
| "loss": 0.1852, | |
| "step": 32400 | |
| }, | |
| { | |
| "epoch": 10.482392667631451, | |
| "grad_norm": 2.0535342693328857, | |
| "learning_rate": 4.837365152010461e-06, | |
| "loss": 0.1866, | |
| "step": 32600 | |
| }, | |
| { | |
| "epoch": 10.546711689982311, | |
| "grad_norm": 2.2602412700653076, | |
| "learning_rate": 4.804838182412553e-06, | |
| "loss": 0.185, | |
| "step": 32800 | |
| }, | |
| { | |
| "epoch": 10.611030712333173, | |
| "grad_norm": 2.098829507827759, | |
| "learning_rate": 4.772147760706114e-06, | |
| "loss": 0.1844, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 10.675349734684033, | |
| "grad_norm": 2.0167388916015625, | |
| "learning_rate": 4.7394573389996735e-06, | |
| "loss": 0.1871, | |
| "step": 33200 | |
| }, | |
| { | |
| "epoch": 10.739668757034893, | |
| "grad_norm": 2.245697259902954, | |
| "learning_rate": 4.706766917293233e-06, | |
| "loss": 0.1866, | |
| "step": 33400 | |
| }, | |
| { | |
| "epoch": 10.803987779385753, | |
| "grad_norm": 1.9759703874588013, | |
| "learning_rate": 4.674076495586794e-06, | |
| "loss": 0.1849, | |
| "step": 33600 | |
| }, | |
| { | |
| "epoch": 10.868306801736614, | |
| "grad_norm": 2.1898162364959717, | |
| "learning_rate": 4.6413860738803535e-06, | |
| "loss": 0.1856, | |
| "step": 33800 | |
| }, | |
| { | |
| "epoch": 10.932625824087474, | |
| "grad_norm": 2.0218801498413086, | |
| "learning_rate": 4.608695652173913e-06, | |
| "loss": 0.1847, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 10.996944846438334, | |
| "grad_norm": 1.9428986310958862, | |
| "learning_rate": 4.576005230467474e-06, | |
| "loss": 0.1863, | |
| "step": 34200 | |
| }, | |
| { | |
| "epoch": 11.061103071233317, | |
| "grad_norm": 1.9156979322433472, | |
| "learning_rate": 4.543314808761033e-06, | |
| "loss": 0.1788, | |
| "step": 34400 | |
| }, | |
| { | |
| "epoch": 11.125422093584177, | |
| "grad_norm": 2.0453121662139893, | |
| "learning_rate": 4.510624387054593e-06, | |
| "loss": 0.1817, | |
| "step": 34600 | |
| }, | |
| { | |
| "epoch": 11.189741115935037, | |
| "grad_norm": 1.9229934215545654, | |
| "learning_rate": 4.477933965348154e-06, | |
| "loss": 0.1808, | |
| "step": 34800 | |
| }, | |
| { | |
| "epoch": 11.254060138285897, | |
| "grad_norm": 2.319345235824585, | |
| "learning_rate": 4.445406995750246e-06, | |
| "loss": 0.1793, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 11.254060138285897, | |
| "eval_loss": 0.20413178205490112, | |
| "eval_runtime": 6186.7151, | |
| "eval_samples_per_second": 4.708, | |
| "eval_steps_per_second": 0.589, | |
| "eval_wer": 29.14897509945317, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 11.31837916063676, | |
| "grad_norm": 2.07478404045105, | |
| "learning_rate": 4.4127165740438055e-06, | |
| "loss": 0.1798, | |
| "step": 35200 | |
| }, | |
| { | |
| "epoch": 11.38269818298762, | |
| "grad_norm": 2.0194761753082275, | |
| "learning_rate": 4.380026152337365e-06, | |
| "loss": 0.1792, | |
| "step": 35400 | |
| }, | |
| { | |
| "epoch": 11.44701720533848, | |
| "grad_norm": 1.903509497642517, | |
| "learning_rate": 4.347335730630926e-06, | |
| "loss": 0.1796, | |
| "step": 35600 | |
| }, | |
| { | |
| "epoch": 11.51133622768934, | |
| "grad_norm": 2.0315921306610107, | |
| "learning_rate": 4.314645308924485e-06, | |
| "loss": 0.1819, | |
| "step": 35800 | |
| }, | |
| { | |
| "epoch": 11.5756552500402, | |
| "grad_norm": 2.0309042930603027, | |
| "learning_rate": 4.281954887218046e-06, | |
| "loss": 0.1805, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 11.63997427239106, | |
| "grad_norm": 2.0719332695007324, | |
| "learning_rate": 4.249264465511606e-06, | |
| "loss": 0.1807, | |
| "step": 36200 | |
| }, | |
| { | |
| "epoch": 11.70429329474192, | |
| "grad_norm": 2.17317271232605, | |
| "learning_rate": 4.216574043805165e-06, | |
| "loss": 0.1832, | |
| "step": 36400 | |
| }, | |
| { | |
| "epoch": 11.76861231709278, | |
| "grad_norm": 2.0271966457366943, | |
| "learning_rate": 4.183883622098726e-06, | |
| "loss": 0.1804, | |
| "step": 36600 | |
| }, | |
| { | |
| "epoch": 11.83293133944364, | |
| "grad_norm": 2.235299825668335, | |
| "learning_rate": 4.151193200392286e-06, | |
| "loss": 0.1826, | |
| "step": 36800 | |
| }, | |
| { | |
| "epoch": 11.897250361794502, | |
| "grad_norm": 1.8323442935943604, | |
| "learning_rate": 4.118666230794378e-06, | |
| "loss": 0.1822, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 11.961569384145362, | |
| "grad_norm": 2.1653242111206055, | |
| "learning_rate": 4.0859758090879374e-06, | |
| "loss": 0.1829, | |
| "step": 37200 | |
| }, | |
| { | |
| "epoch": 12.025727608940343, | |
| "grad_norm": 2.1361212730407715, | |
| "learning_rate": 4.053285387381498e-06, | |
| "loss": 0.1792, | |
| "step": 37400 | |
| }, | |
| { | |
| "epoch": 12.090046631291205, | |
| "grad_norm": 2.1540911197662354, | |
| "learning_rate": 4.020594965675058e-06, | |
| "loss": 0.1779, | |
| "step": 37600 | |
| }, | |
| { | |
| "epoch": 12.154365653642065, | |
| "grad_norm": 2.157705783843994, | |
| "learning_rate": 3.987904543968617e-06, | |
| "loss": 0.1754, | |
| "step": 37800 | |
| }, | |
| { | |
| "epoch": 12.218684675992925, | |
| "grad_norm": 2.100783586502075, | |
| "learning_rate": 3.955214122262178e-06, | |
| "loss": 0.1761, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 12.283003698343785, | |
| "grad_norm": 1.9860684871673584, | |
| "learning_rate": 3.922523700555738e-06, | |
| "loss": 0.178, | |
| "step": 38200 | |
| }, | |
| { | |
| "epoch": 12.347322720694645, | |
| "grad_norm": 2.138315200805664, | |
| "learning_rate": 3.889833278849297e-06, | |
| "loss": 0.1769, | |
| "step": 38400 | |
| }, | |
| { | |
| "epoch": 12.411641743045505, | |
| "grad_norm": 1.8979172706604004, | |
| "learning_rate": 3.857142857142858e-06, | |
| "loss": 0.1761, | |
| "step": 38600 | |
| }, | |
| { | |
| "epoch": 12.475960765396366, | |
| "grad_norm": 2.144052505493164, | |
| "learning_rate": 3.8244524354364175e-06, | |
| "loss": 0.1772, | |
| "step": 38800 | |
| }, | |
| { | |
| "epoch": 12.540279787747226, | |
| "grad_norm": 2.2078232765197754, | |
| "learning_rate": 3.7919254658385097e-06, | |
| "loss": 0.1784, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 12.604598810098086, | |
| "grad_norm": 2.0689291954040527, | |
| "learning_rate": 3.75923504413207e-06, | |
| "loss": 0.179, | |
| "step": 39200 | |
| }, | |
| { | |
| "epoch": 12.668917832448948, | |
| "grad_norm": 2.1173605918884277, | |
| "learning_rate": 3.7265446224256295e-06, | |
| "loss": 0.1761, | |
| "step": 39400 | |
| }, | |
| { | |
| "epoch": 12.733236854799808, | |
| "grad_norm": 2.0470852851867676, | |
| "learning_rate": 3.6938542007191896e-06, | |
| "loss": 0.1791, | |
| "step": 39600 | |
| }, | |
| { | |
| "epoch": 12.797555877150668, | |
| "grad_norm": 2.237996816635132, | |
| "learning_rate": 3.6611637790127497e-06, | |
| "loss": 0.1791, | |
| "step": 39800 | |
| }, | |
| { | |
| "epoch": 12.861874899501528, | |
| "grad_norm": 2.107485055923462, | |
| "learning_rate": 3.62847335730631e-06, | |
| "loss": 0.1776, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 12.861874899501528, | |
| "eval_loss": 0.20320001244544983, | |
| "eval_runtime": 5730.0624, | |
| "eval_samples_per_second": 5.084, | |
| "eval_steps_per_second": 0.636, | |
| "eval_wer": 29.061075282609227, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 12.926193921852388, | |
| "grad_norm": 2.1215012073516846, | |
| "learning_rate": 3.5957829355998696e-06, | |
| "loss": 0.1792, | |
| "step": 40200 | |
| }, | |
| { | |
| "epoch": 12.990512944203248, | |
| "grad_norm": 2.1944947242736816, | |
| "learning_rate": 3.5630925138934297e-06, | |
| "loss": 0.177, | |
| "step": 40400 | |
| }, | |
| { | |
| "epoch": 13.054671168998231, | |
| "grad_norm": 2.1703405380249023, | |
| "learning_rate": 3.53040209218699e-06, | |
| "loss": 0.1754, | |
| "step": 40600 | |
| }, | |
| { | |
| "epoch": 13.118990191349091, | |
| "grad_norm": 2.224743366241455, | |
| "learning_rate": 3.49771167048055e-06, | |
| "loss": 0.1741, | |
| "step": 40800 | |
| }, | |
| { | |
| "epoch": 13.183309213699951, | |
| "grad_norm": 2.370253324508667, | |
| "learning_rate": 3.4650212487741096e-06, | |
| "loss": 0.1727, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 13.247628236050812, | |
| "grad_norm": 1.8502309322357178, | |
| "learning_rate": 3.4324942791762018e-06, | |
| "loss": 0.1732, | |
| "step": 41200 | |
| }, | |
| { | |
| "epoch": 13.311947258401672, | |
| "grad_norm": 2.1553802490234375, | |
| "learning_rate": 3.399803857469762e-06, | |
| "loss": 0.1731, | |
| "step": 41400 | |
| }, | |
| { | |
| "epoch": 13.376266280752532, | |
| "grad_norm": 2.113837242126465, | |
| "learning_rate": 3.3671134357633216e-06, | |
| "loss": 0.1743, | |
| "step": 41600 | |
| }, | |
| { | |
| "epoch": 13.440585303103393, | |
| "grad_norm": 1.9942282438278198, | |
| "learning_rate": 3.3344230140568817e-06, | |
| "loss": 0.1742, | |
| "step": 41800 | |
| }, | |
| { | |
| "epoch": 13.504904325454254, | |
| "grad_norm": 2.089869976043701, | |
| "learning_rate": 3.301732592350442e-06, | |
| "loss": 0.1754, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 13.569223347805114, | |
| "grad_norm": 1.8944735527038574, | |
| "learning_rate": 3.269042170644002e-06, | |
| "loss": 0.1742, | |
| "step": 42200 | |
| }, | |
| { | |
| "epoch": 13.633542370155974, | |
| "grad_norm": 2.06595516204834, | |
| "learning_rate": 3.2363517489375616e-06, | |
| "loss": 0.1736, | |
| "step": 42400 | |
| }, | |
| { | |
| "epoch": 13.697861392506834, | |
| "grad_norm": 2.0173680782318115, | |
| "learning_rate": 3.2036613272311218e-06, | |
| "loss": 0.1738, | |
| "step": 42600 | |
| }, | |
| { | |
| "epoch": 13.762180414857694, | |
| "grad_norm": 2.098628282546997, | |
| "learning_rate": 3.170970905524682e-06, | |
| "loss": 0.1735, | |
| "step": 42800 | |
| }, | |
| { | |
| "epoch": 13.826499437208554, | |
| "grad_norm": 2.226824998855591, | |
| "learning_rate": 3.138280483818241e-06, | |
| "loss": 0.1757, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 13.890818459559414, | |
| "grad_norm": 2.020033121109009, | |
| "learning_rate": 3.1057535142203337e-06, | |
| "loss": 0.1755, | |
| "step": 43200 | |
| }, | |
| { | |
| "epoch": 13.955137481910274, | |
| "grad_norm": 2.29543399810791, | |
| "learning_rate": 3.073063092513894e-06, | |
| "loss": 0.1762, | |
| "step": 43400 | |
| }, | |
| { | |
| "epoch": 14.019295706705257, | |
| "grad_norm": 1.9776209592819214, | |
| "learning_rate": 3.040372670807454e-06, | |
| "loss": 0.1735, | |
| "step": 43600 | |
| }, | |
| { | |
| "epoch": 14.083614729056118, | |
| "grad_norm": 1.7997843027114868, | |
| "learning_rate": 3.0076822491010137e-06, | |
| "loss": 0.1704, | |
| "step": 43800 | |
| }, | |
| { | |
| "epoch": 14.14793375140698, | |
| "grad_norm": 1.968531847000122, | |
| "learning_rate": 2.9749918273945738e-06, | |
| "loss": 0.1717, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 14.21225277375784, | |
| "grad_norm": 2.061990737915039, | |
| "learning_rate": 2.942301405688134e-06, | |
| "loss": 0.1726, | |
| "step": 44200 | |
| }, | |
| { | |
| "epoch": 14.2765717961087, | |
| "grad_norm": 2.0133299827575684, | |
| "learning_rate": 2.909610983981694e-06, | |
| "loss": 0.1714, | |
| "step": 44400 | |
| }, | |
| { | |
| "epoch": 14.34089081845956, | |
| "grad_norm": 2.0535309314727783, | |
| "learning_rate": 2.8769205622752537e-06, | |
| "loss": 0.1717, | |
| "step": 44600 | |
| }, | |
| { | |
| "epoch": 14.40520984081042, | |
| "grad_norm": 2.1071929931640625, | |
| "learning_rate": 2.8442301405688134e-06, | |
| "loss": 0.1722, | |
| "step": 44800 | |
| }, | |
| { | |
| "epoch": 14.46952886316128, | |
| "grad_norm": 2.0523617267608643, | |
| "learning_rate": 2.811539718862373e-06, | |
| "loss": 0.17, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 14.46952886316128, | |
| "eval_loss": 0.20234042406082153, | |
| "eval_runtime": 5780.3456, | |
| "eval_samples_per_second": 5.039, | |
| "eval_steps_per_second": 0.63, | |
| "eval_wer": 28.696460806301626, | |
| "step": 45000 | |
| } | |
| ], | |
| "logging_steps": 200, | |
| "max_steps": 62180, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 20, | |
| "save_steps": 5000, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1.4178134111256576e+20, | |
| "train_batch_size": 32, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |