| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.155569580309246, | |
| "eval_steps": 1000, | |
| "global_step": 5000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.015777847901546228, | |
| "grad_norm": 7.437136650085449, | |
| "learning_rate": 5.000000000000001e-07, | |
| "loss": 1.3977, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.031555695803092455, | |
| "grad_norm": 6.577849864959717, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 1.1112, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.047333543704638686, | |
| "grad_norm": 5.520799160003662, | |
| "learning_rate": 1.5e-06, | |
| "loss": 0.8031, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.06311139160618491, | |
| "grad_norm": 5.328930854797363, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 0.7451, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.07888923950773115, | |
| "grad_norm": 5.012049674987793, | |
| "learning_rate": 2.5e-06, | |
| "loss": 0.682, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.09466708740927737, | |
| "grad_norm": 4.783524036407471, | |
| "learning_rate": 3e-06, | |
| "loss": 0.6447, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.1104449353108236, | |
| "grad_norm": 4.749491214752197, | |
| "learning_rate": 3.5e-06, | |
| "loss": 0.6121, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.12622278321236982, | |
| "grad_norm": 4.470991134643555, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 0.6126, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.14200063111391606, | |
| "grad_norm": 4.39013147354126, | |
| "learning_rate": 4.5e-06, | |
| "loss": 0.5958, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.1577784790154623, | |
| "grad_norm": 4.321168422698975, | |
| "learning_rate": 5e-06, | |
| "loss": 0.5697, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.1735563269170085, | |
| "grad_norm": 4.567933082580566, | |
| "learning_rate": 5.500000000000001e-06, | |
| "loss": 0.5554, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.18933417481855475, | |
| "grad_norm": 4.7674150466918945, | |
| "learning_rate": 6e-06, | |
| "loss": 0.5646, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.20511202272010098, | |
| "grad_norm": 4.203701496124268, | |
| "learning_rate": 6.5000000000000004e-06, | |
| "loss": 0.5316, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.2208898706216472, | |
| "grad_norm": 4.289330005645752, | |
| "learning_rate": 7e-06, | |
| "loss": 0.527, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.23666771852319343, | |
| "grad_norm": 4.419729232788086, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 0.533, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.25244556642473964, | |
| "grad_norm": 3.873534917831421, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 0.5164, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.2682234143262859, | |
| "grad_norm": 3.9782721996307373, | |
| "learning_rate": 8.5e-06, | |
| "loss": 0.4957, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.2840012622278321, | |
| "grad_norm": 4.02784538269043, | |
| "learning_rate": 9e-06, | |
| "loss": 0.5142, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.29977911012937836, | |
| "grad_norm": 4.09860372543335, | |
| "learning_rate": 9.5e-06, | |
| "loss": 0.4987, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.3155569580309246, | |
| "grad_norm": 3.7876222133636475, | |
| "learning_rate": 1e-05, | |
| "loss": 0.5076, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.33133480593247083, | |
| "grad_norm": 4.0808916091918945, | |
| "learning_rate": 9.944444444444445e-06, | |
| "loss": 0.4824, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.347112653834017, | |
| "grad_norm": 3.7431557178497314, | |
| "learning_rate": 9.88888888888889e-06, | |
| "loss": 0.4829, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.36289050173556325, | |
| "grad_norm": 4.337482452392578, | |
| "learning_rate": 9.833333333333333e-06, | |
| "loss": 0.5014, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.3786683496371095, | |
| "grad_norm": 4.180824279785156, | |
| "learning_rate": 9.777777777777779e-06, | |
| "loss": 0.4592, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.39444619753865573, | |
| "grad_norm": 4.251559257507324, | |
| "learning_rate": 9.722222222222223e-06, | |
| "loss": 0.4665, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.41022404544020197, | |
| "grad_norm": 3.684410333633423, | |
| "learning_rate": 9.666666666666667e-06, | |
| "loss": 0.4652, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.4260018933417482, | |
| "grad_norm": 3.8727266788482666, | |
| "learning_rate": 9.611111111111112e-06, | |
| "loss": 0.44, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.4417797412432944, | |
| "grad_norm": 3.3472304344177246, | |
| "learning_rate": 9.555555555555556e-06, | |
| "loss": 0.4557, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.4575575891448406, | |
| "grad_norm": 3.535940170288086, | |
| "learning_rate": 9.5e-06, | |
| "loss": 0.4449, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.47333543704638686, | |
| "grad_norm": 4.009454250335693, | |
| "learning_rate": 9.444444444444445e-06, | |
| "loss": 0.431, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.4891132849479331, | |
| "grad_norm": 3.5649003982543945, | |
| "learning_rate": 9.38888888888889e-06, | |
| "loss": 0.4506, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.5048911328494793, | |
| "grad_norm": 3.044846296310425, | |
| "learning_rate": 9.333333333333334e-06, | |
| "loss": 0.427, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.5206689807510255, | |
| "grad_norm": 3.3232269287109375, | |
| "learning_rate": 9.277777777777778e-06, | |
| "loss": 0.4009, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.5364468286525718, | |
| "grad_norm": 3.964977741241455, | |
| "learning_rate": 9.222222222222224e-06, | |
| "loss": 0.4319, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.552224676554118, | |
| "grad_norm": 3.00532865524292, | |
| "learning_rate": 9.166666666666666e-06, | |
| "loss": 0.4082, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.5680025244556642, | |
| "grad_norm": 3.6661252975463867, | |
| "learning_rate": 9.111111111111112e-06, | |
| "loss": 0.4171, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.5837803723572105, | |
| "grad_norm": 2.911687135696411, | |
| "learning_rate": 9.055555555555556e-06, | |
| "loss": 0.405, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.5995582202587567, | |
| "grad_norm": 3.7377912998199463, | |
| "learning_rate": 9e-06, | |
| "loss": 0.4032, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.615336068160303, | |
| "grad_norm": 3.389277696609497, | |
| "learning_rate": 8.944444444444446e-06, | |
| "loss": 0.3968, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.6311139160618492, | |
| "grad_norm": 3.2546145915985107, | |
| "learning_rate": 8.888888888888888e-06, | |
| "loss": 0.408, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.6311139160618492, | |
| "eval_loss": 0.5, | |
| "eval_runtime": 3003.1872, | |
| "eval_samples_per_second": 2.338, | |
| "eval_steps_per_second": 0.146, | |
| "eval_wer": 0.3734339502318352, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.6468917639633954, | |
| "grad_norm": 3.6185717582702637, | |
| "learning_rate": 8.833333333333334e-06, | |
| "loss": 0.4092, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.6626696118649417, | |
| "grad_norm": 3.227266788482666, | |
| "learning_rate": 8.777777777777778e-06, | |
| "loss": 0.3839, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.6784474597664879, | |
| "grad_norm": 3.281687021255493, | |
| "learning_rate": 8.722222222222224e-06, | |
| "loss": 0.3954, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.694225307668034, | |
| "grad_norm": 3.359879732131958, | |
| "learning_rate": 8.666666666666668e-06, | |
| "loss": 0.4051, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.7100031555695803, | |
| "grad_norm": 2.7654247283935547, | |
| "learning_rate": 8.611111111111112e-06, | |
| "loss": 0.3979, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.7257810034711265, | |
| "grad_norm": 2.9193856716156006, | |
| "learning_rate": 8.555555555555556e-06, | |
| "loss": 0.4075, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.7415588513726727, | |
| "grad_norm": 3.3058297634124756, | |
| "learning_rate": 8.5e-06, | |
| "loss": 0.4176, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.757336699274219, | |
| "grad_norm": 3.126009941101074, | |
| "learning_rate": 8.444444444444446e-06, | |
| "loss": 0.3928, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.7731145471757652, | |
| "grad_norm": 2.937455415725708, | |
| "learning_rate": 8.38888888888889e-06, | |
| "loss": 0.3772, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.7888923950773115, | |
| "grad_norm": 2.8536481857299805, | |
| "learning_rate": 8.333333333333334e-06, | |
| "loss": 0.3968, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.8046702429788577, | |
| "grad_norm": 3.651503086090088, | |
| "learning_rate": 8.277777777777778e-06, | |
| "loss": 0.3758, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.8204480908804039, | |
| "grad_norm": 2.8864688873291016, | |
| "learning_rate": 8.222222222222222e-06, | |
| "loss": 0.3784, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.8362259387819502, | |
| "grad_norm": 3.2489211559295654, | |
| "learning_rate": 8.166666666666668e-06, | |
| "loss": 0.3632, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.8520037866834964, | |
| "grad_norm": 3.0353879928588867, | |
| "learning_rate": 8.111111111111112e-06, | |
| "loss": 0.3564, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.8677816345850426, | |
| "grad_norm": 3.2296388149261475, | |
| "learning_rate": 8.055555555555557e-06, | |
| "loss": 0.3546, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.8835594824865888, | |
| "grad_norm": 3.0037214756011963, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 0.3453, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.899337330388135, | |
| "grad_norm": 3.308013439178467, | |
| "learning_rate": 7.944444444444445e-06, | |
| "loss": 0.3578, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.9151151782896813, | |
| "grad_norm": 3.3210103511810303, | |
| "learning_rate": 7.88888888888889e-06, | |
| "loss": 0.3382, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.9308930261912275, | |
| "grad_norm": 3.2565624713897705, | |
| "learning_rate": 7.833333333333333e-06, | |
| "loss": 0.3718, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 0.9466708740927737, | |
| "grad_norm": 3.0643301010131836, | |
| "learning_rate": 7.77777777777778e-06, | |
| "loss": 0.3478, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.96244872199432, | |
| "grad_norm": 2.628441333770752, | |
| "learning_rate": 7.722222222222223e-06, | |
| "loss": 0.3654, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 0.9782265698958662, | |
| "grad_norm": 2.842191457748413, | |
| "learning_rate": 7.666666666666667e-06, | |
| "loss": 0.3727, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.9940044177974124, | |
| "grad_norm": 2.23713755607605, | |
| "learning_rate": 7.611111111111111e-06, | |
| "loss": 0.3369, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 1.0097822656989586, | |
| "grad_norm": 2.4432640075683594, | |
| "learning_rate": 7.555555555555556e-06, | |
| "loss": 0.2744, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 1.0255601136005048, | |
| "grad_norm": 2.3971242904663086, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 0.2449, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 1.041337961502051, | |
| "grad_norm": 2.4431796073913574, | |
| "learning_rate": 7.444444444444445e-06, | |
| "loss": 0.2499, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 1.0571158094035973, | |
| "grad_norm": 2.3628671169281006, | |
| "learning_rate": 7.38888888888889e-06, | |
| "loss": 0.2698, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 1.0728936573051435, | |
| "grad_norm": 2.0304720401763916, | |
| "learning_rate": 7.333333333333333e-06, | |
| "loss": 0.2418, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 1.0886715052066898, | |
| "grad_norm": 2.804999828338623, | |
| "learning_rate": 7.277777777777778e-06, | |
| "loss": 0.2506, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 1.104449353108236, | |
| "grad_norm": 3.027466297149658, | |
| "learning_rate": 7.222222222222223e-06, | |
| "loss": 0.2568, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 1.1202272010097822, | |
| "grad_norm": 2.6216728687286377, | |
| "learning_rate": 7.166666666666667e-06, | |
| "loss": 0.2417, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 1.1360050489113285, | |
| "grad_norm": 2.6749463081359863, | |
| "learning_rate": 7.111111111111112e-06, | |
| "loss": 0.2479, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 1.1517828968128747, | |
| "grad_norm": 2.4260010719299316, | |
| "learning_rate": 7.055555555555557e-06, | |
| "loss": 0.2385, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 1.167560744714421, | |
| "grad_norm": 2.3928089141845703, | |
| "learning_rate": 7e-06, | |
| "loss": 0.2321, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 1.1833385926159672, | |
| "grad_norm": 2.639481782913208, | |
| "learning_rate": 6.944444444444445e-06, | |
| "loss": 0.2429, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 1.1991164405175134, | |
| "grad_norm": 3.0133748054504395, | |
| "learning_rate": 6.88888888888889e-06, | |
| "loss": 0.252, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 1.2148942884190597, | |
| "grad_norm": 2.657839059829712, | |
| "learning_rate": 6.833333333333334e-06, | |
| "loss": 0.2453, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 1.230672136320606, | |
| "grad_norm": 2.745032787322998, | |
| "learning_rate": 6.777777777777779e-06, | |
| "loss": 0.2454, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 1.2464499842221521, | |
| "grad_norm": 2.7123420238494873, | |
| "learning_rate": 6.7222222222222235e-06, | |
| "loss": 0.2672, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 1.2622278321236984, | |
| "grad_norm": 2.789599657058716, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 0.2455, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.2622278321236984, | |
| "eval_loss": 0.4339774549007416, | |
| "eval_runtime": 2958.2023, | |
| "eval_samples_per_second": 2.374, | |
| "eval_steps_per_second": 0.148, | |
| "eval_wer": 0.32330707196853953, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.2780056800252446, | |
| "grad_norm": 3.1237881183624268, | |
| "learning_rate": 6.6111111111111115e-06, | |
| "loss": 0.2584, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 1.2937835279267909, | |
| "grad_norm": 2.1236650943756104, | |
| "learning_rate": 6.555555555555556e-06, | |
| "loss": 0.2289, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 1.309561375828337, | |
| "grad_norm": 3.098536968231201, | |
| "learning_rate": 6.5000000000000004e-06, | |
| "loss": 0.2392, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 1.325339223729883, | |
| "grad_norm": 2.5509822368621826, | |
| "learning_rate": 6.444444444444445e-06, | |
| "loss": 0.2298, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 1.3411170716314293, | |
| "grad_norm": 2.699902296066284, | |
| "learning_rate": 6.3888888888888885e-06, | |
| "loss": 0.2425, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 1.3568949195329756, | |
| "grad_norm": 2.5050745010375977, | |
| "learning_rate": 6.333333333333333e-06, | |
| "loss": 0.2377, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 1.3726727674345218, | |
| "grad_norm": 2.2431986331939697, | |
| "learning_rate": 6.277777777777778e-06, | |
| "loss": 0.225, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 1.388450615336068, | |
| "grad_norm": 2.3476831912994385, | |
| "learning_rate": 6.222222222222223e-06, | |
| "loss": 0.2458, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 1.4042284632376143, | |
| "grad_norm": 2.1803078651428223, | |
| "learning_rate": 6.166666666666667e-06, | |
| "loss": 0.2459, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 1.4200063111391605, | |
| "grad_norm": 2.475566864013672, | |
| "learning_rate": 6.111111111111112e-06, | |
| "loss": 0.2349, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 1.4357841590407068, | |
| "grad_norm": 2.735623598098755, | |
| "learning_rate": 6.055555555555555e-06, | |
| "loss": 0.2295, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 1.451562006942253, | |
| "grad_norm": 2.5407278537750244, | |
| "learning_rate": 6e-06, | |
| "loss": 0.2367, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 1.4673398548437993, | |
| "grad_norm": 2.189357042312622, | |
| "learning_rate": 5.944444444444445e-06, | |
| "loss": 0.2367, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 1.4831177027453455, | |
| "grad_norm": 2.2761070728302, | |
| "learning_rate": 5.88888888888889e-06, | |
| "loss": 0.2331, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 1.4988955506468917, | |
| "grad_norm": 2.657602071762085, | |
| "learning_rate": 5.833333333333334e-06, | |
| "loss": 0.2346, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 1.514673398548438, | |
| "grad_norm": 2.7713756561279297, | |
| "learning_rate": 5.777777777777778e-06, | |
| "loss": 0.2336, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 1.5304512464499842, | |
| "grad_norm": 2.4739747047424316, | |
| "learning_rate": 5.722222222222222e-06, | |
| "loss": 0.2314, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 1.5462290943515304, | |
| "grad_norm": 2.275024175643921, | |
| "learning_rate": 5.666666666666667e-06, | |
| "loss": 0.2384, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 1.5620069422530767, | |
| "grad_norm": 8.249900817871094, | |
| "learning_rate": 5.611111111111112e-06, | |
| "loss": 0.2137, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 1.577784790154623, | |
| "grad_norm": 2.453125, | |
| "learning_rate": 5.555555555555557e-06, | |
| "loss": 0.2445, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.5935626380561692, | |
| "grad_norm": 2.3670058250427246, | |
| "learning_rate": 5.500000000000001e-06, | |
| "loss": 0.2418, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 1.6093404859577154, | |
| "grad_norm": 2.700669527053833, | |
| "learning_rate": 5.444444444444445e-06, | |
| "loss": 0.2356, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 1.6251183338592616, | |
| "grad_norm": 2.0962114334106445, | |
| "learning_rate": 5.388888888888889e-06, | |
| "loss": 0.2268, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 1.6408961817608079, | |
| "grad_norm": 2.2385094165802, | |
| "learning_rate": 5.333333333333334e-06, | |
| "loss": 0.2292, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 1.656674029662354, | |
| "grad_norm": 2.2060489654541016, | |
| "learning_rate": 5.2777777777777785e-06, | |
| "loss": 0.2255, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 1.6724518775639003, | |
| "grad_norm": 2.859130859375, | |
| "learning_rate": 5.2222222222222226e-06, | |
| "loss": 0.2304, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 1.6882297254654466, | |
| "grad_norm": 2.400465250015259, | |
| "learning_rate": 5.1666666666666675e-06, | |
| "loss": 0.2375, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 1.7040075733669928, | |
| "grad_norm": 2.382413387298584, | |
| "learning_rate": 5.1111111111111115e-06, | |
| "loss": 0.2212, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 1.719785421268539, | |
| "grad_norm": 2.3223814964294434, | |
| "learning_rate": 5.0555555555555555e-06, | |
| "loss": 0.2274, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 1.7355632691700853, | |
| "grad_norm": 2.5504796504974365, | |
| "learning_rate": 5e-06, | |
| "loss": 0.2224, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 1.7513411170716315, | |
| "grad_norm": 1.8646706342697144, | |
| "learning_rate": 4.944444444444445e-06, | |
| "loss": 0.2109, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 1.7671189649731778, | |
| "grad_norm": 2.2399260997772217, | |
| "learning_rate": 4.888888888888889e-06, | |
| "loss": 0.2253, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 1.782896812874724, | |
| "grad_norm": 2.5138754844665527, | |
| "learning_rate": 4.833333333333333e-06, | |
| "loss": 0.2226, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 1.7986746607762703, | |
| "grad_norm": 2.8491804599761963, | |
| "learning_rate": 4.777777777777778e-06, | |
| "loss": 0.2281, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 1.8144525086778165, | |
| "grad_norm": 2.52425217628479, | |
| "learning_rate": 4.722222222222222e-06, | |
| "loss": 0.222, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 1.8302303565793627, | |
| "grad_norm": 2.4254088401794434, | |
| "learning_rate": 4.666666666666667e-06, | |
| "loss": 0.216, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 1.846008204480909, | |
| "grad_norm": 2.4229941368103027, | |
| "learning_rate": 4.611111111111112e-06, | |
| "loss": 0.2178, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 1.861786052382455, | |
| "grad_norm": 2.6721200942993164, | |
| "learning_rate": 4.555555555555556e-06, | |
| "loss": 0.2241, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 1.8775639002840012, | |
| "grad_norm": 2.3381521701812744, | |
| "learning_rate": 4.5e-06, | |
| "loss": 0.2255, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 1.8933417481855475, | |
| "grad_norm": 2.385972023010254, | |
| "learning_rate": 4.444444444444444e-06, | |
| "loss": 0.216, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.8933417481855475, | |
| "eval_loss": 0.40097150206565857, | |
| "eval_runtime": 2940.5315, | |
| "eval_samples_per_second": 2.388, | |
| "eval_steps_per_second": 0.149, | |
| "eval_wer": 0.3000305570686471, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.9091195960870937, | |
| "grad_norm": 2.507140636444092, | |
| "learning_rate": 4.388888888888889e-06, | |
| "loss": 0.2279, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 1.92489744398864, | |
| "grad_norm": 2.303067922592163, | |
| "learning_rate": 4.333333333333334e-06, | |
| "loss": 0.2158, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 1.9406752918901862, | |
| "grad_norm": 2.6614551544189453, | |
| "learning_rate": 4.277777777777778e-06, | |
| "loss": 0.21, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 1.9564531397917324, | |
| "grad_norm": 2.401498794555664, | |
| "learning_rate": 4.222222222222223e-06, | |
| "loss": 0.2301, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 1.9722309876932786, | |
| "grad_norm": 2.147493839263916, | |
| "learning_rate": 4.166666666666667e-06, | |
| "loss": 0.2281, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 1.9880088355948249, | |
| "grad_norm": 2.534454584121704, | |
| "learning_rate": 4.111111111111111e-06, | |
| "loss": 0.2176, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 2.003786683496371, | |
| "grad_norm": 1.805999994277954, | |
| "learning_rate": 4.055555555555556e-06, | |
| "loss": 0.2052, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 2.019564531397917, | |
| "grad_norm": 2.5037739276885986, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 0.1369, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 2.0353423792994634, | |
| "grad_norm": 2.1041407585144043, | |
| "learning_rate": 3.944444444444445e-06, | |
| "loss": 0.1223, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 2.0511202272010096, | |
| "grad_norm": 1.7618393898010254, | |
| "learning_rate": 3.88888888888889e-06, | |
| "loss": 0.1321, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 2.066898075102556, | |
| "grad_norm": 2.2356462478637695, | |
| "learning_rate": 3.833333333333334e-06, | |
| "loss": 0.1309, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 2.082675923004102, | |
| "grad_norm": 1.9169081449508667, | |
| "learning_rate": 3.777777777777778e-06, | |
| "loss": 0.1248, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 2.0984537709056483, | |
| "grad_norm": 2.0058977603912354, | |
| "learning_rate": 3.7222222222222225e-06, | |
| "loss": 0.1295, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 2.1142316188071946, | |
| "grad_norm": 1.9169594049453735, | |
| "learning_rate": 3.6666666666666666e-06, | |
| "loss": 0.1307, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 2.130009466708741, | |
| "grad_norm": 1.9053386449813843, | |
| "learning_rate": 3.6111111111111115e-06, | |
| "loss": 0.1171, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 2.145787314610287, | |
| "grad_norm": 2.123209238052368, | |
| "learning_rate": 3.555555555555556e-06, | |
| "loss": 0.1501, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 2.1615651625118333, | |
| "grad_norm": 2.103135347366333, | |
| "learning_rate": 3.5e-06, | |
| "loss": 0.1286, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 2.1773430104133795, | |
| "grad_norm": 1.901191234588623, | |
| "learning_rate": 3.444444444444445e-06, | |
| "loss": 0.1328, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 2.1931208583149258, | |
| "grad_norm": 1.752928376197815, | |
| "learning_rate": 3.3888888888888893e-06, | |
| "loss": 0.1476, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 2.208898706216472, | |
| "grad_norm": 2.233229160308838, | |
| "learning_rate": 3.3333333333333333e-06, | |
| "loss": 0.133, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 2.2246765541180182, | |
| "grad_norm": 1.8100041151046753, | |
| "learning_rate": 3.277777777777778e-06, | |
| "loss": 0.1401, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 2.2404544020195645, | |
| "grad_norm": 2.2920119762420654, | |
| "learning_rate": 3.2222222222222227e-06, | |
| "loss": 0.1339, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 2.2562322499211107, | |
| "grad_norm": 2.502577543258667, | |
| "learning_rate": 3.1666666666666667e-06, | |
| "loss": 0.1431, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 2.272010097822657, | |
| "grad_norm": 2.1777825355529785, | |
| "learning_rate": 3.1111111111111116e-06, | |
| "loss": 0.1287, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 2.287787945724203, | |
| "grad_norm": 2.158822536468506, | |
| "learning_rate": 3.055555555555556e-06, | |
| "loss": 0.1427, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 2.3035657936257494, | |
| "grad_norm": 2.028738260269165, | |
| "learning_rate": 3e-06, | |
| "loss": 0.1307, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 2.3193436415272957, | |
| "grad_norm": 2.283491849899292, | |
| "learning_rate": 2.944444444444445e-06, | |
| "loss": 0.1284, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 2.335121489428842, | |
| "grad_norm": 2.0819170475006104, | |
| "learning_rate": 2.888888888888889e-06, | |
| "loss": 0.1341, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 2.350899337330388, | |
| "grad_norm": 2.2217302322387695, | |
| "learning_rate": 2.8333333333333335e-06, | |
| "loss": 0.1398, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 2.3666771852319344, | |
| "grad_norm": 1.7553716897964478, | |
| "learning_rate": 2.7777777777777783e-06, | |
| "loss": 0.1284, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 2.3824550331334806, | |
| "grad_norm": 2.2240195274353027, | |
| "learning_rate": 2.7222222222222224e-06, | |
| "loss": 0.1282, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 2.398232881035027, | |
| "grad_norm": 1.973368525505066, | |
| "learning_rate": 2.666666666666667e-06, | |
| "loss": 0.1261, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 2.414010728936573, | |
| "grad_norm": 2.7670340538024902, | |
| "learning_rate": 2.6111111111111113e-06, | |
| "loss": 0.1249, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 2.4297885768381193, | |
| "grad_norm": 1.904808759689331, | |
| "learning_rate": 2.5555555555555557e-06, | |
| "loss": 0.1219, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 2.4455664247396656, | |
| "grad_norm": 2.0226497650146484, | |
| "learning_rate": 2.5e-06, | |
| "loss": 0.1282, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 2.461344272641212, | |
| "grad_norm": 2.153346300125122, | |
| "learning_rate": 2.4444444444444447e-06, | |
| "loss": 0.1274, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 2.477122120542758, | |
| "grad_norm": 2.0116922855377197, | |
| "learning_rate": 2.388888888888889e-06, | |
| "loss": 0.1342, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 2.4928999684443043, | |
| "grad_norm": 1.9735043048858643, | |
| "learning_rate": 2.3333333333333336e-06, | |
| "loss": 0.1328, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 2.5086778163458505, | |
| "grad_norm": 1.958715796470642, | |
| "learning_rate": 2.277777777777778e-06, | |
| "loss": 0.1175, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 2.5244556642473968, | |
| "grad_norm": 2.146303176879883, | |
| "learning_rate": 2.222222222222222e-06, | |
| "loss": 0.1308, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 2.5244556642473968, | |
| "eval_loss": 0.406938761472702, | |
| "eval_runtime": 2943.6463, | |
| "eval_samples_per_second": 2.385, | |
| "eval_steps_per_second": 0.149, | |
| "eval_wer": 0.2964035658770543, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 2.540233512148943, | |
| "grad_norm": 2.378037452697754, | |
| "learning_rate": 2.166666666666667e-06, | |
| "loss": 0.1315, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 2.5560113600504892, | |
| "grad_norm": 1.4860683679580688, | |
| "learning_rate": 2.1111111111111114e-06, | |
| "loss": 0.1256, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 2.5717892079520355, | |
| "grad_norm": 2.182999849319458, | |
| "learning_rate": 2.0555555555555555e-06, | |
| "loss": 0.1245, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 2.5875670558535817, | |
| "grad_norm": 2.181810140609741, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 0.1309, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 2.603344903755128, | |
| "grad_norm": 2.1204681396484375, | |
| "learning_rate": 1.944444444444445e-06, | |
| "loss": 0.1308, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 2.619122751656674, | |
| "grad_norm": 2.2567408084869385, | |
| "learning_rate": 1.888888888888889e-06, | |
| "loss": 0.1292, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 2.6349005995582204, | |
| "grad_norm": 2.3920841217041016, | |
| "learning_rate": 1.8355555555555557e-06, | |
| "loss": 0.1257, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 2.650678447459766, | |
| "grad_norm": 2.260571241378784, | |
| "learning_rate": 1.7800000000000001e-06, | |
| "loss": 0.1342, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 2.6664562953613125, | |
| "grad_norm": 2.4567618370056152, | |
| "learning_rate": 1.7244444444444448e-06, | |
| "loss": 0.1279, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 2.6822341432628587, | |
| "grad_norm": 2.1594390869140625, | |
| "learning_rate": 1.668888888888889e-06, | |
| "loss": 0.1253, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 2.698011991164405, | |
| "grad_norm": 2.029165267944336, | |
| "learning_rate": 1.6133333333333335e-06, | |
| "loss": 0.128, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 2.713789839065951, | |
| "grad_norm": 2.011826515197754, | |
| "learning_rate": 1.5577777777777777e-06, | |
| "loss": 0.1177, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 2.7295676869674974, | |
| "grad_norm": 2.081116199493408, | |
| "learning_rate": 1.5022222222222224e-06, | |
| "loss": 0.1322, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 2.7453455348690436, | |
| "grad_norm": 1.9539333581924438, | |
| "learning_rate": 1.4466666666666669e-06, | |
| "loss": 0.1225, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 2.76112338277059, | |
| "grad_norm": 1.83766770362854, | |
| "learning_rate": 1.3911111111111111e-06, | |
| "loss": 0.1218, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 2.776901230672136, | |
| "grad_norm": 1.8542882204055786, | |
| "learning_rate": 1.3355555555555558e-06, | |
| "loss": 0.1215, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 2.7926790785736824, | |
| "grad_norm": 1.944549798965454, | |
| "learning_rate": 1.28e-06, | |
| "loss": 0.1333, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 2.8084569264752286, | |
| "grad_norm": 1.9153203964233398, | |
| "learning_rate": 1.2244444444444445e-06, | |
| "loss": 0.1312, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 2.824234774376775, | |
| "grad_norm": 1.7963358163833618, | |
| "learning_rate": 1.168888888888889e-06, | |
| "loss": 0.1202, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 2.840012622278321, | |
| "grad_norm": 2.115541696548462, | |
| "learning_rate": 1.1133333333333334e-06, | |
| "loss": 0.1234, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 2.8557904701798673, | |
| "grad_norm": 2.2020208835601807, | |
| "learning_rate": 1.0577777777777779e-06, | |
| "loss": 0.1281, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 2.8715683180814136, | |
| "grad_norm": 2.03625226020813, | |
| "learning_rate": 1.0022222222222223e-06, | |
| "loss": 0.1216, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 2.88734616598296, | |
| "grad_norm": 2.0224132537841797, | |
| "learning_rate": 9.466666666666667e-07, | |
| "loss": 0.1163, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 2.903124013884506, | |
| "grad_norm": 2.030674695968628, | |
| "learning_rate": 8.911111111111112e-07, | |
| "loss": 0.1297, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 2.9189018617860523, | |
| "grad_norm": 1.9184988737106323, | |
| "learning_rate": 8.355555555555556e-07, | |
| "loss": 0.1243, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 2.9346797096875985, | |
| "grad_norm": 1.922864556312561, | |
| "learning_rate": 7.8e-07, | |
| "loss": 0.1175, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 2.9504575575891447, | |
| "grad_norm": 2.3562278747558594, | |
| "learning_rate": 7.244444444444446e-07, | |
| "loss": 0.1352, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 2.966235405490691, | |
| "grad_norm": 1.826926589012146, | |
| "learning_rate": 6.68888888888889e-07, | |
| "loss": 0.129, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 2.982013253392237, | |
| "grad_norm": 1.7427453994750977, | |
| "learning_rate": 6.133333333333333e-07, | |
| "loss": 0.1231, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 2.9977911012937835, | |
| "grad_norm": 1.7461098432540894, | |
| "learning_rate": 5.577777777777779e-07, | |
| "loss": 0.1175, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 3.0135689491953297, | |
| "grad_norm": 1.2887001037597656, | |
| "learning_rate": 5.022222222222222e-07, | |
| "loss": 0.0873, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 3.029346797096876, | |
| "grad_norm": 1.7019718885421753, | |
| "learning_rate": 4.466666666666667e-07, | |
| "loss": 0.0826, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 3.045124644998422, | |
| "grad_norm": 1.7305608987808228, | |
| "learning_rate": 3.9111111111111115e-07, | |
| "loss": 0.0788, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 3.0609024928999684, | |
| "grad_norm": 1.4667015075683594, | |
| "learning_rate": 3.3555555555555556e-07, | |
| "loss": 0.0775, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 3.0766803408015146, | |
| "grad_norm": 2.291264057159424, | |
| "learning_rate": 2.8e-07, | |
| "loss": 0.0802, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 3.092458188703061, | |
| "grad_norm": 1.8721388578414917, | |
| "learning_rate": 2.2444444444444445e-07, | |
| "loss": 0.0789, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 3.108236036604607, | |
| "grad_norm": 1.4734169244766235, | |
| "learning_rate": 1.6888888888888888e-07, | |
| "loss": 0.0919, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 3.1240138845061534, | |
| "grad_norm": 1.3516019582748413, | |
| "learning_rate": 1.1333333333333336e-07, | |
| "loss": 0.0775, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 3.1397917324076996, | |
| "grad_norm": 1.5201116800308228, | |
| "learning_rate": 5.777777777777778e-08, | |
| "loss": 0.0776, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 3.155569580309246, | |
| "grad_norm": 1.4460948705673218, | |
| "learning_rate": 2.2222222222222225e-09, | |
| "loss": 0.0804, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 3.155569580309246, | |
| "eval_loss": 0.4191373586654663, | |
| "eval_runtime": 2987.8843, | |
| "eval_samples_per_second": 2.35, | |
| "eval_steps_per_second": 0.147, | |
| "eval_wer": 0.2910627217048187, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 3.155569580309246, | |
| "step": 5000, | |
| "total_flos": 5.4352838158516224e+20, | |
| "train_loss": 0.2719093270301819, | |
| "train_runtime": 63684.9216, | |
| "train_samples_per_second": 2.512, | |
| "train_steps_per_second": 0.079 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 5000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 4, | |
| "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": 5.4352838158516224e+20, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |