| { | |
| "best_global_step": 3000, | |
| "best_metric": 0.6518804981477404, | |
| "best_model_checkpoint": "./SALAMA_C5/checkpoint-3000", | |
| "epoch": 2.0422055820285907, | |
| "eval_steps": 1000, | |
| "global_step": 3000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.013614703880190605, | |
| "grad_norm": 5.017816066741943, | |
| "learning_rate": 3.8e-07, | |
| "loss": 0.0279, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.02722940776038121, | |
| "grad_norm": 3.6539018154144287, | |
| "learning_rate": 7.8e-07, | |
| "loss": 0.0334, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.04084411164057182, | |
| "grad_norm": 2.4141480922698975, | |
| "learning_rate": 1.1800000000000001e-06, | |
| "loss": 0.0266, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.05445881552076242, | |
| "grad_norm": 2.1062567234039307, | |
| "learning_rate": 1.5800000000000001e-06, | |
| "loss": 0.0197, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.06807351940095303, | |
| "grad_norm": 4.6255106925964355, | |
| "learning_rate": 1.98e-06, | |
| "loss": 0.0279, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.08168822328114364, | |
| "grad_norm": 1.933724284172058, | |
| "learning_rate": 2.38e-06, | |
| "loss": 0.0261, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.09530292716133425, | |
| "grad_norm": 2.3127458095550537, | |
| "learning_rate": 2.7800000000000005e-06, | |
| "loss": 0.0261, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.10891763104152484, | |
| "grad_norm": 2.6971325874328613, | |
| "learning_rate": 3.1800000000000005e-06, | |
| "loss": 0.0186, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.12253233492171545, | |
| "grad_norm": 3.273725986480713, | |
| "learning_rate": 3.58e-06, | |
| "loss": 0.0278, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.13614703880190607, | |
| "grad_norm": 2.669668197631836, | |
| "learning_rate": 3.980000000000001e-06, | |
| "loss": 0.0308, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.14976174268209666, | |
| "grad_norm": 0.6969190239906311, | |
| "learning_rate": 4.38e-06, | |
| "loss": 0.0271, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.16337644656228728, | |
| "grad_norm": 2.6846354007720947, | |
| "learning_rate": 4.78e-06, | |
| "loss": 0.0298, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.17699115044247787, | |
| "grad_norm": 3.2715647220611572, | |
| "learning_rate": 5.18e-06, | |
| "loss": 0.0223, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.1906058543226685, | |
| "grad_norm": 3.50703763961792, | |
| "learning_rate": 5.580000000000001e-06, | |
| "loss": 0.0273, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.2042205582028591, | |
| "grad_norm": 1.7984274625778198, | |
| "learning_rate": 5.98e-06, | |
| "loss": 0.0305, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.21783526208304968, | |
| "grad_norm": 2.7656664848327637, | |
| "learning_rate": 6.380000000000001e-06, | |
| "loss": 0.0484, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.2314499659632403, | |
| "grad_norm": 0.7585025429725647, | |
| "learning_rate": 6.780000000000001e-06, | |
| "loss": 0.0213, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.2450646698434309, | |
| "grad_norm": 4.038341522216797, | |
| "learning_rate": 7.180000000000001e-06, | |
| "loss": 0.0333, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.2586793737236215, | |
| "grad_norm": 2.8464272022247314, | |
| "learning_rate": 7.58e-06, | |
| "loss": 0.0269, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.27229407760381213, | |
| "grad_norm": 3.064093589782715, | |
| "learning_rate": 7.980000000000002e-06, | |
| "loss": 0.0169, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.2859087814840027, | |
| "grad_norm": 5.7664971351623535, | |
| "learning_rate": 8.380000000000001e-06, | |
| "loss": 0.0275, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.2995234853641933, | |
| "grad_norm": 1.4587912559509277, | |
| "learning_rate": 8.78e-06, | |
| "loss": 0.0258, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.3131381892443839, | |
| "grad_norm": 2.867338180541992, | |
| "learning_rate": 9.180000000000002e-06, | |
| "loss": 0.0279, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.32675289312457456, | |
| "grad_norm": 4.9567484855651855, | |
| "learning_rate": 9.58e-06, | |
| "loss": 0.0301, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.34036759700476515, | |
| "grad_norm": 4.536691665649414, | |
| "learning_rate": 9.980000000000001e-06, | |
| "loss": 0.0329, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.35398230088495575, | |
| "grad_norm": 2.4109556674957275, | |
| "learning_rate": 9.95136933708728e-06, | |
| "loss": 0.0351, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.36759700476514634, | |
| "grad_norm": 3.561450958251953, | |
| "learning_rate": 9.900179165600206e-06, | |
| "loss": 0.0368, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.381211708645337, | |
| "grad_norm": 1.7630550861358643, | |
| "learning_rate": 9.848988994113131e-06, | |
| "loss": 0.0388, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.3948264125255276, | |
| "grad_norm": 1.5144799947738647, | |
| "learning_rate": 9.797798822626056e-06, | |
| "loss": 0.0304, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.4084411164057182, | |
| "grad_norm": 4.403763771057129, | |
| "learning_rate": 9.746608651138983e-06, | |
| "loss": 0.033, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.42205582028590877, | |
| "grad_norm": 3.6199100017547607, | |
| "learning_rate": 9.695418479651908e-06, | |
| "loss": 0.0377, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.43567052416609936, | |
| "grad_norm": 3.1659555435180664, | |
| "learning_rate": 9.644228308164833e-06, | |
| "loss": 0.0381, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.44928522804629, | |
| "grad_norm": 4.508129596710205, | |
| "learning_rate": 9.59303813667776e-06, | |
| "loss": 0.0277, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.4628999319264806, | |
| "grad_norm": 3.981079578399658, | |
| "learning_rate": 9.541847965190683e-06, | |
| "loss": 0.0297, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.4765146358066712, | |
| "grad_norm": 2.7824220657348633, | |
| "learning_rate": 9.49065779370361e-06, | |
| "loss": 0.0398, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.4901293396868618, | |
| "grad_norm": 5.651052951812744, | |
| "learning_rate": 9.439467622216535e-06, | |
| "loss": 0.0448, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.5037440435670524, | |
| "grad_norm": 4.3824920654296875, | |
| "learning_rate": 9.38827745072946e-06, | |
| "loss": 0.0314, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.517358747447243, | |
| "grad_norm": 4.516641139984131, | |
| "learning_rate": 9.337087279242385e-06, | |
| "loss": 0.0437, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.5309734513274337, | |
| "grad_norm": 3.0896172523498535, | |
| "learning_rate": 9.285897107755312e-06, | |
| "loss": 0.0389, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.5445881552076243, | |
| "grad_norm": 2.9489505290985107, | |
| "learning_rate": 9.234706936268237e-06, | |
| "loss": 0.036, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.5582028590878149, | |
| "grad_norm": 1.6065051555633545, | |
| "learning_rate": 9.183516764781162e-06, | |
| "loss": 0.0431, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.5718175629680055, | |
| "grad_norm": 2.8498787879943848, | |
| "learning_rate": 9.132326593294089e-06, | |
| "loss": 0.035, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.585432266848196, | |
| "grad_norm": 1.6527795791625977, | |
| "learning_rate": 9.081136421807014e-06, | |
| "loss": 0.0351, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.5990469707283866, | |
| "grad_norm": 3.358445405960083, | |
| "learning_rate": 9.02994625031994e-06, | |
| "loss": 0.0287, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.6126616746085772, | |
| "grad_norm": 3.706794500350952, | |
| "learning_rate": 8.978756078832866e-06, | |
| "loss": 0.0413, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.6262763784887678, | |
| "grad_norm": 3.2217488288879395, | |
| "learning_rate": 8.92756590734579e-06, | |
| "loss": 0.0335, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.6398910823689585, | |
| "grad_norm": 4.742300987243652, | |
| "learning_rate": 8.876375735858716e-06, | |
| "loss": 0.0365, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.6535057862491491, | |
| "grad_norm": 2.290529727935791, | |
| "learning_rate": 8.825185564371641e-06, | |
| "loss": 0.0288, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.6671204901293397, | |
| "grad_norm": 1.5146220922470093, | |
| "learning_rate": 8.773995392884566e-06, | |
| "loss": 0.0379, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.6807351940095303, | |
| "grad_norm": 5.127984046936035, | |
| "learning_rate": 8.722805221397493e-06, | |
| "loss": 0.0256, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.6807351940095303, | |
| "eval_loss": 0.02619771473109722, | |
| "eval_runtime": 4340.6262, | |
| "eval_samples_per_second": 2.707, | |
| "eval_steps_per_second": 0.338, | |
| "eval_wer": 2.0130758995663705, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.6943498978897209, | |
| "grad_norm": 3.8511838912963867, | |
| "learning_rate": 8.671615049910418e-06, | |
| "loss": 0.046, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.7079646017699115, | |
| "grad_norm": 3.2079453468322754, | |
| "learning_rate": 8.620424878423343e-06, | |
| "loss": 0.0366, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.7215793056501021, | |
| "grad_norm": 5.992726802825928, | |
| "learning_rate": 8.56923470693627e-06, | |
| "loss": 0.0507, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.7351940095302927, | |
| "grad_norm": 2.984426498413086, | |
| "learning_rate": 8.518044535449195e-06, | |
| "loss": 0.0364, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.7488087134104833, | |
| "grad_norm": 2.254213333129883, | |
| "learning_rate": 8.46685436396212e-06, | |
| "loss": 0.0345, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.762423417290674, | |
| "grad_norm": 3.1881701946258545, | |
| "learning_rate": 8.415664192475045e-06, | |
| "loss": 0.047, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.7760381211708646, | |
| "grad_norm": 2.6365621089935303, | |
| "learning_rate": 8.36447402098797e-06, | |
| "loss": 0.0299, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.7896528250510552, | |
| "grad_norm": 2.4238040447235107, | |
| "learning_rate": 8.313283849500895e-06, | |
| "loss": 0.0327, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.8032675289312458, | |
| "grad_norm": 7.073357582092285, | |
| "learning_rate": 8.262093678013822e-06, | |
| "loss": 0.038, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.8168822328114363, | |
| "grad_norm": 2.8962574005126953, | |
| "learning_rate": 8.210903506526747e-06, | |
| "loss": 0.0319, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.8304969366916269, | |
| "grad_norm": 4.383892059326172, | |
| "learning_rate": 8.159713335039672e-06, | |
| "loss": 0.0443, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.8441116405718175, | |
| "grad_norm": 2.0938570499420166, | |
| "learning_rate": 8.108523163552599e-06, | |
| "loss": 0.0424, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.8577263444520081, | |
| "grad_norm": 2.2466468811035156, | |
| "learning_rate": 8.057332992065524e-06, | |
| "loss": 0.0382, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.8713410483321987, | |
| "grad_norm": 3.2445216178894043, | |
| "learning_rate": 8.00614282057845e-06, | |
| "loss": 0.0369, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.8849557522123894, | |
| "grad_norm": 3.417433500289917, | |
| "learning_rate": 7.954952649091376e-06, | |
| "loss": 0.0374, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.89857045609258, | |
| "grad_norm": 4.187256336212158, | |
| "learning_rate": 7.903762477604301e-06, | |
| "loss": 0.0385, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.9121851599727706, | |
| "grad_norm": 2.0254714488983154, | |
| "learning_rate": 7.852572306117226e-06, | |
| "loss": 0.0441, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 0.9257998638529612, | |
| "grad_norm": 4.565765857696533, | |
| "learning_rate": 7.801382134630153e-06, | |
| "loss": 0.0352, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 0.9394145677331518, | |
| "grad_norm": 2.339277982711792, | |
| "learning_rate": 7.750191963143076e-06, | |
| "loss": 0.0368, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 0.9530292716133424, | |
| "grad_norm": 1.5078526735305786, | |
| "learning_rate": 7.699001791656003e-06, | |
| "loss": 0.0368, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.966643975493533, | |
| "grad_norm": 2.621042251586914, | |
| "learning_rate": 7.647811620168928e-06, | |
| "loss": 0.0319, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 0.9802586793737236, | |
| "grad_norm": 3.816051483154297, | |
| "learning_rate": 7.596621448681853e-06, | |
| "loss": 0.0336, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 0.9938733832539143, | |
| "grad_norm": 3.447195291519165, | |
| "learning_rate": 7.545431277194779e-06, | |
| "loss": 0.0321, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 1.0074880871341048, | |
| "grad_norm": 1.2242047786712646, | |
| "learning_rate": 7.494241105707705e-06, | |
| "loss": 0.0226, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 1.0211027910142954, | |
| "grad_norm": 2.723759174346924, | |
| "learning_rate": 7.44305093422063e-06, | |
| "loss": 0.0124, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.034717494894486, | |
| "grad_norm": 0.75479656457901, | |
| "learning_rate": 7.391860762733556e-06, | |
| "loss": 0.0126, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 1.0483321987746765, | |
| "grad_norm": 1.7836439609527588, | |
| "learning_rate": 7.340670591246481e-06, | |
| "loss": 0.0134, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 1.0619469026548674, | |
| "grad_norm": 1.289770483970642, | |
| "learning_rate": 7.289480419759407e-06, | |
| "loss": 0.0155, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 1.075561606535058, | |
| "grad_norm": 1.1430002450942993, | |
| "learning_rate": 7.238290248272333e-06, | |
| "loss": 0.0168, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 1.0891763104152485, | |
| "grad_norm": 1.9707790613174438, | |
| "learning_rate": 7.187100076785258e-06, | |
| "loss": 0.0194, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 1.1027910142954391, | |
| "grad_norm": 1.5533260107040405, | |
| "learning_rate": 7.135909905298183e-06, | |
| "loss": 0.0151, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 1.1164057181756297, | |
| "grad_norm": 1.5283340215682983, | |
| "learning_rate": 7.084719733811108e-06, | |
| "loss": 0.0124, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 1.1300204220558203, | |
| "grad_norm": 2.581458330154419, | |
| "learning_rate": 7.033529562324034e-06, | |
| "loss": 0.0167, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 1.143635125936011, | |
| "grad_norm": 2.616727352142334, | |
| "learning_rate": 6.98233939083696e-06, | |
| "loss": 0.0193, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 1.1572498298162015, | |
| "grad_norm": 1.738170862197876, | |
| "learning_rate": 6.931149219349885e-06, | |
| "loss": 0.0123, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 1.170864533696392, | |
| "grad_norm": 1.8732948303222656, | |
| "learning_rate": 6.879959047862811e-06, | |
| "loss": 0.0139, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 1.1844792375765827, | |
| "grad_norm": 2.220024585723877, | |
| "learning_rate": 6.828768876375736e-06, | |
| "loss": 0.0115, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 1.1980939414567733, | |
| "grad_norm": 1.57091224193573, | |
| "learning_rate": 6.777578704888662e-06, | |
| "loss": 0.0162, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 1.2117086453369639, | |
| "grad_norm": 3.6890385150909424, | |
| "learning_rate": 6.726388533401588e-06, | |
| "loss": 0.0218, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 1.2253233492171545, | |
| "grad_norm": 3.3784091472625732, | |
| "learning_rate": 6.675198361914513e-06, | |
| "loss": 0.0176, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 1.238938053097345, | |
| "grad_norm": 2.2989323139190674, | |
| "learning_rate": 6.624008190427439e-06, | |
| "loss": 0.0113, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 1.2525527569775359, | |
| "grad_norm": 3.2158448696136475, | |
| "learning_rate": 6.572818018940365e-06, | |
| "loss": 0.0158, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 1.2661674608577265, | |
| "grad_norm": 2.4740653038024902, | |
| "learning_rate": 6.521627847453289e-06, | |
| "loss": 0.013, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 1.279782164737917, | |
| "grad_norm": 1.7962628602981567, | |
| "learning_rate": 6.470437675966215e-06, | |
| "loss": 0.0159, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 1.2933968686181077, | |
| "grad_norm": 3.0249545574188232, | |
| "learning_rate": 6.41924750447914e-06, | |
| "loss": 0.0135, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 1.3070115724982982, | |
| "grad_norm": 0.7751892805099487, | |
| "learning_rate": 6.368057332992066e-06, | |
| "loss": 0.016, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 1.3206262763784888, | |
| "grad_norm": 2.894857168197632, | |
| "learning_rate": 6.316867161504991e-06, | |
| "loss": 0.0128, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 1.3342409802586794, | |
| "grad_norm": 2.7285149097442627, | |
| "learning_rate": 6.265676990017917e-06, | |
| "loss": 0.0166, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 1.34785568413887, | |
| "grad_norm": 0.9751483201980591, | |
| "learning_rate": 6.214486818530843e-06, | |
| "loss": 0.0147, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 1.3614703880190606, | |
| "grad_norm": 1.6021450757980347, | |
| "learning_rate": 6.163296647043768e-06, | |
| "loss": 0.0151, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.3614703880190606, | |
| "eval_loss": 0.015386924147605896, | |
| "eval_runtime": 4340.0308, | |
| "eval_samples_per_second": 2.707, | |
| "eval_steps_per_second": 0.338, | |
| "eval_wer": 1.172619104597624, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.3750850918992512, | |
| "grad_norm": 2.4471325874328613, | |
| "learning_rate": 6.112106475556694e-06, | |
| "loss": 0.013, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 1.3886997957794418, | |
| "grad_norm": 2.017465353012085, | |
| "learning_rate": 6.06091630406962e-06, | |
| "loss": 0.0148, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 1.4023144996596324, | |
| "grad_norm": 1.9365452527999878, | |
| "learning_rate": 6.009726132582545e-06, | |
| "loss": 0.0137, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 1.415929203539823, | |
| "grad_norm": 2.5946829319000244, | |
| "learning_rate": 5.958535961095469e-06, | |
| "loss": 0.0181, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 1.4295439074200136, | |
| "grad_norm": 1.8529382944107056, | |
| "learning_rate": 5.907345789608395e-06, | |
| "loss": 0.0119, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 1.4431586113002042, | |
| "grad_norm": 2.656736135482788, | |
| "learning_rate": 5.856155618121321e-06, | |
| "loss": 0.022, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 1.4567733151803948, | |
| "grad_norm": 3.045898914337158, | |
| "learning_rate": 5.804965446634246e-06, | |
| "loss": 0.0141, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 1.4703880190605854, | |
| "grad_norm": 1.8601387739181519, | |
| "learning_rate": 5.753775275147172e-06, | |
| "loss": 0.0152, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 1.484002722940776, | |
| "grad_norm": 3.429184675216675, | |
| "learning_rate": 5.702585103660098e-06, | |
| "loss": 0.0147, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 1.4976174268209665, | |
| "grad_norm": 2.976245880126953, | |
| "learning_rate": 5.651394932173023e-06, | |
| "loss": 0.0163, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 1.5112321307011571, | |
| "grad_norm": 1.0114331245422363, | |
| "learning_rate": 5.600204760685949e-06, | |
| "loss": 0.012, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 1.5248468345813477, | |
| "grad_norm": 1.8622580766677856, | |
| "learning_rate": 5.549014589198875e-06, | |
| "loss": 0.0101, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 1.5384615384615383, | |
| "grad_norm": 1.6890689134597778, | |
| "learning_rate": 5.4978244177118e-06, | |
| "loss": 0.0139, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 1.552076242341729, | |
| "grad_norm": 2.004462718963623, | |
| "learning_rate": 5.446634246224726e-06, | |
| "loss": 0.0109, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 1.5656909462219195, | |
| "grad_norm": 1.9925123453140259, | |
| "learning_rate": 5.395444074737651e-06, | |
| "loss": 0.0142, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 1.5793056501021103, | |
| "grad_norm": 1.5560657978057861, | |
| "learning_rate": 5.344253903250576e-06, | |
| "loss": 0.0129, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 1.592920353982301, | |
| "grad_norm": 1.2889870405197144, | |
| "learning_rate": 5.293063731763501e-06, | |
| "loss": 0.0128, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 1.6065350578624915, | |
| "grad_norm": 2.232189416885376, | |
| "learning_rate": 5.241873560276427e-06, | |
| "loss": 0.0129, | |
| "step": 2360 | |
| }, | |
| { | |
| "epoch": 1.620149761742682, | |
| "grad_norm": 0.7273709774017334, | |
| "learning_rate": 5.190683388789353e-06, | |
| "loss": 0.0133, | |
| "step": 2380 | |
| }, | |
| { | |
| "epoch": 1.6337644656228727, | |
| "grad_norm": 1.8503831624984741, | |
| "learning_rate": 5.139493217302278e-06, | |
| "loss": 0.0138, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 1.6473791695030633, | |
| "grad_norm": 1.786586880683899, | |
| "learning_rate": 5.088303045815204e-06, | |
| "loss": 0.0152, | |
| "step": 2420 | |
| }, | |
| { | |
| "epoch": 1.6609938733832539, | |
| "grad_norm": 2.061833381652832, | |
| "learning_rate": 5.037112874328129e-06, | |
| "loss": 0.0141, | |
| "step": 2440 | |
| }, | |
| { | |
| "epoch": 1.6746085772634445, | |
| "grad_norm": 1.8314088582992554, | |
| "learning_rate": 4.985922702841055e-06, | |
| "loss": 0.0124, | |
| "step": 2460 | |
| }, | |
| { | |
| "epoch": 1.6882232811436353, | |
| "grad_norm": 2.991048812866211, | |
| "learning_rate": 4.934732531353981e-06, | |
| "loss": 0.0132, | |
| "step": 2480 | |
| }, | |
| { | |
| "epoch": 1.7018379850238259, | |
| "grad_norm": 1.280178189277649, | |
| "learning_rate": 4.883542359866906e-06, | |
| "loss": 0.0136, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.7154526889040165, | |
| "grad_norm": 1.5085817575454712, | |
| "learning_rate": 4.832352188379831e-06, | |
| "loss": 0.015, | |
| "step": 2520 | |
| }, | |
| { | |
| "epoch": 1.729067392784207, | |
| "grad_norm": 4.28915548324585, | |
| "learning_rate": 4.781162016892757e-06, | |
| "loss": 0.0129, | |
| "step": 2540 | |
| }, | |
| { | |
| "epoch": 1.7426820966643977, | |
| "grad_norm": 2.7937703132629395, | |
| "learning_rate": 4.729971845405683e-06, | |
| "loss": 0.0158, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 1.7562968005445883, | |
| "grad_norm": 3.536447048187256, | |
| "learning_rate": 4.678781673918608e-06, | |
| "loss": 0.0155, | |
| "step": 2580 | |
| }, | |
| { | |
| "epoch": 1.7699115044247788, | |
| "grad_norm": 1.4017407894134521, | |
| "learning_rate": 4.627591502431534e-06, | |
| "loss": 0.0118, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 1.7835262083049694, | |
| "grad_norm": 1.5913890600204468, | |
| "learning_rate": 4.576401330944459e-06, | |
| "loss": 0.0096, | |
| "step": 2620 | |
| }, | |
| { | |
| "epoch": 1.79714091218516, | |
| "grad_norm": 0.7930202484130859, | |
| "learning_rate": 4.525211159457384e-06, | |
| "loss": 0.0116, | |
| "step": 2640 | |
| }, | |
| { | |
| "epoch": 1.8107556160653506, | |
| "grad_norm": 1.5710654258728027, | |
| "learning_rate": 4.47402098797031e-06, | |
| "loss": 0.0129, | |
| "step": 2660 | |
| }, | |
| { | |
| "epoch": 1.8243703199455412, | |
| "grad_norm": 0.65106201171875, | |
| "learning_rate": 4.422830816483236e-06, | |
| "loss": 0.0098, | |
| "step": 2680 | |
| }, | |
| { | |
| "epoch": 1.8379850238257318, | |
| "grad_norm": 2.7859442234039307, | |
| "learning_rate": 4.371640644996161e-06, | |
| "loss": 0.0173, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 1.8515997277059224, | |
| "grad_norm": 1.7729928493499756, | |
| "learning_rate": 4.320450473509087e-06, | |
| "loss": 0.0113, | |
| "step": 2720 | |
| }, | |
| { | |
| "epoch": 1.865214431586113, | |
| "grad_norm": 0.9945047497749329, | |
| "learning_rate": 4.269260302022012e-06, | |
| "loss": 0.0111, | |
| "step": 2740 | |
| }, | |
| { | |
| "epoch": 1.8788291354663036, | |
| "grad_norm": 2.976438283920288, | |
| "learning_rate": 4.218070130534938e-06, | |
| "loss": 0.0133, | |
| "step": 2760 | |
| }, | |
| { | |
| "epoch": 1.8924438393464942, | |
| "grad_norm": 2.28800106048584, | |
| "learning_rate": 4.166879959047863e-06, | |
| "loss": 0.0129, | |
| "step": 2780 | |
| }, | |
| { | |
| "epoch": 1.9060585432266848, | |
| "grad_norm": 0.837317943572998, | |
| "learning_rate": 4.115689787560789e-06, | |
| "loss": 0.0136, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 1.9196732471068754, | |
| "grad_norm": 1.1099382638931274, | |
| "learning_rate": 4.064499616073714e-06, | |
| "loss": 0.0139, | |
| "step": 2820 | |
| }, | |
| { | |
| "epoch": 1.933287950987066, | |
| "grad_norm": 1.3402103185653687, | |
| "learning_rate": 4.01330944458664e-06, | |
| "loss": 0.0144, | |
| "step": 2840 | |
| }, | |
| { | |
| "epoch": 1.9469026548672566, | |
| "grad_norm": 1.1998571157455444, | |
| "learning_rate": 3.962119273099565e-06, | |
| "loss": 0.0097, | |
| "step": 2860 | |
| }, | |
| { | |
| "epoch": 1.9605173587474471, | |
| "grad_norm": 1.3589485883712769, | |
| "learning_rate": 3.913488610186844e-06, | |
| "loss": 0.0136, | |
| "step": 2880 | |
| }, | |
| { | |
| "epoch": 1.9741320626276377, | |
| "grad_norm": 1.45253586769104, | |
| "learning_rate": 3.862298438699769e-06, | |
| "loss": 0.0108, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 1.9877467665078283, | |
| "grad_norm": 1.3955638408660889, | |
| "learning_rate": 3.8111082672126953e-06, | |
| "loss": 0.0102, | |
| "step": 2920 | |
| }, | |
| { | |
| "epoch": 2.001361470388019, | |
| "grad_norm": 0.19237089157104492, | |
| "learning_rate": 3.759918095725621e-06, | |
| "loss": 0.0134, | |
| "step": 2940 | |
| }, | |
| { | |
| "epoch": 2.0149761742682095, | |
| "grad_norm": 0.6027423143386841, | |
| "learning_rate": 3.7087279242385467e-06, | |
| "loss": 0.0045, | |
| "step": 2960 | |
| }, | |
| { | |
| "epoch": 2.0285908781484, | |
| "grad_norm": 0.41271352767944336, | |
| "learning_rate": 3.6575377527514722e-06, | |
| "loss": 0.0049, | |
| "step": 2980 | |
| }, | |
| { | |
| "epoch": 2.0422055820285907, | |
| "grad_norm": 0.31510305404663086, | |
| "learning_rate": 3.6063475812643977e-06, | |
| "loss": 0.004, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 2.0422055820285907, | |
| "eval_loss": 0.009533572010695934, | |
| "eval_runtime": 4348.4613, | |
| "eval_samples_per_second": 2.702, | |
| "eval_steps_per_second": 0.338, | |
| "eval_wer": 0.6518804981477404, | |
| "step": 3000 | |
| } | |
| ], | |
| "logging_steps": 20, | |
| "max_steps": 4407, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 1000, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 2.769784955191296e+19, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |