| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.4344629729245113, | |
| "eval_steps": 2000, | |
| "global_step": 4000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "grad_norm": 0.9492120146751404, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 0.9769, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 1.344916582107544, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 1.0718, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 0.6680417060852051, | |
| "learning_rate": 3e-06, | |
| "loss": 0.8901, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 0.8369797468185425, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 0.9398, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 0.5738756656646729, | |
| "learning_rate": 5e-06, | |
| "loss": 0.9152, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 1.028475284576416, | |
| "learning_rate": 6e-06, | |
| "loss": 0.849, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 1.2893372774124146, | |
| "learning_rate": 7e-06, | |
| "loss": 0.7312, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 0.8779547810554504, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 0.6707, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 0.8972748517990112, | |
| "learning_rate": 9e-06, | |
| "loss": 0.6413, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 1.505399227142334, | |
| "learning_rate": 1e-05, | |
| "loss": 0.6129, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 0.6442121267318726, | |
| "learning_rate": 9.989898989898991e-06, | |
| "loss": 0.5579, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 0.623290479183197, | |
| "learning_rate": 9.97979797979798e-06, | |
| "loss": 0.6022, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 0.7216657996177673, | |
| "learning_rate": 9.96969696969697e-06, | |
| "loss": 0.5481, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 0.7031328678131104, | |
| "learning_rate": 9.95959595959596e-06, | |
| "loss": 0.5594, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 0.7411965727806091, | |
| "learning_rate": 9.94949494949495e-06, | |
| "loss": 0.6241, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 0.8735277652740479, | |
| "learning_rate": 9.939393939393939e-06, | |
| "loss": 0.498, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 1.1062073707580566, | |
| "learning_rate": 9.92929292929293e-06, | |
| "loss": 0.6463, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 0.7930333614349365, | |
| "learning_rate": 9.91919191919192e-06, | |
| "loss": 0.6231, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 0.8774802684783936, | |
| "learning_rate": 9.90909090909091e-06, | |
| "loss": 0.5895, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 0.929542064666748, | |
| "learning_rate": 9.8989898989899e-06, | |
| "loss": 0.632, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 0.6024735569953918, | |
| "learning_rate": 9.88888888888889e-06, | |
| "loss": 0.5682, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 0.5887441635131836, | |
| "learning_rate": 9.87878787878788e-06, | |
| "loss": 0.5192, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 0.7813496589660645, | |
| "learning_rate": 9.86868686868687e-06, | |
| "loss": 0.5841, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 0.5949407815933228, | |
| "learning_rate": 9.85858585858586e-06, | |
| "loss": 0.5263, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 0.8114556670188904, | |
| "learning_rate": 9.84848484848485e-06, | |
| "loss": 0.5513, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 0.7692855596542358, | |
| "learning_rate": 9.838383838383839e-06, | |
| "loss": 0.5217, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 0.8319826126098633, | |
| "learning_rate": 9.828282828282829e-06, | |
| "loss": 0.4386, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 0.6725042462348938, | |
| "learning_rate": 9.81818181818182e-06, | |
| "loss": 0.4919, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 0.779315173625946, | |
| "learning_rate": 9.80808080808081e-06, | |
| "loss": 0.5281, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 0.8005223274230957, | |
| "learning_rate": 9.797979797979798e-06, | |
| "loss": 0.5206, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 0.6395801901817322, | |
| "learning_rate": 9.787878787878788e-06, | |
| "loss": 0.474, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 1.1016992330551147, | |
| "learning_rate": 9.777777777777779e-06, | |
| "loss": 0.5403, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "grad_norm": 1.395189881324768, | |
| "learning_rate": 9.767676767676767e-06, | |
| "loss": 0.5181, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "grad_norm": 0.801499605178833, | |
| "learning_rate": 9.757575757575758e-06, | |
| "loss": 0.5296, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "grad_norm": 0.7606889605522156, | |
| "learning_rate": 9.747474747474748e-06, | |
| "loss": 0.5406, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "grad_norm": 0.580170750617981, | |
| "learning_rate": 9.737373737373738e-06, | |
| "loss": 0.541, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "grad_norm": 0.6730871796607971, | |
| "learning_rate": 9.727272727272728e-06, | |
| "loss": 0.5665, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "grad_norm": 1.240430474281311, | |
| "learning_rate": 9.717171717171719e-06, | |
| "loss": 0.5148, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "grad_norm": 0.9439683556556702, | |
| "learning_rate": 9.707070707070709e-06, | |
| "loss": 0.513, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "grad_norm": 0.6674547791481018, | |
| "learning_rate": 9.696969696969698e-06, | |
| "loss": 0.5202, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "grad_norm": 0.8516045808792114, | |
| "learning_rate": 9.686868686868688e-06, | |
| "loss": 0.5162, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "grad_norm": 0.8700432181358337, | |
| "learning_rate": 9.676767676767678e-06, | |
| "loss": 0.5392, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "grad_norm": 0.5687388777732849, | |
| "learning_rate": 9.666666666666667e-06, | |
| "loss": 0.5106, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "grad_norm": 1.2382631301879883, | |
| "learning_rate": 9.656565656565657e-06, | |
| "loss": 0.5074, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "grad_norm": 0.8921974301338196, | |
| "learning_rate": 9.646464646464647e-06, | |
| "loss": 0.561, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "grad_norm": 1.3508048057556152, | |
| "learning_rate": 9.636363636363638e-06, | |
| "loss": 0.5484, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "grad_norm": 1.1822205781936646, | |
| "learning_rate": 9.626262626262626e-06, | |
| "loss": 0.5719, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "grad_norm": 0.6617271304130554, | |
| "learning_rate": 9.616161616161616e-06, | |
| "loss": 0.535, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "grad_norm": 0.6614571809768677, | |
| "learning_rate": 9.606060606060607e-06, | |
| "loss": 0.4549, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "grad_norm": 0.8642953038215637, | |
| "learning_rate": 9.595959595959597e-06, | |
| "loss": 0.4789, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "grad_norm": 0.614743173122406, | |
| "learning_rate": 9.585858585858586e-06, | |
| "loss": 0.4854, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "grad_norm": 0.970829427242279, | |
| "learning_rate": 9.575757575757576e-06, | |
| "loss": 0.5196, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "grad_norm": 0.7311980128288269, | |
| "learning_rate": 9.565656565656566e-06, | |
| "loss": 0.5106, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "grad_norm": 0.765849769115448, | |
| "learning_rate": 9.555555555555556e-06, | |
| "loss": 0.5782, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "grad_norm": 1.0889312028884888, | |
| "learning_rate": 9.545454545454547e-06, | |
| "loss": 0.5824, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "grad_norm": 0.7402384877204895, | |
| "learning_rate": 9.535353535353537e-06, | |
| "loss": 0.5005, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "grad_norm": 0.707028329372406, | |
| "learning_rate": 9.525252525252526e-06, | |
| "loss": 0.5233, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "grad_norm": 0.8338315486907959, | |
| "learning_rate": 9.515151515151516e-06, | |
| "loss": 0.4694, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "grad_norm": 0.7450662851333618, | |
| "learning_rate": 9.505050505050506e-06, | |
| "loss": 0.4762, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "grad_norm": 0.7595840692520142, | |
| "learning_rate": 9.494949494949497e-06, | |
| "loss": 0.5018, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "grad_norm": 0.5880123376846313, | |
| "learning_rate": 9.484848484848485e-06, | |
| "loss": 0.5228, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "grad_norm": 0.9635146260261536, | |
| "learning_rate": 9.474747474747475e-06, | |
| "loss": 0.4987, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "grad_norm": 1.1274373531341553, | |
| "learning_rate": 9.464646464646466e-06, | |
| "loss": 0.5085, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "grad_norm": 1.1324831247329712, | |
| "learning_rate": 9.454545454545456e-06, | |
| "loss": 0.504, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "grad_norm": 0.5410157442092896, | |
| "learning_rate": 9.444444444444445e-06, | |
| "loss": 0.4619, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "grad_norm": 0.7583281993865967, | |
| "learning_rate": 9.434343434343435e-06, | |
| "loss": 0.5162, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "grad_norm": 0.6546668410301208, | |
| "learning_rate": 9.424242424242425e-06, | |
| "loss": 0.4969, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "grad_norm": 0.6070376634597778, | |
| "learning_rate": 9.414141414141414e-06, | |
| "loss": 0.4805, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "grad_norm": 1.0108693838119507, | |
| "learning_rate": 9.404040404040404e-06, | |
| "loss": 0.4808, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "grad_norm": 0.8799183368682861, | |
| "learning_rate": 9.393939393939396e-06, | |
| "loss": 0.5082, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "grad_norm": 1.3432070016860962, | |
| "learning_rate": 9.383838383838385e-06, | |
| "loss": 0.4353, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "grad_norm": 0.6518195271492004, | |
| "learning_rate": 9.373737373737375e-06, | |
| "loss": 0.4933, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "grad_norm": 0.6736329793930054, | |
| "learning_rate": 9.363636363636365e-06, | |
| "loss": 0.5342, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "grad_norm": 0.776785135269165, | |
| "learning_rate": 9.353535353535354e-06, | |
| "loss": 0.5162, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "grad_norm": 0.9443957805633545, | |
| "learning_rate": 9.343434343434344e-06, | |
| "loss": 0.4486, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "grad_norm": 0.8882728815078735, | |
| "learning_rate": 9.333333333333334e-06, | |
| "loss": 0.4774, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "grad_norm": 0.819239616394043, | |
| "learning_rate": 9.323232323232325e-06, | |
| "loss": 0.5, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "grad_norm": 0.8829065561294556, | |
| "learning_rate": 9.313131313131313e-06, | |
| "loss": 0.4655, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "grad_norm": 1.1993345022201538, | |
| "learning_rate": 9.303030303030303e-06, | |
| "loss": 0.5392, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "grad_norm": 0.681409478187561, | |
| "learning_rate": 9.292929292929294e-06, | |
| "loss": 0.5076, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "grad_norm": 1.075088381767273, | |
| "learning_rate": 9.282828282828284e-06, | |
| "loss": 0.4953, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "grad_norm": 0.7040195465087891, | |
| "learning_rate": 9.272727272727273e-06, | |
| "loss": 0.5596, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "grad_norm": 1.1210192441940308, | |
| "learning_rate": 9.262626262626263e-06, | |
| "loss": 0.5299, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "grad_norm": 0.583011269569397, | |
| "learning_rate": 9.252525252525253e-06, | |
| "loss": 0.497, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "grad_norm": 0.6583887338638306, | |
| "learning_rate": 9.242424242424244e-06, | |
| "loss": 0.5556, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "grad_norm": 0.8040810227394104, | |
| "learning_rate": 9.232323232323232e-06, | |
| "loss": 0.5257, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "grad_norm": 0.9269919991493225, | |
| "learning_rate": 9.222222222222224e-06, | |
| "loss": 0.4421, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "grad_norm": 0.9947918653488159, | |
| "learning_rate": 9.212121212121213e-06, | |
| "loss": 0.5297, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "grad_norm": 0.6900811791419983, | |
| "learning_rate": 9.202020202020203e-06, | |
| "loss": 0.4833, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "grad_norm": 0.8033557534217834, | |
| "learning_rate": 9.191919191919193e-06, | |
| "loss": 0.4894, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "grad_norm": 0.635124683380127, | |
| "learning_rate": 9.181818181818184e-06, | |
| "loss": 0.4554, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "grad_norm": 0.7293840646743774, | |
| "learning_rate": 9.171717171717172e-06, | |
| "loss": 0.4693, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "grad_norm": 0.7628031373023987, | |
| "learning_rate": 9.161616161616162e-06, | |
| "loss": 0.5181, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "grad_norm": 1.0783181190490723, | |
| "learning_rate": 9.151515151515153e-06, | |
| "loss": 0.4632, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "grad_norm": 0.5340379476547241, | |
| "learning_rate": 9.141414141414143e-06, | |
| "loss": 0.4664, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "grad_norm": 0.9029551148414612, | |
| "learning_rate": 9.131313131313132e-06, | |
| "loss": 0.5333, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "grad_norm": 0.7257616519927979, | |
| "learning_rate": 9.121212121212122e-06, | |
| "loss": 0.5168, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "grad_norm": 0.761325478553772, | |
| "learning_rate": 9.111111111111112e-06, | |
| "loss": 0.5606, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "grad_norm": 0.8582245707511902, | |
| "learning_rate": 9.1010101010101e-06, | |
| "loss": 0.4332, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "grad_norm": 0.8598415851593018, | |
| "learning_rate": 9.090909090909091e-06, | |
| "loss": 0.5884, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "grad_norm": 0.8292351365089417, | |
| "learning_rate": 9.080808080808081e-06, | |
| "loss": 0.4848, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "grad_norm": 1.0559266805648804, | |
| "learning_rate": 9.070707070707072e-06, | |
| "loss": 0.4588, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "grad_norm": 0.6693033576011658, | |
| "learning_rate": 9.06060606060606e-06, | |
| "loss": 0.5333, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "grad_norm": 0.8114706873893738, | |
| "learning_rate": 9.050505050505052e-06, | |
| "loss": 0.5166, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "grad_norm": 0.8659316301345825, | |
| "learning_rate": 9.040404040404042e-06, | |
| "loss": 0.4504, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "grad_norm": 0.9083582758903503, | |
| "learning_rate": 9.030303030303031e-06, | |
| "loss": 0.5611, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "grad_norm": 0.6691566109657288, | |
| "learning_rate": 9.020202020202021e-06, | |
| "loss": 0.5192, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "grad_norm": 0.5889317989349365, | |
| "learning_rate": 9.010101010101012e-06, | |
| "loss": 0.4515, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "grad_norm": 0.9215373992919922, | |
| "learning_rate": 9e-06, | |
| "loss": 0.4776, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "grad_norm": 0.7439729571342468, | |
| "learning_rate": 8.98989898989899e-06, | |
| "loss": 0.4656, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "grad_norm": 1.1780657768249512, | |
| "learning_rate": 8.97979797979798e-06, | |
| "loss": 0.4933, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "grad_norm": 0.9686077833175659, | |
| "learning_rate": 8.969696969696971e-06, | |
| "loss": 0.5167, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "grad_norm": 0.829994261264801, | |
| "learning_rate": 8.95959595959596e-06, | |
| "loss": 0.491, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "grad_norm": 0.6313827633857727, | |
| "learning_rate": 8.94949494949495e-06, | |
| "loss": 0.4864, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "grad_norm": 0.596537709236145, | |
| "learning_rate": 8.93939393939394e-06, | |
| "loss": 0.4807, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "grad_norm": 0.8714896440505981, | |
| "learning_rate": 8.92929292929293e-06, | |
| "loss": 0.512, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "grad_norm": 0.9466399550437927, | |
| "learning_rate": 8.919191919191919e-06, | |
| "loss": 0.4883, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "grad_norm": 0.7337993383407593, | |
| "learning_rate": 8.90909090909091e-06, | |
| "loss": 0.4757, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "grad_norm": 0.7684504985809326, | |
| "learning_rate": 8.8989898989899e-06, | |
| "loss": 0.5224, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "grad_norm": 0.6455299854278564, | |
| "learning_rate": 8.888888888888888e-06, | |
| "loss": 0.5346, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "grad_norm": 0.7279661893844604, | |
| "learning_rate": 8.87878787878788e-06, | |
| "loss": 0.4845, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "grad_norm": 0.7996425032615662, | |
| "learning_rate": 8.86868686868687e-06, | |
| "loss": 0.4812, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "grad_norm": 0.7299336791038513, | |
| "learning_rate": 8.85858585858586e-06, | |
| "loss": 0.4985, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "grad_norm": 0.7462379932403564, | |
| "learning_rate": 8.84848484848485e-06, | |
| "loss": 0.5742, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "grad_norm": 0.7165307998657227, | |
| "learning_rate": 8.83838383838384e-06, | |
| "loss": 0.4627, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "grad_norm": 0.7239411473274231, | |
| "learning_rate": 8.82828282828283e-06, | |
| "loss": 0.5469, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "grad_norm": 0.5761345028877258, | |
| "learning_rate": 8.818181818181819e-06, | |
| "loss": 0.4391, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "grad_norm": 0.8207817077636719, | |
| "learning_rate": 8.808080808080809e-06, | |
| "loss": 0.5477, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "grad_norm": 0.9331930875778198, | |
| "learning_rate": 8.7979797979798e-06, | |
| "loss": 0.4975, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "grad_norm": 1.0401968955993652, | |
| "learning_rate": 8.787878787878788e-06, | |
| "loss": 0.5149, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "grad_norm": 0.7848596572875977, | |
| "learning_rate": 8.777777777777778e-06, | |
| "loss": 0.5448, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "grad_norm": 1.5979989767074585, | |
| "learning_rate": 8.767676767676768e-06, | |
| "loss": 0.4861, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "grad_norm": 0.7734145522117615, | |
| "learning_rate": 8.757575757575759e-06, | |
| "loss": 0.5955, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "grad_norm": 0.9506089687347412, | |
| "learning_rate": 8.747474747474747e-06, | |
| "loss": 0.5212, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "grad_norm": 0.7118907570838928, | |
| "learning_rate": 8.737373737373738e-06, | |
| "loss": 0.474, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "grad_norm": 1.0045005083084106, | |
| "learning_rate": 8.727272727272728e-06, | |
| "loss": 0.509, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "grad_norm": 1.1283302307128906, | |
| "learning_rate": 8.717171717171718e-06, | |
| "loss": 0.5096, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "grad_norm": 1.3884085416793823, | |
| "learning_rate": 8.707070707070707e-06, | |
| "loss": 0.5008, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "grad_norm": 0.7452566027641296, | |
| "learning_rate": 8.696969696969699e-06, | |
| "loss": 0.4792, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "grad_norm": 1.3434414863586426, | |
| "learning_rate": 8.686868686868687e-06, | |
| "loss": 0.5385, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "grad_norm": 0.9600369930267334, | |
| "learning_rate": 8.676767676767678e-06, | |
| "loss": 0.4743, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "grad_norm": 0.6895599961280823, | |
| "learning_rate": 8.666666666666668e-06, | |
| "loss": 0.5639, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "grad_norm": 0.9460670948028564, | |
| "learning_rate": 8.656565656565658e-06, | |
| "loss": 0.5957, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "grad_norm": 0.6181186437606812, | |
| "learning_rate": 8.646464646464647e-06, | |
| "loss": 0.5032, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "grad_norm": 0.6992838382720947, | |
| "learning_rate": 8.636363636363637e-06, | |
| "loss": 0.5371, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "grad_norm": 0.6007195711135864, | |
| "learning_rate": 8.626262626262627e-06, | |
| "loss": 0.4446, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "grad_norm": 0.8171485662460327, | |
| "learning_rate": 8.616161616161618e-06, | |
| "loss": 0.4835, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "grad_norm": 0.9369902610778809, | |
| "learning_rate": 8.606060606060606e-06, | |
| "loss": 0.5437, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "grad_norm": 1.1894652843475342, | |
| "learning_rate": 8.595959595959596e-06, | |
| "loss": 0.5151, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "grad_norm": 0.694837212562561, | |
| "learning_rate": 8.585858585858587e-06, | |
| "loss": 0.488, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "grad_norm": 0.8813522458076477, | |
| "learning_rate": 8.575757575757575e-06, | |
| "loss": 0.5129, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "grad_norm": 1.0121350288391113, | |
| "learning_rate": 8.565656565656566e-06, | |
| "loss": 0.5165, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "grad_norm": 0.7935439944267273, | |
| "learning_rate": 8.555555555555556e-06, | |
| "loss": 0.4736, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "grad_norm": 0.7670463919639587, | |
| "learning_rate": 8.545454545454546e-06, | |
| "loss": 0.5214, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "grad_norm": 1.212927222251892, | |
| "learning_rate": 8.535353535353535e-06, | |
| "loss": 0.5125, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "grad_norm": 0.7966919541358948, | |
| "learning_rate": 8.525252525252527e-06, | |
| "loss": 0.4823, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "grad_norm": 1.0880494117736816, | |
| "learning_rate": 8.515151515151517e-06, | |
| "loss": 0.478, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "grad_norm": 1.0308737754821777, | |
| "learning_rate": 8.505050505050506e-06, | |
| "loss": 0.5368, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "grad_norm": 0.7291275262832642, | |
| "learning_rate": 8.494949494949496e-06, | |
| "loss": 0.4838, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "grad_norm": 0.6764214038848877, | |
| "learning_rate": 8.484848484848486e-06, | |
| "loss": 0.4882, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "grad_norm": 1.051628828048706, | |
| "learning_rate": 8.474747474747475e-06, | |
| "loss": 0.4564, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "grad_norm": 0.8614614605903625, | |
| "learning_rate": 8.464646464646465e-06, | |
| "loss": 0.5632, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "grad_norm": 1.1045228242874146, | |
| "learning_rate": 8.454545454545455e-06, | |
| "loss": 0.4535, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "grad_norm": 0.8160364031791687, | |
| "learning_rate": 8.444444444444446e-06, | |
| "loss": 0.4964, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "grad_norm": 0.7776429653167725, | |
| "learning_rate": 8.434343434343434e-06, | |
| "loss": 0.4554, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "grad_norm": 0.7589672207832336, | |
| "learning_rate": 8.424242424242425e-06, | |
| "loss": 0.4458, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "grad_norm": 0.825233519077301, | |
| "learning_rate": 8.414141414141415e-06, | |
| "loss": 0.4814, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "grad_norm": 0.8226912617683411, | |
| "learning_rate": 8.404040404040405e-06, | |
| "loss": 0.4892, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "grad_norm": 0.9273412823677063, | |
| "learning_rate": 8.393939393939394e-06, | |
| "loss": 0.5143, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "grad_norm": 0.9043828248977661, | |
| "learning_rate": 8.383838383838384e-06, | |
| "loss": 0.4598, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "grad_norm": 2.2596805095672607, | |
| "learning_rate": 8.373737373737374e-06, | |
| "loss": 0.5415, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "grad_norm": 1.0041202306747437, | |
| "learning_rate": 8.363636363636365e-06, | |
| "loss": 0.5899, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "grad_norm": 0.9188370704650879, | |
| "learning_rate": 8.353535353535355e-06, | |
| "loss": 0.5119, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "grad_norm": 0.7778961062431335, | |
| "learning_rate": 8.343434343434345e-06, | |
| "loss": 0.5237, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "grad_norm": 0.7438649535179138, | |
| "learning_rate": 8.333333333333334e-06, | |
| "loss": 0.4999, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "grad_norm": 0.5649489760398865, | |
| "learning_rate": 8.323232323232324e-06, | |
| "loss": 0.552, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "grad_norm": 0.5625451803207397, | |
| "learning_rate": 8.313131313131314e-06, | |
| "loss": 0.4549, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "grad_norm": 1.3711755275726318, | |
| "learning_rate": 8.303030303030305e-06, | |
| "loss": 0.445, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "grad_norm": 1.4339165687561035, | |
| "learning_rate": 8.292929292929293e-06, | |
| "loss": 0.4975, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "grad_norm": 0.8113200068473816, | |
| "learning_rate": 8.282828282828283e-06, | |
| "loss": 0.5288, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "grad_norm": 1.1567124128341675, | |
| "learning_rate": 8.272727272727274e-06, | |
| "loss": 0.4669, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "grad_norm": 0.7966761589050293, | |
| "learning_rate": 8.262626262626264e-06, | |
| "loss": 0.4856, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "grad_norm": 1.0181186199188232, | |
| "learning_rate": 8.252525252525253e-06, | |
| "loss": 0.4576, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "grad_norm": 0.611566960811615, | |
| "learning_rate": 8.242424242424243e-06, | |
| "loss": 0.496, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "grad_norm": 0.6482832431793213, | |
| "learning_rate": 8.232323232323233e-06, | |
| "loss": 0.4601, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "grad_norm": 0.7550622820854187, | |
| "learning_rate": 8.222222222222222e-06, | |
| "loss": 0.5036, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "grad_norm": 0.7835694551467896, | |
| "learning_rate": 8.212121212121212e-06, | |
| "loss": 0.5617, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "grad_norm": 0.7926068305969238, | |
| "learning_rate": 8.202020202020202e-06, | |
| "loss": 0.4327, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "grad_norm": 0.786851167678833, | |
| "learning_rate": 8.191919191919193e-06, | |
| "loss": 0.4654, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "grad_norm": 0.9023171663284302, | |
| "learning_rate": 8.181818181818183e-06, | |
| "loss": 0.5426, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "grad_norm": 1.0345401763916016, | |
| "learning_rate": 8.171717171717173e-06, | |
| "loss": 0.52, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "grad_norm": 0.945004940032959, | |
| "learning_rate": 8.161616161616162e-06, | |
| "loss": 0.5512, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "grad_norm": 0.658362090587616, | |
| "learning_rate": 8.151515151515152e-06, | |
| "loss": 0.513, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "grad_norm": 0.6390058398246765, | |
| "learning_rate": 8.141414141414142e-06, | |
| "loss": 0.5652, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "grad_norm": 0.7705880403518677, | |
| "learning_rate": 8.131313131313133e-06, | |
| "loss": 0.4929, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "grad_norm": 0.5400047302246094, | |
| "learning_rate": 8.121212121212121e-06, | |
| "loss": 0.4932, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "grad_norm": 0.9128320217132568, | |
| "learning_rate": 8.111111111111112e-06, | |
| "loss": 0.5085, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "grad_norm": 1.0019017457962036, | |
| "learning_rate": 8.101010101010102e-06, | |
| "loss": 0.4552, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "grad_norm": 0.818148136138916, | |
| "learning_rate": 8.090909090909092e-06, | |
| "loss": 0.5517, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "grad_norm": 0.8848174810409546, | |
| "learning_rate": 8.08080808080808e-06, | |
| "loss": 0.5094, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "eval_loss": 0.6485620141029358, | |
| "eval_runtime": 340.5095, | |
| "eval_samples_per_second": 2.937, | |
| "eval_steps_per_second": 2.937, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "grad_norm": 0.9389908313751221, | |
| "learning_rate": 8.070707070707071e-06, | |
| "loss": 0.5228, | |
| "step": 2010 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "grad_norm": 0.901942789554596, | |
| "learning_rate": 8.060606060606061e-06, | |
| "loss": 0.4322, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "grad_norm": 0.6482081413269043, | |
| "learning_rate": 8.050505050505052e-06, | |
| "loss": 0.4123, | |
| "step": 2030 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "grad_norm": 1.188251256942749, | |
| "learning_rate": 8.04040404040404e-06, | |
| "loss": 0.3863, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "grad_norm": 0.6995498538017273, | |
| "learning_rate": 8.03030303030303e-06, | |
| "loss": 0.4834, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "grad_norm": 0.7236158847808838, | |
| "learning_rate": 8.02020202020202e-06, | |
| "loss": 0.5279, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "grad_norm": 0.6699327826499939, | |
| "learning_rate": 8.010101010101011e-06, | |
| "loss": 0.5294, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "grad_norm": 0.7172369956970215, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 0.4882, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "grad_norm": 0.7741474509239197, | |
| "learning_rate": 7.989898989898992e-06, | |
| "loss": 0.4928, | |
| "step": 2090 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "grad_norm": 0.7598722577095032, | |
| "learning_rate": 7.97979797979798e-06, | |
| "loss": 0.477, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "grad_norm": 0.7849257588386536, | |
| "learning_rate": 7.96969696969697e-06, | |
| "loss": 0.497, | |
| "step": 2110 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "grad_norm": 0.7920640707015991, | |
| "learning_rate": 7.95959595959596e-06, | |
| "loss": 0.4974, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "grad_norm": 0.7092992067337036, | |
| "learning_rate": 7.949494949494951e-06, | |
| "loss": 0.5234, | |
| "step": 2130 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "grad_norm": 0.8268713355064392, | |
| "learning_rate": 7.93939393939394e-06, | |
| "loss": 0.3966, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "grad_norm": 1.1046198606491089, | |
| "learning_rate": 7.92929292929293e-06, | |
| "loss": 0.452, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "grad_norm": 0.8234682083129883, | |
| "learning_rate": 7.91919191919192e-06, | |
| "loss": 0.5364, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "grad_norm": 0.8524590730667114, | |
| "learning_rate": 7.909090909090909e-06, | |
| "loss": 0.3922, | |
| "step": 2170 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "grad_norm": 0.964320182800293, | |
| "learning_rate": 7.898989898989899e-06, | |
| "loss": 0.4991, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "grad_norm": 1.2167562246322632, | |
| "learning_rate": 7.88888888888889e-06, | |
| "loss": 0.4237, | |
| "step": 2190 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "grad_norm": 0.893385648727417, | |
| "learning_rate": 7.87878787878788e-06, | |
| "loss": 0.503, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "grad_norm": 1.4781544208526611, | |
| "learning_rate": 7.868686868686868e-06, | |
| "loss": 0.4084, | |
| "step": 2210 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "grad_norm": 1.0282255411148071, | |
| "learning_rate": 7.858585858585859e-06, | |
| "loss": 0.5531, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "grad_norm": 0.5956897139549255, | |
| "learning_rate": 7.848484848484849e-06, | |
| "loss": 0.421, | |
| "step": 2230 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "grad_norm": 0.674526572227478, | |
| "learning_rate": 7.838383838383839e-06, | |
| "loss": 0.5221, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "grad_norm": 0.7602831125259399, | |
| "learning_rate": 7.82828282828283e-06, | |
| "loss": 0.4826, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "grad_norm": 0.637804388999939, | |
| "learning_rate": 7.81818181818182e-06, | |
| "loss": 0.4965, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "grad_norm": 1.0468482971191406, | |
| "learning_rate": 7.808080808080808e-06, | |
| "loss": 0.4352, | |
| "step": 2270 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "grad_norm": 0.8232371807098389, | |
| "learning_rate": 7.797979797979799e-06, | |
| "loss": 0.4314, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "grad_norm": 1.2416061162948608, | |
| "learning_rate": 7.787878787878789e-06, | |
| "loss": 0.4559, | |
| "step": 2290 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "grad_norm": 0.7591462135314941, | |
| "learning_rate": 7.77777777777778e-06, | |
| "loss": 0.5443, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "grad_norm": 0.7769295573234558, | |
| "learning_rate": 7.767676767676768e-06, | |
| "loss": 0.4915, | |
| "step": 2310 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "grad_norm": 0.7007648944854736, | |
| "learning_rate": 7.757575757575758e-06, | |
| "loss": 0.4102, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "grad_norm": 0.8512477874755859, | |
| "learning_rate": 7.747474747474748e-06, | |
| "loss": 0.488, | |
| "step": 2330 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "grad_norm": 1.0806859731674194, | |
| "learning_rate": 7.737373737373739e-06, | |
| "loss": 0.4482, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "grad_norm": 0.8377964496612549, | |
| "learning_rate": 7.727272727272727e-06, | |
| "loss": 0.4714, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "grad_norm": 0.9962327480316162, | |
| "learning_rate": 7.717171717171717e-06, | |
| "loss": 0.4354, | |
| "step": 2360 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "grad_norm": 0.7444832921028137, | |
| "learning_rate": 7.707070707070708e-06, | |
| "loss": 0.5651, | |
| "step": 2370 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "grad_norm": 0.7137871384620667, | |
| "learning_rate": 7.696969696969696e-06, | |
| "loss": 0.4773, | |
| "step": 2380 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "grad_norm": 0.8521376252174377, | |
| "learning_rate": 7.686868686868687e-06, | |
| "loss": 0.4961, | |
| "step": 2390 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "grad_norm": 0.9442594647407532, | |
| "learning_rate": 7.676767676767677e-06, | |
| "loss": 0.4627, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "grad_norm": 1.169360876083374, | |
| "learning_rate": 7.666666666666667e-06, | |
| "loss": 0.4601, | |
| "step": 2410 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "grad_norm": 0.8181536197662354, | |
| "learning_rate": 7.656565656565658e-06, | |
| "loss": 0.4635, | |
| "step": 2420 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "grad_norm": 0.9113516211509705, | |
| "learning_rate": 7.646464646464648e-06, | |
| "loss": 0.4219, | |
| "step": 2430 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "grad_norm": 1.26584792137146, | |
| "learning_rate": 7.636363636363638e-06, | |
| "loss": 0.5165, | |
| "step": 2440 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "grad_norm": 0.6445989608764648, | |
| "learning_rate": 7.6262626262626275e-06, | |
| "loss": 0.4755, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "grad_norm": 0.7551100850105286, | |
| "learning_rate": 7.616161616161617e-06, | |
| "loss": 0.4976, | |
| "step": 2460 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "grad_norm": 1.0578736066818237, | |
| "learning_rate": 7.606060606060606e-06, | |
| "loss": 0.524, | |
| "step": 2470 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "grad_norm": 1.0422261953353882, | |
| "learning_rate": 7.595959595959597e-06, | |
| "loss": 0.5136, | |
| "step": 2480 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "grad_norm": 1.0524308681488037, | |
| "learning_rate": 7.585858585858586e-06, | |
| "loss": 0.5058, | |
| "step": 2490 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "grad_norm": 0.8334051966667175, | |
| "learning_rate": 7.5757575757575764e-06, | |
| "loss": 0.4487, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "grad_norm": 0.9613906145095825, | |
| "learning_rate": 7.565656565656566e-06, | |
| "loss": 0.5055, | |
| "step": 2510 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "grad_norm": 0.9347293972969055, | |
| "learning_rate": 7.555555555555556e-06, | |
| "loss": 0.4672, | |
| "step": 2520 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "grad_norm": 1.0498868227005005, | |
| "learning_rate": 7.545454545454546e-06, | |
| "loss": 0.4336, | |
| "step": 2530 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "grad_norm": 1.0005099773406982, | |
| "learning_rate": 7.535353535353536e-06, | |
| "loss": 0.523, | |
| "step": 2540 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "grad_norm": 1.027875542640686, | |
| "learning_rate": 7.525252525252525e-06, | |
| "loss": 0.5088, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "grad_norm": 0.7960171103477478, | |
| "learning_rate": 7.515151515151516e-06, | |
| "loss": 0.4636, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "grad_norm": 1.1945937871932983, | |
| "learning_rate": 7.505050505050505e-06, | |
| "loss": 0.4736, | |
| "step": 2570 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "grad_norm": 0.9849948287010193, | |
| "learning_rate": 7.494949494949496e-06, | |
| "loss": 0.4766, | |
| "step": 2580 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "grad_norm": 0.7381343841552734, | |
| "learning_rate": 7.484848484848486e-06, | |
| "loss": 0.4427, | |
| "step": 2590 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "grad_norm": 0.964435875415802, | |
| "learning_rate": 7.474747474747476e-06, | |
| "loss": 0.5179, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "grad_norm": 1.0832308530807495, | |
| "learning_rate": 7.464646464646465e-06, | |
| "loss": 0.5179, | |
| "step": 2610 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "grad_norm": 0.8275842070579529, | |
| "learning_rate": 7.454545454545456e-06, | |
| "loss": 0.3957, | |
| "step": 2620 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "grad_norm": 1.7618038654327393, | |
| "learning_rate": 7.444444444444445e-06, | |
| "loss": 0.4626, | |
| "step": 2630 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "grad_norm": 0.9667460918426514, | |
| "learning_rate": 7.434343434343435e-06, | |
| "loss": 0.4338, | |
| "step": 2640 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "grad_norm": 0.8243513107299805, | |
| "learning_rate": 7.424242424242425e-06, | |
| "loss": 0.496, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "grad_norm": 0.8097424507141113, | |
| "learning_rate": 7.414141414141415e-06, | |
| "loss": 0.4911, | |
| "step": 2660 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "grad_norm": 1.219351053237915, | |
| "learning_rate": 7.4040404040404045e-06, | |
| "loss": 0.5163, | |
| "step": 2670 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "grad_norm": 0.7272489666938782, | |
| "learning_rate": 7.393939393939395e-06, | |
| "loss": 0.4783, | |
| "step": 2680 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "grad_norm": 0.9799202084541321, | |
| "learning_rate": 7.383838383838384e-06, | |
| "loss": 0.4638, | |
| "step": 2690 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "grad_norm": 1.1537474393844604, | |
| "learning_rate": 7.373737373737374e-06, | |
| "loss": 0.4184, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "grad_norm": 0.750483512878418, | |
| "learning_rate": 7.363636363636364e-06, | |
| "loss": 0.4544, | |
| "step": 2710 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "grad_norm": 0.915256679058075, | |
| "learning_rate": 7.353535353535353e-06, | |
| "loss": 0.4579, | |
| "step": 2720 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "grad_norm": 1.0220776796340942, | |
| "learning_rate": 7.343434343434344e-06, | |
| "loss": 0.4568, | |
| "step": 2730 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "grad_norm": 1.0540516376495361, | |
| "learning_rate": 7.333333333333333e-06, | |
| "loss": 0.5228, | |
| "step": 2740 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "grad_norm": 2.014601230621338, | |
| "learning_rate": 7.323232323232324e-06, | |
| "loss": 0.4482, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "grad_norm": 0.9005866646766663, | |
| "learning_rate": 7.3131313131313146e-06, | |
| "loss": 0.4868, | |
| "step": 2760 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "grad_norm": 1.0538028478622437, | |
| "learning_rate": 7.303030303030304e-06, | |
| "loss": 0.4884, | |
| "step": 2770 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 0.9115027785301208, | |
| "learning_rate": 7.2929292929292934e-06, | |
| "loss": 0.4421, | |
| "step": 2780 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 1.5715361833572388, | |
| "learning_rate": 7.282828282828284e-06, | |
| "loss": 0.4442, | |
| "step": 2790 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 0.8492701649665833, | |
| "learning_rate": 7.272727272727273e-06, | |
| "loss": 0.4828, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "grad_norm": 0.7459146976470947, | |
| "learning_rate": 7.2626262626262635e-06, | |
| "loss": 0.4649, | |
| "step": 2810 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "grad_norm": 0.6268482804298401, | |
| "learning_rate": 7.252525252525253e-06, | |
| "loss": 0.4521, | |
| "step": 2820 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "grad_norm": 0.9125592112541199, | |
| "learning_rate": 7.242424242424243e-06, | |
| "loss": 0.57, | |
| "step": 2830 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "grad_norm": 0.8162791132926941, | |
| "learning_rate": 7.232323232323233e-06, | |
| "loss": 0.4746, | |
| "step": 2840 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "grad_norm": 0.9345506429672241, | |
| "learning_rate": 7.222222222222223e-06, | |
| "loss": 0.4777, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "grad_norm": 0.6382574439048767, | |
| "learning_rate": 7.212121212121212e-06, | |
| "loss": 0.4832, | |
| "step": 2860 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "grad_norm": 0.9409840703010559, | |
| "learning_rate": 7.202020202020203e-06, | |
| "loss": 0.4659, | |
| "step": 2870 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "grad_norm": 1.0053200721740723, | |
| "learning_rate": 7.191919191919192e-06, | |
| "loss": 0.4041, | |
| "step": 2880 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "grad_norm": 0.8394672870635986, | |
| "learning_rate": 7.181818181818182e-06, | |
| "loss": 0.4477, | |
| "step": 2890 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "grad_norm": 0.8685116171836853, | |
| "learning_rate": 7.171717171717172e-06, | |
| "loss": 0.4155, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "grad_norm": 0.8150361776351929, | |
| "learning_rate": 7.161616161616162e-06, | |
| "loss": 0.3901, | |
| "step": 2910 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "grad_norm": 0.7799510955810547, | |
| "learning_rate": 7.151515151515152e-06, | |
| "loss": 0.4273, | |
| "step": 2920 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "grad_norm": 1.1197009086608887, | |
| "learning_rate": 7.141414141414143e-06, | |
| "loss": 0.447, | |
| "step": 2930 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "grad_norm": 0.8753387331962585, | |
| "learning_rate": 7.131313131313132e-06, | |
| "loss": 0.3847, | |
| "step": 2940 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "grad_norm": 0.9633245468139648, | |
| "learning_rate": 7.121212121212122e-06, | |
| "loss": 0.4985, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "grad_norm": 1.0391559600830078, | |
| "learning_rate": 7.111111111111112e-06, | |
| "loss": 0.4476, | |
| "step": 2960 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "grad_norm": 0.9432191252708435, | |
| "learning_rate": 7.101010101010102e-06, | |
| "loss": 0.4669, | |
| "step": 2970 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "grad_norm": 1.078410267829895, | |
| "learning_rate": 7.0909090909090916e-06, | |
| "loss": 0.401, | |
| "step": 2980 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "grad_norm": 0.9405683875083923, | |
| "learning_rate": 7.080808080808082e-06, | |
| "loss": 0.4744, | |
| "step": 2990 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "grad_norm": 1.4189893007278442, | |
| "learning_rate": 7.070707070707071e-06, | |
| "loss": 0.4334, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "grad_norm": 0.9169256091117859, | |
| "learning_rate": 7.060606060606061e-06, | |
| "loss": 0.4234, | |
| "step": 3010 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "grad_norm": 0.8117393851280212, | |
| "learning_rate": 7.050505050505051e-06, | |
| "loss": 0.4676, | |
| "step": 3020 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "grad_norm": 0.967890739440918, | |
| "learning_rate": 7.0404040404040404e-06, | |
| "loss": 0.4497, | |
| "step": 3030 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "grad_norm": 1.0548313856124878, | |
| "learning_rate": 7.030303030303031e-06, | |
| "loss": 0.4429, | |
| "step": 3040 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "grad_norm": 1.071303367614746, | |
| "learning_rate": 7.02020202020202e-06, | |
| "loss": 0.4251, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "grad_norm": 1.1772490739822388, | |
| "learning_rate": 7.0101010101010105e-06, | |
| "loss": 0.4897, | |
| "step": 3060 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "grad_norm": 1.3903322219848633, | |
| "learning_rate": 7e-06, | |
| "loss": 0.4953, | |
| "step": 3070 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "grad_norm": 0.8463280200958252, | |
| "learning_rate": 6.98989898989899e-06, | |
| "loss": 0.3923, | |
| "step": 3080 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "grad_norm": 1.146888256072998, | |
| "learning_rate": 6.979797979797981e-06, | |
| "loss": 0.4723, | |
| "step": 3090 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "grad_norm": 1.0877312421798706, | |
| "learning_rate": 6.969696969696971e-06, | |
| "loss": 0.4822, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "grad_norm": 0.7444977164268494, | |
| "learning_rate": 6.95959595959596e-06, | |
| "loss": 0.4591, | |
| "step": 3110 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "grad_norm": 0.7259607911109924, | |
| "learning_rate": 6.9494949494949505e-06, | |
| "loss": 0.4564, | |
| "step": 3120 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "grad_norm": 0.9535048604011536, | |
| "learning_rate": 6.93939393939394e-06, | |
| "loss": 0.4402, | |
| "step": 3130 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "grad_norm": 0.9428088068962097, | |
| "learning_rate": 6.92929292929293e-06, | |
| "loss": 0.4039, | |
| "step": 3140 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "grad_norm": 0.9858347177505493, | |
| "learning_rate": 6.91919191919192e-06, | |
| "loss": 0.4334, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "grad_norm": 1.0837122201919556, | |
| "learning_rate": 6.90909090909091e-06, | |
| "loss": 0.3957, | |
| "step": 3160 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "grad_norm": 1.0605705976486206, | |
| "learning_rate": 6.898989898989899e-06, | |
| "loss": 0.4195, | |
| "step": 3170 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "grad_norm": 1.0856525897979736, | |
| "learning_rate": 6.88888888888889e-06, | |
| "loss": 0.4286, | |
| "step": 3180 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "grad_norm": 1.02509605884552, | |
| "learning_rate": 6.878787878787879e-06, | |
| "loss": 0.4413, | |
| "step": 3190 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "grad_norm": 0.8847616314888, | |
| "learning_rate": 6.868686868686869e-06, | |
| "loss": 0.4117, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "grad_norm": 1.1464060544967651, | |
| "learning_rate": 6.858585858585859e-06, | |
| "loss": 0.44, | |
| "step": 3210 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "grad_norm": 0.9589990973472595, | |
| "learning_rate": 6.848484848484849e-06, | |
| "loss": 0.4896, | |
| "step": 3220 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "grad_norm": 1.080541968345642, | |
| "learning_rate": 6.8383838383838386e-06, | |
| "loss": 0.4203, | |
| "step": 3230 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "grad_norm": 1.3150702714920044, | |
| "learning_rate": 6.828282828282828e-06, | |
| "loss": 0.4213, | |
| "step": 3240 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "grad_norm": 1.0854105949401855, | |
| "learning_rate": 6.818181818181818e-06, | |
| "loss": 0.4538, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "grad_norm": 0.7211188673973083, | |
| "learning_rate": 6.808080808080809e-06, | |
| "loss": 0.4332, | |
| "step": 3260 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "grad_norm": 1.0323004722595215, | |
| "learning_rate": 6.797979797979799e-06, | |
| "loss": 0.465, | |
| "step": 3270 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "grad_norm": 0.9176574349403381, | |
| "learning_rate": 6.787878787878789e-06, | |
| "loss": 0.4898, | |
| "step": 3280 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "grad_norm": 1.2033518552780151, | |
| "learning_rate": 6.777777777777779e-06, | |
| "loss": 0.4652, | |
| "step": 3290 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "grad_norm": 0.6069207191467285, | |
| "learning_rate": 6.767676767676769e-06, | |
| "loss": 0.4663, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "grad_norm": 0.9627435803413391, | |
| "learning_rate": 6.757575757575758e-06, | |
| "loss": 0.4992, | |
| "step": 3310 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "grad_norm": 0.7333098649978638, | |
| "learning_rate": 6.747474747474749e-06, | |
| "loss": 0.4876, | |
| "step": 3320 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "grad_norm": 0.6735036969184875, | |
| "learning_rate": 6.737373737373738e-06, | |
| "loss": 0.4797, | |
| "step": 3330 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "grad_norm": 0.8416649699211121, | |
| "learning_rate": 6.7272727272727275e-06, | |
| "loss": 0.3889, | |
| "step": 3340 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "grad_norm": 1.227149486541748, | |
| "learning_rate": 6.717171717171718e-06, | |
| "loss": 0.4677, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "grad_norm": 1.5126155614852905, | |
| "learning_rate": 6.707070707070707e-06, | |
| "loss": 0.4444, | |
| "step": 3360 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "grad_norm": 1.1622529029846191, | |
| "learning_rate": 6.6969696969696975e-06, | |
| "loss": 0.4372, | |
| "step": 3370 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "grad_norm": 0.975615918636322, | |
| "learning_rate": 6.686868686868687e-06, | |
| "loss": 0.4037, | |
| "step": 3380 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "grad_norm": 0.8992089033126831, | |
| "learning_rate": 6.676767676767677e-06, | |
| "loss": 0.3785, | |
| "step": 3390 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "grad_norm": 1.4439970254898071, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 0.5169, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "grad_norm": 0.9601142406463623, | |
| "learning_rate": 6.656565656565657e-06, | |
| "loss": 0.4396, | |
| "step": 3410 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "grad_norm": 1.0457966327667236, | |
| "learning_rate": 6.646464646464646e-06, | |
| "loss": 0.4065, | |
| "step": 3420 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "grad_norm": 1.0014325380325317, | |
| "learning_rate": 6.6363636363636375e-06, | |
| "loss": 0.4616, | |
| "step": 3430 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "grad_norm": 1.3321492671966553, | |
| "learning_rate": 6.626262626262627e-06, | |
| "loss": 0.4164, | |
| "step": 3440 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "grad_norm": 0.9500308036804199, | |
| "learning_rate": 6.616161616161617e-06, | |
| "loss": 0.4156, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "grad_norm": 1.1907135248184204, | |
| "learning_rate": 6.606060606060607e-06, | |
| "loss": 0.4353, | |
| "step": 3460 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "grad_norm": 1.7826789617538452, | |
| "learning_rate": 6.595959595959597e-06, | |
| "loss": 0.4764, | |
| "step": 3470 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "grad_norm": 1.0160069465637207, | |
| "learning_rate": 6.585858585858586e-06, | |
| "loss": 0.4594, | |
| "step": 3480 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "grad_norm": 1.0795533657073975, | |
| "learning_rate": 6.575757575757577e-06, | |
| "loss": 0.4557, | |
| "step": 3490 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "grad_norm": 1.2484716176986694, | |
| "learning_rate": 6.565656565656566e-06, | |
| "loss": 0.5057, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "grad_norm": 1.6144826412200928, | |
| "learning_rate": 6.555555555555556e-06, | |
| "loss": 0.4554, | |
| "step": 3510 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "grad_norm": 1.3575657606124878, | |
| "learning_rate": 6.545454545454546e-06, | |
| "loss": 0.3373, | |
| "step": 3520 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "grad_norm": 0.9021044373512268, | |
| "learning_rate": 6.535353535353536e-06, | |
| "loss": 0.4765, | |
| "step": 3530 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "grad_norm": 1.0576077699661255, | |
| "learning_rate": 6.525252525252526e-06, | |
| "loss": 0.4344, | |
| "step": 3540 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "grad_norm": 1.0687928199768066, | |
| "learning_rate": 6.515151515151516e-06, | |
| "loss": 0.5022, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "grad_norm": 0.9124734401702881, | |
| "learning_rate": 6.505050505050505e-06, | |
| "loss": 0.4133, | |
| "step": 3560 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "grad_norm": 0.8715507984161377, | |
| "learning_rate": 6.494949494949495e-06, | |
| "loss": 0.454, | |
| "step": 3570 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "grad_norm": 1.0889540910720825, | |
| "learning_rate": 6.484848484848485e-06, | |
| "loss": 0.443, | |
| "step": 3580 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "grad_norm": 0.9323639273643494, | |
| "learning_rate": 6.4747474747474745e-06, | |
| "loss": 0.4983, | |
| "step": 3590 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "grad_norm": 0.6946113109588623, | |
| "learning_rate": 6.464646464646466e-06, | |
| "loss": 0.4364, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "grad_norm": 1.1879271268844604, | |
| "learning_rate": 6.454545454545456e-06, | |
| "loss": 0.4816, | |
| "step": 3610 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "grad_norm": 1.337172031402588, | |
| "learning_rate": 6.444444444444445e-06, | |
| "loss": 0.4963, | |
| "step": 3620 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "grad_norm": 0.9384005069732666, | |
| "learning_rate": 6.434343434343436e-06, | |
| "loss": 0.5185, | |
| "step": 3630 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "grad_norm": 1.2666473388671875, | |
| "learning_rate": 6.424242424242425e-06, | |
| "loss": 0.389, | |
| "step": 3640 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "grad_norm": 1.037668228149414, | |
| "learning_rate": 6.4141414141414145e-06, | |
| "loss": 0.4529, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "grad_norm": 1.4652140140533447, | |
| "learning_rate": 6.404040404040405e-06, | |
| "loss": 0.3915, | |
| "step": 3660 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "grad_norm": 1.1823006868362427, | |
| "learning_rate": 6.393939393939394e-06, | |
| "loss": 0.5123, | |
| "step": 3670 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "grad_norm": 1.0706424713134766, | |
| "learning_rate": 6.3838383838383845e-06, | |
| "loss": 0.4989, | |
| "step": 3680 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "grad_norm": 0.9993262887001038, | |
| "learning_rate": 6.373737373737374e-06, | |
| "loss": 0.4268, | |
| "step": 3690 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "grad_norm": 0.9947906732559204, | |
| "learning_rate": 6.363636363636364e-06, | |
| "loss": 0.4246, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "grad_norm": 0.8563311696052551, | |
| "learning_rate": 6.353535353535354e-06, | |
| "loss": 0.4439, | |
| "step": 3710 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "grad_norm": 1.2519429922103882, | |
| "learning_rate": 6.343434343434344e-06, | |
| "loss": 0.3973, | |
| "step": 3720 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "grad_norm": 1.2925148010253906, | |
| "learning_rate": 6.333333333333333e-06, | |
| "loss": 0.4179, | |
| "step": 3730 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "grad_norm": 1.1731756925582886, | |
| "learning_rate": 6.323232323232324e-06, | |
| "loss": 0.4748, | |
| "step": 3740 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "grad_norm": 1.030536413192749, | |
| "learning_rate": 6.313131313131313e-06, | |
| "loss": 0.4131, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "grad_norm": 1.1279093027114868, | |
| "learning_rate": 6.303030303030303e-06, | |
| "loss": 0.4143, | |
| "step": 3760 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "grad_norm": 1.1602801084518433, | |
| "learning_rate": 6.292929292929294e-06, | |
| "loss": 0.4718, | |
| "step": 3770 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "grad_norm": 1.5314010381698608, | |
| "learning_rate": 6.282828282828284e-06, | |
| "loss": 0.5463, | |
| "step": 3780 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "grad_norm": 1.6124476194381714, | |
| "learning_rate": 6.2727272727272734e-06, | |
| "loss": 0.4857, | |
| "step": 3790 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "grad_norm": 1.185355305671692, | |
| "learning_rate": 6.262626262626264e-06, | |
| "loss": 0.4167, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "grad_norm": 1.1900933980941772, | |
| "learning_rate": 6.252525252525253e-06, | |
| "loss": 0.4658, | |
| "step": 3810 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "grad_norm": 1.2930880784988403, | |
| "learning_rate": 6.2424242424242434e-06, | |
| "loss": 0.4912, | |
| "step": 3820 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "grad_norm": 1.1819838285446167, | |
| "learning_rate": 6.232323232323233e-06, | |
| "loss": 0.4129, | |
| "step": 3830 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "grad_norm": 1.315292477607727, | |
| "learning_rate": 6.222222222222223e-06, | |
| "loss": 0.477, | |
| "step": 3840 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "grad_norm": 1.007524013519287, | |
| "learning_rate": 6.212121212121213e-06, | |
| "loss": 0.414, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "grad_norm": 1.1948471069335938, | |
| "learning_rate": 6.202020202020203e-06, | |
| "loss": 0.4576, | |
| "step": 3860 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "grad_norm": 1.3339303731918335, | |
| "learning_rate": 6.191919191919192e-06, | |
| "loss": 0.5547, | |
| "step": 3870 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "grad_norm": 1.2568715810775757, | |
| "learning_rate": 6.181818181818182e-06, | |
| "loss": 0.3941, | |
| "step": 3880 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "grad_norm": 1.2084232568740845, | |
| "learning_rate": 6.171717171717172e-06, | |
| "loss": 0.462, | |
| "step": 3890 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "grad_norm": 1.0328818559646606, | |
| "learning_rate": 6.1616161616161615e-06, | |
| "loss": 0.4484, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "grad_norm": 1.9673748016357422, | |
| "learning_rate": 6.151515151515152e-06, | |
| "loss": 0.4543, | |
| "step": 3910 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "grad_norm": 1.4105424880981445, | |
| "learning_rate": 6.141414141414141e-06, | |
| "loss": 0.4964, | |
| "step": 3920 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "grad_norm": 0.7082942128181458, | |
| "learning_rate": 6.1313131313131315e-06, | |
| "loss": 0.5297, | |
| "step": 3930 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "grad_norm": 1.2604327201843262, | |
| "learning_rate": 6.121212121212121e-06, | |
| "loss": 0.4207, | |
| "step": 3940 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "grad_norm": 0.999332070350647, | |
| "learning_rate": 6.111111111111112e-06, | |
| "loss": 0.4791, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "grad_norm": 1.139615535736084, | |
| "learning_rate": 6.1010101010101015e-06, | |
| "loss": 0.4281, | |
| "step": 3960 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "grad_norm": 1.1164727210998535, | |
| "learning_rate": 6.090909090909092e-06, | |
| "loss": 0.3807, | |
| "step": 3970 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "grad_norm": 1.0829392671585083, | |
| "learning_rate": 6.080808080808081e-06, | |
| "loss": 0.4996, | |
| "step": 3980 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "grad_norm": 0.8551573753356934, | |
| "learning_rate": 6.0707070707070715e-06, | |
| "loss": 0.4759, | |
| "step": 3990 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "grad_norm": 1.224134087562561, | |
| "learning_rate": 6.060606060606061e-06, | |
| "loss": 0.4244, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "eval_loss": 0.5992664694786072, | |
| "eval_runtime": 334.8169, | |
| "eval_samples_per_second": 2.987, | |
| "eval_steps_per_second": 2.987, | |
| "step": 4000 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 10000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 4, | |
| "save_steps": 2000, | |
| "total_flos": 3.26411004936192e+17, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |