| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 10.030090270812437, | |
| "eval_steps": 1000, | |
| "global_step": 5000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.05015045135406219, | |
| "grad_norm": 7.0412750244140625, | |
| "learning_rate": 5.000000000000001e-07, | |
| "loss": 1.2267, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.10030090270812438, | |
| "grad_norm": 4.570739269256592, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 1.0287, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.15045135406218657, | |
| "grad_norm": 4.061439514160156, | |
| "learning_rate": 1.5e-06, | |
| "loss": 0.716, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.20060180541624875, | |
| "grad_norm": 3.5382726192474365, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 0.5682, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.25075225677031093, | |
| "grad_norm": 3.460435152053833, | |
| "learning_rate": 2.5e-06, | |
| "loss": 0.5493, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.30090270812437314, | |
| "grad_norm": 3.3355889320373535, | |
| "learning_rate": 3e-06, | |
| "loss": 0.4992, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.3510531594784353, | |
| "grad_norm": 3.1046106815338135, | |
| "learning_rate": 3.5e-06, | |
| "loss": 0.495, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.4012036108324975, | |
| "grad_norm": 3.444261074066162, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 0.4769, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.45135406218655966, | |
| "grad_norm": 3.0431270599365234, | |
| "learning_rate": 4.5e-06, | |
| "loss": 0.4349, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.5015045135406219, | |
| "grad_norm": 2.8672142028808594, | |
| "learning_rate": 5e-06, | |
| "loss": 0.4178, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.551654964894684, | |
| "grad_norm": 3.4275848865509033, | |
| "learning_rate": 5.500000000000001e-06, | |
| "loss": 0.417, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.6018054162487463, | |
| "grad_norm": 2.5542633533477783, | |
| "learning_rate": 6e-06, | |
| "loss": 0.409, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.6519558676028084, | |
| "grad_norm": 2.6586201190948486, | |
| "learning_rate": 6.5000000000000004e-06, | |
| "loss": 0.4022, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.7021063189568706, | |
| "grad_norm": 3.106419324874878, | |
| "learning_rate": 7e-06, | |
| "loss": 0.4231, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.7522567703109327, | |
| "grad_norm": 2.45819354057312, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 0.3943, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.802407221664995, | |
| "grad_norm": 2.844208002090454, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 0.3942, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.8525576730190572, | |
| "grad_norm": 2.8705875873565674, | |
| "learning_rate": 8.5e-06, | |
| "loss": 0.3947, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.9027081243731193, | |
| "grad_norm": 2.5415868759155273, | |
| "learning_rate": 9e-06, | |
| "loss": 0.3928, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.9528585757271816, | |
| "grad_norm": 3.9096789360046387, | |
| "learning_rate": 9.5e-06, | |
| "loss": 0.3802, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 1.0030090270812437, | |
| "grad_norm": 2.4224791526794434, | |
| "learning_rate": 1e-05, | |
| "loss": 0.3726, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.053159478435306, | |
| "grad_norm": 3.116126298904419, | |
| "learning_rate": 9.944444444444445e-06, | |
| "loss": 0.282, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 1.103309929789368, | |
| "grad_norm": 3.0114049911499023, | |
| "learning_rate": 9.88888888888889e-06, | |
| "loss": 0.3006, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.1534603811434303, | |
| "grad_norm": 2.352905035018921, | |
| "learning_rate": 9.833333333333333e-06, | |
| "loss": 0.2841, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 1.2036108324974926, | |
| "grad_norm": 2.395094871520996, | |
| "learning_rate": 9.777777777777779e-06, | |
| "loss": 0.2905, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.2537612838515546, | |
| "grad_norm": 2.6098897457122803, | |
| "learning_rate": 9.722222222222223e-06, | |
| "loss": 0.276, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 1.3039117352056169, | |
| "grad_norm": 2.0559678077697754, | |
| "learning_rate": 9.666666666666667e-06, | |
| "loss": 0.2741, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.3540621865596791, | |
| "grad_norm": 2.6575767993927, | |
| "learning_rate": 9.611111111111112e-06, | |
| "loss": 0.2903, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 1.4042126379137412, | |
| "grad_norm": 2.2294228076934814, | |
| "learning_rate": 9.555555555555556e-06, | |
| "loss": 0.2818, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.4543630892678034, | |
| "grad_norm": 2.2873806953430176, | |
| "learning_rate": 9.5e-06, | |
| "loss": 0.2778, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 1.5045135406218657, | |
| "grad_norm": 3.035785436630249, | |
| "learning_rate": 9.444444444444445e-06, | |
| "loss": 0.2958, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.5546639919759278, | |
| "grad_norm": 2.3122036457061768, | |
| "learning_rate": 9.38888888888889e-06, | |
| "loss": 0.2719, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 1.60481444332999, | |
| "grad_norm": 2.1870856285095215, | |
| "learning_rate": 9.333333333333334e-06, | |
| "loss": 0.2696, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.6549648946840523, | |
| "grad_norm": 2.747004270553589, | |
| "learning_rate": 9.277777777777778e-06, | |
| "loss": 0.276, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 1.7051153460381143, | |
| "grad_norm": 2.084136486053467, | |
| "learning_rate": 9.222222222222224e-06, | |
| "loss": 0.2979, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.7552657973921764, | |
| "grad_norm": 2.6127214431762695, | |
| "learning_rate": 9.166666666666666e-06, | |
| "loss": 0.2824, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 1.8054162487462388, | |
| "grad_norm": 2.4067749977111816, | |
| "learning_rate": 9.111111111111112e-06, | |
| "loss": 0.2671, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.855566700100301, | |
| "grad_norm": 2.577610731124878, | |
| "learning_rate": 9.055555555555556e-06, | |
| "loss": 0.2744, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 1.905717151454363, | |
| "grad_norm": 2.2848598957061768, | |
| "learning_rate": 9e-06, | |
| "loss": 0.2658, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.9558676028084254, | |
| "grad_norm": 2.707517385482788, | |
| "learning_rate": 8.944444444444446e-06, | |
| "loss": 0.2824, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 2.0060180541624875, | |
| "grad_norm": 1.714905023574829, | |
| "learning_rate": 8.891111111111111e-06, | |
| "loss": 0.2506, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.0060180541624875, | |
| "eval_loss": 0.34494394063949585, | |
| "eval_runtime": 4608.7156, | |
| "eval_samples_per_second": 2.335, | |
| "eval_steps_per_second": 0.146, | |
| "eval_wer": 0.21134514589689432, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.0561685055165495, | |
| "grad_norm": 1.9843376874923706, | |
| "learning_rate": 8.835555555555557e-06, | |
| "loss": 0.1393, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 2.106318956870612, | |
| "grad_norm": 2.0423927307128906, | |
| "learning_rate": 8.78e-06, | |
| "loss": 0.1513, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 2.156469408224674, | |
| "grad_norm": 2.1197094917297363, | |
| "learning_rate": 8.724444444444445e-06, | |
| "loss": 0.1454, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 2.206619859578736, | |
| "grad_norm": 1.8990999460220337, | |
| "learning_rate": 8.66888888888889e-06, | |
| "loss": 0.1484, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 2.2567703109327986, | |
| "grad_norm": 1.6921278238296509, | |
| "learning_rate": 8.613333333333333e-06, | |
| "loss": 0.1461, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 2.3069207622868606, | |
| "grad_norm": 1.9527013301849365, | |
| "learning_rate": 8.557777777777778e-06, | |
| "loss": 0.1557, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 2.3570712136409226, | |
| "grad_norm": 2.20961332321167, | |
| "learning_rate": 8.502222222222223e-06, | |
| "loss": 0.1459, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 2.407221664994985, | |
| "grad_norm": 2.1781654357910156, | |
| "learning_rate": 8.446666666666668e-06, | |
| "loss": 0.156, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 2.457372116349047, | |
| "grad_norm": 2.2167611122131348, | |
| "learning_rate": 8.391111111111112e-06, | |
| "loss": 0.1527, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 2.507522567703109, | |
| "grad_norm": 2.0940988063812256, | |
| "learning_rate": 8.335555555555556e-06, | |
| "loss": 0.1535, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 2.5576730190571713, | |
| "grad_norm": 2.141779899597168, | |
| "learning_rate": 8.28e-06, | |
| "loss": 0.15, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 2.6078234704112337, | |
| "grad_norm": 1.9819048643112183, | |
| "learning_rate": 8.224444444444444e-06, | |
| "loss": 0.1483, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 2.657973921765296, | |
| "grad_norm": 1.7420545816421509, | |
| "learning_rate": 8.16888888888889e-06, | |
| "loss": 0.1487, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 2.7081243731193583, | |
| "grad_norm": 2.012843608856201, | |
| "learning_rate": 8.113333333333334e-06, | |
| "loss": 0.1549, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 2.7582748244734203, | |
| "grad_norm": 2.0659139156341553, | |
| "learning_rate": 8.057777777777778e-06, | |
| "loss": 0.1515, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 2.8084252758274824, | |
| "grad_norm": 2.046651840209961, | |
| "learning_rate": 8.002222222222222e-06, | |
| "loss": 0.1414, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 2.8585757271815444, | |
| "grad_norm": 2.0306413173675537, | |
| "learning_rate": 7.946666666666666e-06, | |
| "loss": 0.1486, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 2.908726178535607, | |
| "grad_norm": 2.124835729598999, | |
| "learning_rate": 7.891111111111112e-06, | |
| "loss": 0.1598, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 2.958876629889669, | |
| "grad_norm": 2.09136700630188, | |
| "learning_rate": 7.835555555555556e-06, | |
| "loss": 0.1507, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 3.0090270812437314, | |
| "grad_norm": 1.2488484382629395, | |
| "learning_rate": 7.78e-06, | |
| "loss": 0.1431, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 3.0591775325977935, | |
| "grad_norm": 1.3443397283554077, | |
| "learning_rate": 7.724444444444446e-06, | |
| "loss": 0.0791, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 3.1093279839518555, | |
| "grad_norm": 1.274675726890564, | |
| "learning_rate": 7.66888888888889e-06, | |
| "loss": 0.0807, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 3.1594784353059175, | |
| "grad_norm": 1.38137948513031, | |
| "learning_rate": 7.613333333333334e-06, | |
| "loss": 0.0814, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 3.20962888665998, | |
| "grad_norm": 1.5349384546279907, | |
| "learning_rate": 7.557777777777779e-06, | |
| "loss": 0.0792, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 3.259779338014042, | |
| "grad_norm": 1.2551178932189941, | |
| "learning_rate": 7.502222222222223e-06, | |
| "loss": 0.0786, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 3.3099297893681046, | |
| "grad_norm": 1.400694489479065, | |
| "learning_rate": 7.446666666666668e-06, | |
| "loss": 0.0827, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 3.3600802407221666, | |
| "grad_norm": 1.382521390914917, | |
| "learning_rate": 7.3911111111111125e-06, | |
| "loss": 0.0814, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 3.4102306920762286, | |
| "grad_norm": 1.4124006032943726, | |
| "learning_rate": 7.335555555555556e-06, | |
| "loss": 0.087, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 3.4603811434302907, | |
| "grad_norm": 1.130927562713623, | |
| "learning_rate": 7.280000000000001e-06, | |
| "loss": 0.0911, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 3.510531594784353, | |
| "grad_norm": 1.6954821348190308, | |
| "learning_rate": 7.224444444444445e-06, | |
| "loss": 0.0856, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 3.560682046138415, | |
| "grad_norm": 1.1873488426208496, | |
| "learning_rate": 7.1688888888888895e-06, | |
| "loss": 0.0829, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 3.6108324974924777, | |
| "grad_norm": 1.500501275062561, | |
| "learning_rate": 7.113333333333334e-06, | |
| "loss": 0.0828, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 3.6609829488465397, | |
| "grad_norm": 1.2689720392227173, | |
| "learning_rate": 7.057777777777778e-06, | |
| "loss": 0.0828, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 3.711133400200602, | |
| "grad_norm": 1.4504294395446777, | |
| "learning_rate": 7.0022222222222225e-06, | |
| "loss": 0.0875, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 3.761283851554664, | |
| "grad_norm": 1.286098599433899, | |
| "learning_rate": 6.946666666666667e-06, | |
| "loss": 0.082, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 3.8114343029087263, | |
| "grad_norm": 1.1930783987045288, | |
| "learning_rate": 6.891111111111111e-06, | |
| "loss": 0.0817, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 3.8615847542627884, | |
| "grad_norm": 1.4965267181396484, | |
| "learning_rate": 6.835555555555556e-06, | |
| "loss": 0.0808, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 3.9117352056168504, | |
| "grad_norm": 1.3505349159240723, | |
| "learning_rate": 6.780000000000001e-06, | |
| "loss": 0.0847, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 3.961885656970913, | |
| "grad_norm": 2.0747885704040527, | |
| "learning_rate": 6.724444444444444e-06, | |
| "loss": 0.0846, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 4.012036108324975, | |
| "grad_norm": 0.9830118417739868, | |
| "learning_rate": 6.668888888888889e-06, | |
| "loss": 0.0783, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 4.012036108324975, | |
| "eval_loss": 0.3656960725784302, | |
| "eval_runtime": 4598.7916, | |
| "eval_samples_per_second": 2.34, | |
| "eval_steps_per_second": 0.146, | |
| "eval_wer": 0.2009700383321599, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 4.062186559679037, | |
| "grad_norm": 1.436523675918579, | |
| "learning_rate": 6.613333333333334e-06, | |
| "loss": 0.0531, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 4.112337011033099, | |
| "grad_norm": 0.6627805233001709, | |
| "learning_rate": 6.557777777777778e-06, | |
| "loss": 0.0505, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 4.162487462387162, | |
| "grad_norm": 1.2392983436584473, | |
| "learning_rate": 6.502222222222223e-06, | |
| "loss": 0.0548, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 4.212637913741224, | |
| "grad_norm": 1.0560728311538696, | |
| "learning_rate": 6.446666666666668e-06, | |
| "loss": 0.0492, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 4.262788365095286, | |
| "grad_norm": 1.5553325414657593, | |
| "learning_rate": 6.391111111111111e-06, | |
| "loss": 0.0519, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 4.312938816449348, | |
| "grad_norm": 1.320204257965088, | |
| "learning_rate": 6.335555555555556e-06, | |
| "loss": 0.0534, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 4.36308926780341, | |
| "grad_norm": 0.9548274278640747, | |
| "learning_rate": 6.280000000000001e-06, | |
| "loss": 0.0506, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 4.413239719157472, | |
| "grad_norm": 1.0882978439331055, | |
| "learning_rate": 6.224444444444445e-06, | |
| "loss": 0.0513, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 4.463390170511534, | |
| "grad_norm": 1.3011126518249512, | |
| "learning_rate": 6.16888888888889e-06, | |
| "loss": 0.052, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 4.513540621865597, | |
| "grad_norm": 0.8610795140266418, | |
| "learning_rate": 6.113333333333333e-06, | |
| "loss": 0.0502, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 4.563691073219659, | |
| "grad_norm": 1.2139874696731567, | |
| "learning_rate": 6.057777777777778e-06, | |
| "loss": 0.0526, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 4.613841524573721, | |
| "grad_norm": 0.9586848020553589, | |
| "learning_rate": 6.002222222222223e-06, | |
| "loss": 0.0523, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 4.663991975927783, | |
| "grad_norm": 0.9463987350463867, | |
| "learning_rate": 5.946666666666668e-06, | |
| "loss": 0.0534, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 4.714142427281845, | |
| "grad_norm": 1.0908218622207642, | |
| "learning_rate": 5.891111111111112e-06, | |
| "loss": 0.0522, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 4.764292878635907, | |
| "grad_norm": 1.063152551651001, | |
| "learning_rate": 5.8355555555555565e-06, | |
| "loss": 0.0507, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 4.81444332998997, | |
| "grad_norm": 0.8035433292388916, | |
| "learning_rate": 5.78e-06, | |
| "loss": 0.052, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 4.864593781344032, | |
| "grad_norm": 1.3851343393325806, | |
| "learning_rate": 5.724444444444445e-06, | |
| "loss": 0.0511, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 4.914744232698094, | |
| "grad_norm": 0.9622118473052979, | |
| "learning_rate": 5.6688888888888895e-06, | |
| "loss": 0.0538, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 4.964894684052156, | |
| "grad_norm": 0.7414006590843201, | |
| "learning_rate": 5.613333333333334e-06, | |
| "loss": 0.0501, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 5.015045135406218, | |
| "grad_norm": 0.7375747561454773, | |
| "learning_rate": 5.557777777777778e-06, | |
| "loss": 0.0489, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 5.0651955867602805, | |
| "grad_norm": 0.4028901755809784, | |
| "learning_rate": 5.5022222222222224e-06, | |
| "loss": 0.0413, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 5.115346038114343, | |
| "grad_norm": 0.49548035860061646, | |
| "learning_rate": 5.4466666666666665e-06, | |
| "loss": 0.041, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 5.1654964894684054, | |
| "grad_norm": 0.7137969732284546, | |
| "learning_rate": 5.391111111111111e-06, | |
| "loss": 0.0393, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 5.2156469408224675, | |
| "grad_norm": 1.015681266784668, | |
| "learning_rate": 5.335555555555556e-06, | |
| "loss": 0.0396, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 5.2657973921765295, | |
| "grad_norm": 1.030476450920105, | |
| "learning_rate": 5.28e-06, | |
| "loss": 0.0386, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 5.315947843530592, | |
| "grad_norm": 1.2187836170196533, | |
| "learning_rate": 5.224444444444445e-06, | |
| "loss": 0.0391, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 5.366098294884654, | |
| "grad_norm": 0.6549712419509888, | |
| "learning_rate": 5.168888888888889e-06, | |
| "loss": 0.0388, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 5.4162487462387165, | |
| "grad_norm": 0.6556330919265747, | |
| "learning_rate": 5.113333333333333e-06, | |
| "loss": 0.0399, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 5.466399197592779, | |
| "grad_norm": 1.0968986749649048, | |
| "learning_rate": 5.057777777777778e-06, | |
| "loss": 0.039, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 5.516549648946841, | |
| "grad_norm": 0.8567794561386108, | |
| "learning_rate": 5.002222222222223e-06, | |
| "loss": 0.0395, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 5.566700100300903, | |
| "grad_norm": 0.48124751448631287, | |
| "learning_rate": 4.946666666666667e-06, | |
| "loss": 0.0385, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 5.616850551654965, | |
| "grad_norm": 0.16856200993061066, | |
| "learning_rate": 4.891111111111111e-06, | |
| "loss": 0.0381, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 5.667001003009027, | |
| "grad_norm": 0.4240492284297943, | |
| "learning_rate": 4.835555555555556e-06, | |
| "loss": 0.0401, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 5.717151454363089, | |
| "grad_norm": 0.7855473756790161, | |
| "learning_rate": 4.78e-06, | |
| "loss": 0.0397, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 5.767301905717152, | |
| "grad_norm": 1.1043782234191895, | |
| "learning_rate": 4.724444444444445e-06, | |
| "loss": 0.0384, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 5.817452357071214, | |
| "grad_norm": 0.4021261930465698, | |
| "learning_rate": 4.66888888888889e-06, | |
| "loss": 0.0387, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 5.867602808425276, | |
| "grad_norm": 0.9103265404701233, | |
| "learning_rate": 4.613333333333334e-06, | |
| "loss": 0.0429, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 5.917753259779338, | |
| "grad_norm": 2.1955697536468506, | |
| "learning_rate": 4.557777777777778e-06, | |
| "loss": 0.0395, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 5.9679037111334, | |
| "grad_norm": 0.39817795157432556, | |
| "learning_rate": 4.502222222222223e-06, | |
| "loss": 0.0386, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 6.018054162487463, | |
| "grad_norm": 0.6511321067810059, | |
| "learning_rate": 4.446666666666667e-06, | |
| "loss": 0.0377, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 6.018054162487463, | |
| "eval_loss": 0.4252609610557556, | |
| "eval_runtime": 4620.2895, | |
| "eval_samples_per_second": 2.33, | |
| "eval_steps_per_second": 0.146, | |
| "eval_wer": 0.19864272862395368, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 6.068204613841525, | |
| "grad_norm": 0.8066951036453247, | |
| "learning_rate": 4.391111111111112e-06, | |
| "loss": 0.0351, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 6.118355065195587, | |
| "grad_norm": 1.054024338722229, | |
| "learning_rate": 4.3355555555555565e-06, | |
| "loss": 0.0364, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 6.168505516549649, | |
| "grad_norm": 0.4796682894229889, | |
| "learning_rate": 4.2800000000000005e-06, | |
| "loss": 0.0348, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 6.218655967903711, | |
| "grad_norm": 1.1482007503509521, | |
| "learning_rate": 4.2244444444444446e-06, | |
| "loss": 0.0344, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 6.268806419257773, | |
| "grad_norm": 0.5434701442718506, | |
| "learning_rate": 4.168888888888889e-06, | |
| "loss": 0.0346, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 6.318956870611835, | |
| "grad_norm": 0.18123650550842285, | |
| "learning_rate": 4.1133333333333335e-06, | |
| "loss": 0.0342, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 6.369107321965898, | |
| "grad_norm": 0.48640918731689453, | |
| "learning_rate": 4.057777777777778e-06, | |
| "loss": 0.0345, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 6.41925777331996, | |
| "grad_norm": 0.16967743635177612, | |
| "learning_rate": 4.002222222222222e-06, | |
| "loss": 0.0339, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 6.469408224674022, | |
| "grad_norm": 1.0342961549758911, | |
| "learning_rate": 3.946666666666667e-06, | |
| "loss": 0.035, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 6.519558676028084, | |
| "grad_norm": 0.4811567962169647, | |
| "learning_rate": 3.891111111111111e-06, | |
| "loss": 0.0346, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 6.569709127382146, | |
| "grad_norm": 0.41370365023612976, | |
| "learning_rate": 3.835555555555555e-06, | |
| "loss": 0.0347, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 6.619859578736209, | |
| "grad_norm": 0.32207295298576355, | |
| "learning_rate": 3.7800000000000002e-06, | |
| "loss": 0.0341, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 6.670010030090271, | |
| "grad_norm": 0.6185954213142395, | |
| "learning_rate": 3.724444444444445e-06, | |
| "loss": 0.034, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 6.720160481444333, | |
| "grad_norm": 0.23153382539749146, | |
| "learning_rate": 3.668888888888889e-06, | |
| "loss": 0.0345, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 6.770310932798395, | |
| "grad_norm": 0.8901228308677673, | |
| "learning_rate": 3.6133333333333336e-06, | |
| "loss": 0.0342, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 6.820461384152457, | |
| "grad_norm": 0.6802942156791687, | |
| "learning_rate": 3.5577777777777785e-06, | |
| "loss": 0.0355, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 6.870611835506519, | |
| "grad_norm": 0.41077226400375366, | |
| "learning_rate": 3.5022222222222225e-06, | |
| "loss": 0.0358, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 6.920762286860581, | |
| "grad_norm": 0.5837184190750122, | |
| "learning_rate": 3.446666666666667e-06, | |
| "loss": 0.0336, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 6.970912738214644, | |
| "grad_norm": 0.5418899655342102, | |
| "learning_rate": 3.391111111111111e-06, | |
| "loss": 0.0337, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 7.021063189568706, | |
| "grad_norm": 0.4523651599884033, | |
| "learning_rate": 3.335555555555556e-06, | |
| "loss": 0.0336, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 7.071213640922768, | |
| "grad_norm": 0.20040252804756165, | |
| "learning_rate": 3.2800000000000004e-06, | |
| "loss": 0.0332, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 7.12136409227683, | |
| "grad_norm": 0.8929922580718994, | |
| "learning_rate": 3.2244444444444444e-06, | |
| "loss": 0.0333, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 7.1715145436308925, | |
| "grad_norm": 0.19636617600917816, | |
| "learning_rate": 3.1688888888888893e-06, | |
| "loss": 0.0329, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 7.2216649949849545, | |
| "grad_norm": 0.5786032676696777, | |
| "learning_rate": 3.1133333333333337e-06, | |
| "loss": 0.0323, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 7.271815446339017, | |
| "grad_norm": 0.14882159233093262, | |
| "learning_rate": 3.0577777777777778e-06, | |
| "loss": 0.0323, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 7.3219658976930795, | |
| "grad_norm": 0.6977226138114929, | |
| "learning_rate": 3.0022222222222227e-06, | |
| "loss": 0.0337, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 7.3721163490471415, | |
| "grad_norm": 0.27530452609062195, | |
| "learning_rate": 2.946666666666667e-06, | |
| "loss": 0.034, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 7.422266800401204, | |
| "grad_norm": 0.16913309693336487, | |
| "learning_rate": 2.891111111111111e-06, | |
| "loss": 0.0333, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 7.472417251755266, | |
| "grad_norm": 0.42673996090888977, | |
| "learning_rate": 2.835555555555556e-06, | |
| "loss": 0.0349, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 7.522567703109328, | |
| "grad_norm": 1.0766425132751465, | |
| "learning_rate": 2.7800000000000005e-06, | |
| "loss": 0.0328, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 7.57271815446339, | |
| "grad_norm": 0.1712706983089447, | |
| "learning_rate": 2.7244444444444445e-06, | |
| "loss": 0.0332, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 7.622868605817453, | |
| "grad_norm": 0.3342408239841461, | |
| "learning_rate": 2.6688888888888894e-06, | |
| "loss": 0.0328, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 7.673019057171515, | |
| "grad_norm": 0.15629036724567413, | |
| "learning_rate": 2.6133333333333334e-06, | |
| "loss": 0.0334, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 7.723169508525577, | |
| "grad_norm": 0.923606276512146, | |
| "learning_rate": 2.557777777777778e-06, | |
| "loss": 0.0339, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 7.773319959879639, | |
| "grad_norm": 0.9420749545097351, | |
| "learning_rate": 2.5022222222222224e-06, | |
| "loss": 0.033, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 7.823470411233701, | |
| "grad_norm": 0.7262980341911316, | |
| "learning_rate": 2.446666666666667e-06, | |
| "loss": 0.0347, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 7.873620862587764, | |
| "grad_norm": 0.44713786244392395, | |
| "learning_rate": 2.3911111111111113e-06, | |
| "loss": 0.0319, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 7.923771313941826, | |
| "grad_norm": 0.16295509040355682, | |
| "learning_rate": 2.3355555555555557e-06, | |
| "loss": 0.0326, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 7.973921765295888, | |
| "grad_norm": 0.7272210717201233, | |
| "learning_rate": 2.28e-06, | |
| "loss": 0.0318, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 8.02407221664995, | |
| "grad_norm": 0.36710575222969055, | |
| "learning_rate": 2.2244444444444447e-06, | |
| "loss": 0.0326, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 8.02407221664995, | |
| "eval_loss": 0.4601472318172455, | |
| "eval_runtime": 4510.0106, | |
| "eval_samples_per_second": 2.386, | |
| "eval_steps_per_second": 0.149, | |
| "eval_wer": 0.19771376046311506, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 8.074222668004012, | |
| "grad_norm": 0.6596123576164246, | |
| "learning_rate": 2.168888888888889e-06, | |
| "loss": 0.0318, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 8.124373119358074, | |
| "grad_norm": 0.12991444766521454, | |
| "learning_rate": 2.1133333333333336e-06, | |
| "loss": 0.0329, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 8.174523570712136, | |
| "grad_norm": 0.17968349158763885, | |
| "learning_rate": 2.057777777777778e-06, | |
| "loss": 0.0315, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 8.224674022066198, | |
| "grad_norm": 0.2529272139072418, | |
| "learning_rate": 2.0022222222222225e-06, | |
| "loss": 0.0347, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 8.27482447342026, | |
| "grad_norm": 0.31072911620140076, | |
| "learning_rate": 1.9466666666666665e-06, | |
| "loss": 0.0329, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 8.324974924774324, | |
| "grad_norm": 0.3404042422771454, | |
| "learning_rate": 1.8911111111111114e-06, | |
| "loss": 0.0317, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 8.375125376128386, | |
| "grad_norm": 0.6961965560913086, | |
| "learning_rate": 1.8355555555555557e-06, | |
| "loss": 0.032, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 8.425275827482448, | |
| "grad_norm": 0.3431238830089569, | |
| "learning_rate": 1.7800000000000001e-06, | |
| "loss": 0.0327, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 8.47542627883651, | |
| "grad_norm": 0.9905340671539307, | |
| "learning_rate": 1.7244444444444448e-06, | |
| "loss": 0.032, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 8.525576730190572, | |
| "grad_norm": 0.17600752413272858, | |
| "learning_rate": 1.668888888888889e-06, | |
| "loss": 0.0321, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 8.575727181544634, | |
| "grad_norm": 0.4973747432231903, | |
| "learning_rate": 1.6133333333333335e-06, | |
| "loss": 0.0314, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 8.625877632898696, | |
| "grad_norm": 0.6604855060577393, | |
| "learning_rate": 1.5577777777777777e-06, | |
| "loss": 0.0325, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 8.676028084252758, | |
| "grad_norm": 0.46838390827178955, | |
| "learning_rate": 1.5022222222222224e-06, | |
| "loss": 0.0325, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 8.72617853560682, | |
| "grad_norm": 0.30177703499794006, | |
| "learning_rate": 1.4466666666666669e-06, | |
| "loss": 0.0316, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 8.776328986960882, | |
| "grad_norm": 0.6524584293365479, | |
| "learning_rate": 1.3911111111111111e-06, | |
| "loss": 0.0321, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 8.826479438314944, | |
| "grad_norm": 0.9478035569190979, | |
| "learning_rate": 1.3355555555555558e-06, | |
| "loss": 0.0318, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 8.876629889669006, | |
| "grad_norm": 0.1826964169740677, | |
| "learning_rate": 1.28e-06, | |
| "loss": 0.032, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 8.926780341023068, | |
| "grad_norm": 0.35147854685783386, | |
| "learning_rate": 1.2244444444444445e-06, | |
| "loss": 0.0323, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 8.976930792377132, | |
| "grad_norm": 0.12909051775932312, | |
| "learning_rate": 1.168888888888889e-06, | |
| "loss": 0.0327, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 9.027081243731194, | |
| "grad_norm": 0.18831254541873932, | |
| "learning_rate": 1.1133333333333334e-06, | |
| "loss": 0.0325, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 9.077231695085256, | |
| "grad_norm": 0.17683148384094238, | |
| "learning_rate": 1.0577777777777779e-06, | |
| "loss": 0.0328, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 9.127382146439318, | |
| "grad_norm": 0.47570478916168213, | |
| "learning_rate": 1.0022222222222223e-06, | |
| "loss": 0.0316, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 9.17753259779338, | |
| "grad_norm": 0.5286629796028137, | |
| "learning_rate": 9.466666666666667e-07, | |
| "loss": 0.0322, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 9.227683049147442, | |
| "grad_norm": 0.7967525720596313, | |
| "learning_rate": 8.911111111111112e-07, | |
| "loss": 0.0311, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 9.277833500501504, | |
| "grad_norm": 0.6575748324394226, | |
| "learning_rate": 8.355555555555556e-07, | |
| "loss": 0.0322, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 9.327983951855567, | |
| "grad_norm": 0.705969512462616, | |
| "learning_rate": 7.8e-07, | |
| "loss": 0.0316, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 9.378134403209629, | |
| "grad_norm": 0.385044664144516, | |
| "learning_rate": 7.244444444444446e-07, | |
| "loss": 0.0316, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 9.42828485456369, | |
| "grad_norm": 1.0980923175811768, | |
| "learning_rate": 6.68888888888889e-07, | |
| "loss": 0.0318, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 9.478435305917753, | |
| "grad_norm": 0.8150089979171753, | |
| "learning_rate": 6.133333333333333e-07, | |
| "loss": 0.0328, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 9.528585757271815, | |
| "grad_norm": 0.7437707185745239, | |
| "learning_rate": 5.577777777777779e-07, | |
| "loss": 0.0319, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 9.578736208625877, | |
| "grad_norm": 0.18542395532131195, | |
| "learning_rate": 5.022222222222222e-07, | |
| "loss": 0.0316, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 9.62888665997994, | |
| "grad_norm": 0.22823591530323029, | |
| "learning_rate": 4.466666666666667e-07, | |
| "loss": 0.0321, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 9.679037111334003, | |
| "grad_norm": 0.5161811113357544, | |
| "learning_rate": 3.9111111111111115e-07, | |
| "loss": 0.0323, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 9.729187562688065, | |
| "grad_norm": 0.797174334526062, | |
| "learning_rate": 3.3555555555555556e-07, | |
| "loss": 0.032, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 9.779338014042127, | |
| "grad_norm": 0.7822555899620056, | |
| "learning_rate": 2.8e-07, | |
| "loss": 0.0313, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 9.829488465396189, | |
| "grad_norm": 0.7201440930366516, | |
| "learning_rate": 2.2444444444444445e-07, | |
| "loss": 0.0319, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 9.87963891675025, | |
| "grad_norm": 0.06772810220718384, | |
| "learning_rate": 1.6888888888888888e-07, | |
| "loss": 0.032, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 9.929789368104313, | |
| "grad_norm": 0.12714877724647522, | |
| "learning_rate": 1.1333333333333336e-07, | |
| "loss": 0.0315, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 9.979939819458375, | |
| "grad_norm": 0.15611818432807922, | |
| "learning_rate": 5.777777777777778e-08, | |
| "loss": 0.0317, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 10.030090270812437, | |
| "grad_norm": 0.24346430599689484, | |
| "learning_rate": 2.2222222222222225e-09, | |
| "loss": 0.0313, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 10.030090270812437, | |
| "eval_loss": 0.4825207591056824, | |
| "eval_runtime": 4514.4484, | |
| "eval_samples_per_second": 2.384, | |
| "eval_steps_per_second": 0.149, | |
| "eval_wer": 0.19794844715637958, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 10.030090270812437, | |
| "step": 5000, | |
| "total_flos": 5.43090104303616e+20, | |
| "train_loss": 0.1254427011013031, | |
| "train_runtime": 51254.2105, | |
| "train_samples_per_second": 3.122, | |
| "train_steps_per_second": 0.098 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 5000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 11, | |
| "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": 5.43090104303616e+20, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |