| { | |
| "best_metric": 0.12016157805919647, | |
| "best_model_checkpoint": "autotrain-ledky-lrgbn/checkpoint-13926", | |
| "epoch": 3.0, | |
| "eval_steps": 500, | |
| "global_step": 13926, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.005385609651012494, | |
| "grad_norm": 1.214734673500061, | |
| "learning_rate": 8.973438621679828e-07, | |
| "loss": 0.696, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.010771219302024989, | |
| "grad_norm": 0.9307985901832581, | |
| "learning_rate": 1.7946877243359655e-06, | |
| "loss": 0.6901, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.016156828953037484, | |
| "grad_norm": 1.196394443511963, | |
| "learning_rate": 2.6920315865039484e-06, | |
| "loss": 0.6818, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.021542438604049977, | |
| "grad_norm": 1.0026447772979736, | |
| "learning_rate": 3.589375448671931e-06, | |
| "loss": 0.6614, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.026928048255062474, | |
| "grad_norm": 1.2615079879760742, | |
| "learning_rate": 4.486719310839914e-06, | |
| "loss": 0.6209, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.03231365790607497, | |
| "grad_norm": 2.165177583694458, | |
| "learning_rate": 5.384063173007897e-06, | |
| "loss": 0.5555, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.03769926755708746, | |
| "grad_norm": 1.6065446138381958, | |
| "learning_rate": 6.2814070351758795e-06, | |
| "loss": 0.4827, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.043084877208099955, | |
| "grad_norm": 1.00612473487854, | |
| "learning_rate": 7.178750897343862e-06, | |
| "loss": 0.415, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.048470486859112455, | |
| "grad_norm": 1.8842555284500122, | |
| "learning_rate": 8.076094759511846e-06, | |
| "loss": 0.3724, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.05385609651012495, | |
| "grad_norm": 1.875260353088379, | |
| "learning_rate": 8.973438621679828e-06, | |
| "loss": 0.2979, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.05924170616113744, | |
| "grad_norm": 2.970935106277466, | |
| "learning_rate": 9.870782483847811e-06, | |
| "loss": 0.2478, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.06462731581214994, | |
| "grad_norm": 0.5585727691650391, | |
| "learning_rate": 1.0768126346015794e-05, | |
| "loss": 0.2456, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.07001292546316243, | |
| "grad_norm": 0.5557870864868164, | |
| "learning_rate": 1.1665470208183776e-05, | |
| "loss": 0.2346, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.07539853511417492, | |
| "grad_norm": 5.3282365798950195, | |
| "learning_rate": 1.2562814070351759e-05, | |
| "loss": 0.2192, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.08078414476518742, | |
| "grad_norm": 0.8856891989707947, | |
| "learning_rate": 1.3460157932519743e-05, | |
| "loss": 0.2935, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.08616975441619991, | |
| "grad_norm": 0.9343020915985107, | |
| "learning_rate": 1.4357501794687724e-05, | |
| "loss": 0.2117, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.0915553640672124, | |
| "grad_norm": 8.641404151916504, | |
| "learning_rate": 1.5254845656855707e-05, | |
| "loss": 0.2256, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.09694097371822491, | |
| "grad_norm": 0.31070542335510254, | |
| "learning_rate": 1.615218951902369e-05, | |
| "loss": 0.2155, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.1023265833692374, | |
| "grad_norm": 2.8748903274536133, | |
| "learning_rate": 1.7049533381191674e-05, | |
| "loss": 0.2272, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.1077121930202499, | |
| "grad_norm": 0.5525055527687073, | |
| "learning_rate": 1.7946877243359657e-05, | |
| "loss": 0.3538, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.11309780267126239, | |
| "grad_norm": 23.755407333374023, | |
| "learning_rate": 1.884422110552764e-05, | |
| "loss": 0.1872, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.11848341232227488, | |
| "grad_norm": 34.48551559448242, | |
| "learning_rate": 1.9741564967695622e-05, | |
| "loss": 0.1676, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.12386902197328738, | |
| "grad_norm": 1.0720394849777222, | |
| "learning_rate": 2.0638908829863605e-05, | |
| "loss": 0.258, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.12925463162429987, | |
| "grad_norm": 0.2333178073167801, | |
| "learning_rate": 2.1536252692031587e-05, | |
| "loss": 0.2356, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.13464024127531238, | |
| "grad_norm": 28.830148696899414, | |
| "learning_rate": 2.243359655419957e-05, | |
| "loss": 0.2483, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.14002585092632486, | |
| "grad_norm": 0.38484323024749756, | |
| "learning_rate": 2.3330940416367553e-05, | |
| "loss": 0.2576, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.14541146057733736, | |
| "grad_norm": 9.458002090454102, | |
| "learning_rate": 2.4228284278535535e-05, | |
| "loss": 0.2542, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.15079707022834984, | |
| "grad_norm": 0.28468650579452515, | |
| "learning_rate": 2.5125628140703518e-05, | |
| "loss": 0.2451, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.15618267987936235, | |
| "grad_norm": 4.210727691650391, | |
| "learning_rate": 2.6022972002871504e-05, | |
| "loss": 0.2512, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.16156828953037483, | |
| "grad_norm": 0.2801736295223236, | |
| "learning_rate": 2.6920315865039487e-05, | |
| "loss": 0.2245, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.16695389918138734, | |
| "grad_norm": 0.18515126407146454, | |
| "learning_rate": 2.781765972720747e-05, | |
| "loss": 0.2455, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.17233950883239982, | |
| "grad_norm": 24.687742233276367, | |
| "learning_rate": 2.871500358937545e-05, | |
| "loss": 0.1717, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.17772511848341233, | |
| "grad_norm": 0.31296807527542114, | |
| "learning_rate": 2.961234745154343e-05, | |
| "loss": 0.3381, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.1831107281344248, | |
| "grad_norm": 0.2845638692378998, | |
| "learning_rate": 3.0509691313711414e-05, | |
| "loss": 0.2039, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.1884963377854373, | |
| "grad_norm": 9.822659492492676, | |
| "learning_rate": 3.14070351758794e-05, | |
| "loss": 0.2695, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.19388194743644982, | |
| "grad_norm": 11.07955265045166, | |
| "learning_rate": 3.230437903804738e-05, | |
| "loss": 0.1941, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.1992675570874623, | |
| "grad_norm": 0.13381847739219666, | |
| "learning_rate": 3.3201722900215365e-05, | |
| "loss": 0.1602, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.2046531667384748, | |
| "grad_norm": 0.3931732177734375, | |
| "learning_rate": 3.409906676238335e-05, | |
| "loss": 0.1306, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.2100387763894873, | |
| "grad_norm": 59.89822006225586, | |
| "learning_rate": 3.499641062455133e-05, | |
| "loss": 0.218, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.2154243860404998, | |
| "grad_norm": 2.6214139461517334, | |
| "learning_rate": 3.5893754486719313e-05, | |
| "loss": 0.1886, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.22080999569151227, | |
| "grad_norm": 1.7539465427398682, | |
| "learning_rate": 3.6791098348887296e-05, | |
| "loss": 0.1999, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.22619560534252478, | |
| "grad_norm": 0.13422812521457672, | |
| "learning_rate": 3.768844221105528e-05, | |
| "loss": 0.1967, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.23158121499353726, | |
| "grad_norm": 17.16714859008789, | |
| "learning_rate": 3.858578607322326e-05, | |
| "loss": 0.2197, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.23696682464454977, | |
| "grad_norm": 4.463374614715576, | |
| "learning_rate": 3.9483129935391244e-05, | |
| "loss": 0.1428, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.24235243429556225, | |
| "grad_norm": 14.286873817443848, | |
| "learning_rate": 4.038047379755923e-05, | |
| "loss": 0.154, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.24773804394657475, | |
| "grad_norm": 1.0160866975784302, | |
| "learning_rate": 4.127781765972721e-05, | |
| "loss": 0.1008, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.25312365359758726, | |
| "grad_norm": 0.08231418579816818, | |
| "learning_rate": 4.217516152189519e-05, | |
| "loss": 0.1716, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.25850926324859974, | |
| "grad_norm": 0.4267788231372833, | |
| "learning_rate": 4.3072505384063175e-05, | |
| "loss": 0.2282, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.2638948728996122, | |
| "grad_norm": 8.705976486206055, | |
| "learning_rate": 4.396984924623116e-05, | |
| "loss": 0.1687, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.26928048255062476, | |
| "grad_norm": 0.08920421451330185, | |
| "learning_rate": 4.486719310839914e-05, | |
| "loss": 0.1566, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.27466609220163724, | |
| "grad_norm": 136.6876220703125, | |
| "learning_rate": 4.576453697056712e-05, | |
| "loss": 0.3009, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.2800517018526497, | |
| "grad_norm": 97.63652038574219, | |
| "learning_rate": 4.6661880832735106e-05, | |
| "loss": 0.1277, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.2854373115036622, | |
| "grad_norm": 0.1028209999203682, | |
| "learning_rate": 4.755922469490309e-05, | |
| "loss": 0.1079, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.29082292115467473, | |
| "grad_norm": 0.1444273144006729, | |
| "learning_rate": 4.845656855707107e-05, | |
| "loss": 0.2049, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.2962085308056872, | |
| "grad_norm": 0.07809091359376907, | |
| "learning_rate": 4.9353912419239054e-05, | |
| "loss": 0.1565, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.3015941404566997, | |
| "grad_norm": 27.804719924926758, | |
| "learning_rate": 4.9972073725365035e-05, | |
| "loss": 0.1187, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.30697975010771217, | |
| "grad_norm": 2.5913608074188232, | |
| "learning_rate": 4.987233703024017e-05, | |
| "loss": 0.2541, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.3123653597587247, | |
| "grad_norm": 0.08927720040082932, | |
| "learning_rate": 4.97726003351153e-05, | |
| "loss": 0.1258, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.3177509694097372, | |
| "grad_norm": 11.248839378356934, | |
| "learning_rate": 4.967286363999043e-05, | |
| "loss": 0.2244, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 0.32313657906074966, | |
| "grad_norm": 4.807493686676025, | |
| "learning_rate": 4.957312694486556e-05, | |
| "loss": 0.1561, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.3285221887117622, | |
| "grad_norm": 2.7352848052978516, | |
| "learning_rate": 4.9473390249740684e-05, | |
| "loss": 0.2706, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 0.3339077983627747, | |
| "grad_norm": 2.721325159072876, | |
| "learning_rate": 4.937365355461582e-05, | |
| "loss": 0.1181, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.33929340801378716, | |
| "grad_norm": 0.08835005760192871, | |
| "learning_rate": 4.927391685949095e-05, | |
| "loss": 0.0849, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 0.34467901766479964, | |
| "grad_norm": 0.06279947608709335, | |
| "learning_rate": 4.917418016436607e-05, | |
| "loss": 0.0603, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.35006462731581217, | |
| "grad_norm": 5.033148765563965, | |
| "learning_rate": 4.9074443469241206e-05, | |
| "loss": 0.1982, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 0.35545023696682465, | |
| "grad_norm": 19.948888778686523, | |
| "learning_rate": 4.897470677411633e-05, | |
| "loss": 0.1951, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.36083584661783713, | |
| "grad_norm": 0.15091180801391602, | |
| "learning_rate": 4.887497007899147e-05, | |
| "loss": 0.1967, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 0.3662214562688496, | |
| "grad_norm": 11.530139923095703, | |
| "learning_rate": 4.8775233383866594e-05, | |
| "loss": 0.2055, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.37160706591986215, | |
| "grad_norm": 1.3459250926971436, | |
| "learning_rate": 4.867549668874172e-05, | |
| "loss": 0.165, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 0.3769926755708746, | |
| "grad_norm": 0.1419544219970703, | |
| "learning_rate": 4.8575759993616855e-05, | |
| "loss": 0.2761, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.3823782852218871, | |
| "grad_norm": 0.0771302655339241, | |
| "learning_rate": 4.847602329849198e-05, | |
| "loss": 0.1745, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 0.38776389487289964, | |
| "grad_norm": 7.464870452880859, | |
| "learning_rate": 4.8376286603367116e-05, | |
| "loss": 0.1824, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.3931495045239121, | |
| "grad_norm": 8.839171409606934, | |
| "learning_rate": 4.827654990824224e-05, | |
| "loss": 0.2336, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 0.3985351141749246, | |
| "grad_norm": 2.038132667541504, | |
| "learning_rate": 4.817681321311737e-05, | |
| "loss": 0.3063, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.4039207238259371, | |
| "grad_norm": 1.8243613243103027, | |
| "learning_rate": 4.8077076517992504e-05, | |
| "loss": 0.1202, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 0.4093063334769496, | |
| "grad_norm": 0.09693591296672821, | |
| "learning_rate": 4.797733982286763e-05, | |
| "loss": 0.1442, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.4146919431279621, | |
| "grad_norm": 4.965816020965576, | |
| "learning_rate": 4.787760312774276e-05, | |
| "loss": 0.2029, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 0.4200775527789746, | |
| "grad_norm": 0.06871023029088974, | |
| "learning_rate": 4.777786643261789e-05, | |
| "loss": 0.0826, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.42546316242998705, | |
| "grad_norm": 0.0709209218621254, | |
| "learning_rate": 4.767812973749302e-05, | |
| "loss": 0.1575, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 0.4308487720809996, | |
| "grad_norm": 1.0321829319000244, | |
| "learning_rate": 4.7578393042368154e-05, | |
| "loss": 0.1481, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.43623438173201207, | |
| "grad_norm": 0.05471009761095047, | |
| "learning_rate": 4.747865634724328e-05, | |
| "loss": 0.0653, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 0.44161999138302455, | |
| "grad_norm": 0.5579153299331665, | |
| "learning_rate": 4.737891965211841e-05, | |
| "loss": 0.1818, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.447005601034037, | |
| "grad_norm": 7.742904186248779, | |
| "learning_rate": 4.727918295699354e-05, | |
| "loss": 0.1663, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 0.45239121068504956, | |
| "grad_norm": 0.26534390449523926, | |
| "learning_rate": 4.717944626186867e-05, | |
| "loss": 0.2548, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.45777682033606204, | |
| "grad_norm": 0.060189589858055115, | |
| "learning_rate": 4.7079709566743796e-05, | |
| "loss": 0.1214, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 0.4631624299870745, | |
| "grad_norm": 2.2301554679870605, | |
| "learning_rate": 4.697997287161893e-05, | |
| "loss": 0.1254, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.46854803963808705, | |
| "grad_norm": 31.325176239013672, | |
| "learning_rate": 4.688023617649406e-05, | |
| "loss": 0.2186, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 0.47393364928909953, | |
| "grad_norm": 0.8820040822029114, | |
| "learning_rate": 4.678049948136919e-05, | |
| "loss": 0.1814, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.479319258940112, | |
| "grad_norm": 2.4866137504577637, | |
| "learning_rate": 4.668076278624432e-05, | |
| "loss": 0.1301, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 0.4847048685911245, | |
| "grad_norm": 0.0820331797003746, | |
| "learning_rate": 4.6581026091119445e-05, | |
| "loss": 0.1845, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.49009047824213703, | |
| "grad_norm": 36.052513122558594, | |
| "learning_rate": 4.648128939599458e-05, | |
| "loss": 0.1782, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 0.4954760878931495, | |
| "grad_norm": 0.08009187877178192, | |
| "learning_rate": 4.6381552700869706e-05, | |
| "loss": 0.1326, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.500861697544162, | |
| "grad_norm": 1.324997067451477, | |
| "learning_rate": 4.628181600574484e-05, | |
| "loss": 0.2862, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 0.5062473071951745, | |
| "grad_norm": 0.17933067679405212, | |
| "learning_rate": 4.618207931061996e-05, | |
| "loss": 0.1677, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.511632916846187, | |
| "grad_norm": 0.22597669064998627, | |
| "learning_rate": 4.6082342615495094e-05, | |
| "loss": 0.2077, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 0.5170185264971995, | |
| "grad_norm": 1.9283539056777954, | |
| "learning_rate": 4.598260592037023e-05, | |
| "loss": 0.227, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.522404136148212, | |
| "grad_norm": 0.21102119982242584, | |
| "learning_rate": 4.5882869225245355e-05, | |
| "loss": 0.2177, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 0.5277897457992244, | |
| "grad_norm": 16.233251571655273, | |
| "learning_rate": 4.578313253012048e-05, | |
| "loss": 0.1349, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 0.533175355450237, | |
| "grad_norm": 12.861350059509277, | |
| "learning_rate": 4.568339583499561e-05, | |
| "loss": 0.2012, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 0.5385609651012495, | |
| "grad_norm": 29.975788116455078, | |
| "learning_rate": 4.5583659139870744e-05, | |
| "loss": 0.1332, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.5439465747522619, | |
| "grad_norm": 0.08081357926130295, | |
| "learning_rate": 4.548392244474588e-05, | |
| "loss": 0.1514, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 0.5493321844032745, | |
| "grad_norm": 72.80885314941406, | |
| "learning_rate": 4.5384185749621005e-05, | |
| "loss": 0.2106, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 0.5547177940542869, | |
| "grad_norm": 0.0727340430021286, | |
| "learning_rate": 4.528444905449613e-05, | |
| "loss": 0.1395, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 0.5601034037052994, | |
| "grad_norm": 0.10599514842033386, | |
| "learning_rate": 4.518471235937126e-05, | |
| "loss": 0.1318, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.565489013356312, | |
| "grad_norm": 0.09759501367807388, | |
| "learning_rate": 4.508497566424639e-05, | |
| "loss": 0.0796, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 0.5708746230073244, | |
| "grad_norm": 0.15663489699363708, | |
| "learning_rate": 4.4985238969121527e-05, | |
| "loss": 0.1019, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 0.5762602326583369, | |
| "grad_norm": 0.13662685453891754, | |
| "learning_rate": 4.488550227399665e-05, | |
| "loss": 0.1396, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 0.5816458423093495, | |
| "grad_norm": 0.10818319767713547, | |
| "learning_rate": 4.478576557887178e-05, | |
| "loss": 0.1455, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.5870314519603619, | |
| "grad_norm": 0.11137444525957108, | |
| "learning_rate": 4.468602888374691e-05, | |
| "loss": 0.1616, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 0.5924170616113744, | |
| "grad_norm": 0.1452583223581314, | |
| "learning_rate": 4.458629218862204e-05, | |
| "loss": 0.1445, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 0.597802671262387, | |
| "grad_norm": 10.563451766967773, | |
| "learning_rate": 4.448655549349717e-05, | |
| "loss": 0.1514, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 0.6031882809133994, | |
| "grad_norm": 14.584248542785645, | |
| "learning_rate": 4.4386818798372296e-05, | |
| "loss": 0.1717, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.6085738905644119, | |
| "grad_norm": 12.885149955749512, | |
| "learning_rate": 4.428708210324743e-05, | |
| "loss": 0.2395, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 0.6139595002154243, | |
| "grad_norm": 2.0140767097473145, | |
| "learning_rate": 4.418734540812256e-05, | |
| "loss": 0.2966, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 0.6193451098664369, | |
| "grad_norm": 0.11748749762773514, | |
| "learning_rate": 4.4087608712997684e-05, | |
| "loss": 0.083, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 0.6247307195174494, | |
| "grad_norm": 0.058700062334537506, | |
| "learning_rate": 4.398787201787282e-05, | |
| "loss": 0.1189, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.6301163291684618, | |
| "grad_norm": 1.4539530277252197, | |
| "learning_rate": 4.3888135322747945e-05, | |
| "loss": 0.1992, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 0.6355019388194744, | |
| "grad_norm": 0.36291152238845825, | |
| "learning_rate": 4.378839862762308e-05, | |
| "loss": 0.0857, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 0.6408875484704869, | |
| "grad_norm": 1.1221033334732056, | |
| "learning_rate": 4.3688661932498206e-05, | |
| "loss": 0.2207, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 0.6462731581214993, | |
| "grad_norm": 1.3202532529830933, | |
| "learning_rate": 4.3588925237373333e-05, | |
| "loss": 0.1268, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.6516587677725119, | |
| "grad_norm": 0.23690970242023468, | |
| "learning_rate": 4.348918854224847e-05, | |
| "loss": 0.2074, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 0.6570443774235244, | |
| "grad_norm": 0.39078834652900696, | |
| "learning_rate": 4.3389451847123595e-05, | |
| "loss": 0.0893, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 0.6624299870745368, | |
| "grad_norm": 2.198420286178589, | |
| "learning_rate": 4.328971515199873e-05, | |
| "loss": 0.1987, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 0.6678155967255494, | |
| "grad_norm": 0.14714717864990234, | |
| "learning_rate": 4.3189978456873856e-05, | |
| "loss": 0.1911, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 0.6732012063765618, | |
| "grad_norm": 0.07894087582826614, | |
| "learning_rate": 4.309024176174898e-05, | |
| "loss": 0.1334, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 0.6785868160275743, | |
| "grad_norm": 0.11621283739805222, | |
| "learning_rate": 4.2990505066624117e-05, | |
| "loss": 0.0956, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 0.6839724256785868, | |
| "grad_norm": 0.15330371260643005, | |
| "learning_rate": 4.2890768371499244e-05, | |
| "loss": 0.1856, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 0.6893580353295993, | |
| "grad_norm": 100.52692413330078, | |
| "learning_rate": 4.279103167637437e-05, | |
| "loss": 0.1754, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.6947436449806118, | |
| "grad_norm": 1.5307726860046387, | |
| "learning_rate": 4.2691294981249505e-05, | |
| "loss": 0.1644, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 0.7001292546316243, | |
| "grad_norm": 0.1815415918827057, | |
| "learning_rate": 4.259155828612463e-05, | |
| "loss": 0.1715, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 0.7055148642826368, | |
| "grad_norm": 0.06501147896051407, | |
| "learning_rate": 4.2491821590999766e-05, | |
| "loss": 0.0786, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 0.7109004739336493, | |
| "grad_norm": 13.024656295776367, | |
| "learning_rate": 4.239208489587489e-05, | |
| "loss": 0.1299, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 0.7162860835846618, | |
| "grad_norm": 0.061265505850315094, | |
| "learning_rate": 4.229234820075002e-05, | |
| "loss": 0.1553, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 0.7216716932356743, | |
| "grad_norm": 0.15151949226856232, | |
| "learning_rate": 4.2192611505625154e-05, | |
| "loss": 0.1563, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 0.7270573028866868, | |
| "grad_norm": 0.37678295373916626, | |
| "learning_rate": 4.209287481050028e-05, | |
| "loss": 0.1824, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 0.7324429125376992, | |
| "grad_norm": 0.22883647680282593, | |
| "learning_rate": 4.1993138115375415e-05, | |
| "loss": 0.2303, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 0.7378285221887118, | |
| "grad_norm": 0.123573899269104, | |
| "learning_rate": 4.189340142025054e-05, | |
| "loss": 0.1554, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 0.7432141318397243, | |
| "grad_norm": 7.3934431076049805, | |
| "learning_rate": 4.179366472512567e-05, | |
| "loss": 0.1949, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 0.7485997414907367, | |
| "grad_norm": 0.07116091251373291, | |
| "learning_rate": 4.16939280300008e-05, | |
| "loss": 0.1022, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 0.7539853511417492, | |
| "grad_norm": 0.6062744855880737, | |
| "learning_rate": 4.159419133487593e-05, | |
| "loss": 0.1311, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.7593709607927618, | |
| "grad_norm": 0.19308426976203918, | |
| "learning_rate": 4.149445463975106e-05, | |
| "loss": 0.1942, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 0.7647565704437742, | |
| "grad_norm": 0.08045301586389542, | |
| "learning_rate": 4.139471794462619e-05, | |
| "loss": 0.1055, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 0.7701421800947867, | |
| "grad_norm": 47.74348449707031, | |
| "learning_rate": 4.129498124950132e-05, | |
| "loss": 0.0844, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 0.7755277897457993, | |
| "grad_norm": 14.923918724060059, | |
| "learning_rate": 4.119524455437645e-05, | |
| "loss": 0.0943, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.7809133993968117, | |
| "grad_norm": 0.3901768624782562, | |
| "learning_rate": 4.109550785925157e-05, | |
| "loss": 0.2203, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 0.7862990090478242, | |
| "grad_norm": 0.05978202819824219, | |
| "learning_rate": 4.0995771164126707e-05, | |
| "loss": 0.0804, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 0.7916846186988367, | |
| "grad_norm": 8.785200119018555, | |
| "learning_rate": 4.089603446900184e-05, | |
| "loss": 0.1765, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 0.7970702283498492, | |
| "grad_norm": 1.523232102394104, | |
| "learning_rate": 4.079629777387697e-05, | |
| "loss": 0.1302, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 0.8024558380008617, | |
| "grad_norm": 0.05809938907623291, | |
| "learning_rate": 4.06965610787521e-05, | |
| "loss": 0.0481, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 0.8078414476518742, | |
| "grad_norm": 0.05040917173027992, | |
| "learning_rate": 4.059682438362722e-05, | |
| "loss": 0.0768, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 0.8132270573028867, | |
| "grad_norm": 0.07563222199678421, | |
| "learning_rate": 4.0497087688502356e-05, | |
| "loss": 0.1397, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 0.8186126669538992, | |
| "grad_norm": 0.4528738260269165, | |
| "learning_rate": 4.039735099337749e-05, | |
| "loss": 0.3308, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 0.8239982766049117, | |
| "grad_norm": 1.0809814929962158, | |
| "learning_rate": 4.029761429825262e-05, | |
| "loss": 0.1919, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 0.8293838862559242, | |
| "grad_norm": 0.2343398779630661, | |
| "learning_rate": 4.0197877603127744e-05, | |
| "loss": 0.1351, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 0.8347694959069367, | |
| "grad_norm": 1.747331976890564, | |
| "learning_rate": 4.009814090800287e-05, | |
| "loss": 0.1744, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 0.8401551055579491, | |
| "grad_norm": 0.06216764450073242, | |
| "learning_rate": 3.9998404212878005e-05, | |
| "loss": 0.0587, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 0.8455407152089617, | |
| "grad_norm": 0.33079928159713745, | |
| "learning_rate": 3.989866751775314e-05, | |
| "loss": 0.1353, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 0.8509263248599741, | |
| "grad_norm": 1.1662975549697876, | |
| "learning_rate": 3.979893082262826e-05, | |
| "loss": 0.2162, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 0.8563119345109866, | |
| "grad_norm": 0.2112116664648056, | |
| "learning_rate": 3.969919412750339e-05, | |
| "loss": 0.1332, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 0.8616975441619992, | |
| "grad_norm": 0.2449023723602295, | |
| "learning_rate": 3.959945743237852e-05, | |
| "loss": 0.1461, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.8670831538130116, | |
| "grad_norm": 0.17366133630275726, | |
| "learning_rate": 3.9499720737253654e-05, | |
| "loss": 0.0463, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 0.8724687634640241, | |
| "grad_norm": 5.110925674438477, | |
| "learning_rate": 3.939998404212878e-05, | |
| "loss": 0.2415, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 0.8778543731150367, | |
| "grad_norm": 0.30593565106391907, | |
| "learning_rate": 3.930024734700391e-05, | |
| "loss": 0.1456, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 0.8832399827660491, | |
| "grad_norm": 0.19217516481876373, | |
| "learning_rate": 3.920051065187904e-05, | |
| "loss": 0.161, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 0.8886255924170616, | |
| "grad_norm": 1.0149356126785278, | |
| "learning_rate": 3.910077395675417e-05, | |
| "loss": 0.1787, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 0.894011202068074, | |
| "grad_norm": 5.444819450378418, | |
| "learning_rate": 3.90010372616293e-05, | |
| "loss": 0.1539, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 0.8993968117190866, | |
| "grad_norm": 1.2180671691894531, | |
| "learning_rate": 3.890130056650443e-05, | |
| "loss": 0.115, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 0.9047824213700991, | |
| "grad_norm": 0.16885504126548767, | |
| "learning_rate": 3.880156387137956e-05, | |
| "loss": 0.2378, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 0.9101680310211115, | |
| "grad_norm": 0.19795845448970795, | |
| "learning_rate": 3.870182717625469e-05, | |
| "loss": 0.1481, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 0.9155536406721241, | |
| "grad_norm": 0.2421010136604309, | |
| "learning_rate": 3.860209048112982e-05, | |
| "loss": 0.1496, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 0.9209392503231366, | |
| "grad_norm": 6.898841857910156, | |
| "learning_rate": 3.8502353786004946e-05, | |
| "loss": 0.1823, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 0.926324859974149, | |
| "grad_norm": 0.2529371678829193, | |
| "learning_rate": 3.840261709088008e-05, | |
| "loss": 0.2487, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 0.9317104696251616, | |
| "grad_norm": 0.39406079053878784, | |
| "learning_rate": 3.830288039575521e-05, | |
| "loss": 0.2116, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 0.9370960792761741, | |
| "grad_norm": 0.07333202660083771, | |
| "learning_rate": 3.820314370063034e-05, | |
| "loss": 0.0279, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 0.9424816889271865, | |
| "grad_norm": 1.0374135971069336, | |
| "learning_rate": 3.810340700550547e-05, | |
| "loss": 0.1611, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 0.9478672985781991, | |
| "grad_norm": 0.12242020666599274, | |
| "learning_rate": 3.8003670310380595e-05, | |
| "loss": 0.1268, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 0.9532529082292115, | |
| "grad_norm": 0.7784824371337891, | |
| "learning_rate": 3.790393361525573e-05, | |
| "loss": 0.2008, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 0.958638517880224, | |
| "grad_norm": 0.3692557215690613, | |
| "learning_rate": 3.7804196920130856e-05, | |
| "loss": 0.1814, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 0.9640241275312366, | |
| "grad_norm": 0.30838918685913086, | |
| "learning_rate": 3.770446022500599e-05, | |
| "loss": 0.0937, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 0.969409737182249, | |
| "grad_norm": 8.506460189819336, | |
| "learning_rate": 3.760472352988112e-05, | |
| "loss": 0.2372, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.9747953468332615, | |
| "grad_norm": 0.11787492781877518, | |
| "learning_rate": 3.7504986834756244e-05, | |
| "loss": 0.0906, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 0.9801809564842741, | |
| "grad_norm": 0.18946515023708344, | |
| "learning_rate": 3.740525013963138e-05, | |
| "loss": 0.2561, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 0.9855665661352865, | |
| "grad_norm": 0.10240988433361053, | |
| "learning_rate": 3.7305513444506505e-05, | |
| "loss": 0.087, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 0.990952175786299, | |
| "grad_norm": 2.1701135635375977, | |
| "learning_rate": 3.720577674938163e-05, | |
| "loss": 0.1067, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 0.9963377854373116, | |
| "grad_norm": 0.10697437822818756, | |
| "learning_rate": 3.7106040054256766e-05, | |
| "loss": 0.1298, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.9611117095766455, | |
| "eval_auc": 0.9901169654651506, | |
| "eval_f1": 0.961078167115903, | |
| "eval_loss": 0.13669981062412262, | |
| "eval_precision": 0.9587007958700796, | |
| "eval_recall": 0.9634673584089927, | |
| "eval_runtime": 1681.5869, | |
| "eval_samples_per_second": 5.52, | |
| "eval_steps_per_second": 0.346, | |
| "step": 4642 | |
| }, | |
| { | |
| "epoch": 1.001723395088324, | |
| "grad_norm": 2.8891570568084717, | |
| "learning_rate": 3.700630335913189e-05, | |
| "loss": 0.2125, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 1.0071090047393365, | |
| "grad_norm": 10.658247947692871, | |
| "learning_rate": 3.690656666400703e-05, | |
| "loss": 0.1635, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 1.012494614390349, | |
| "grad_norm": 0.06771805137395859, | |
| "learning_rate": 3.680682996888215e-05, | |
| "loss": 0.0609, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 1.0178802240413616, | |
| "grad_norm": 0.10092346370220184, | |
| "learning_rate": 3.670709327375728e-05, | |
| "loss": 0.1142, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 1.023265833692374, | |
| "grad_norm": 0.09358137845993042, | |
| "learning_rate": 3.6607356578632415e-05, | |
| "loss": 0.134, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 1.0286514433433864, | |
| "grad_norm": 0.10109369456768036, | |
| "learning_rate": 3.650761988350754e-05, | |
| "loss": 0.0835, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 1.034037052994399, | |
| "grad_norm": 0.06669428944587708, | |
| "learning_rate": 3.640788318838267e-05, | |
| "loss": 0.0729, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 1.0394226626454115, | |
| "grad_norm": 0.26679906249046326, | |
| "learning_rate": 3.63081464932578e-05, | |
| "loss": 0.1923, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 1.044808272296424, | |
| "grad_norm": 0.1384529024362564, | |
| "learning_rate": 3.620840979813293e-05, | |
| "loss": 0.0301, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 1.0501938819474363, | |
| "grad_norm": 0.04985777288675308, | |
| "learning_rate": 3.6108673103008065e-05, | |
| "loss": 0.0582, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 1.0555794915984489, | |
| "grad_norm": 0.17264176905155182, | |
| "learning_rate": 3.600893640788319e-05, | |
| "loss": 0.1751, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 1.0609651012494614, | |
| "grad_norm": 0.41739097237586975, | |
| "learning_rate": 3.590919971275832e-05, | |
| "loss": 0.0556, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 1.066350710900474, | |
| "grad_norm": 0.049522146582603455, | |
| "learning_rate": 3.5809463017633446e-05, | |
| "loss": 0.0233, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 1.0717363205514865, | |
| "grad_norm": 18.618080139160156, | |
| "learning_rate": 3.570972632250858e-05, | |
| "loss": 0.0875, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 1.077121930202499, | |
| "grad_norm": 0.055007074028253555, | |
| "learning_rate": 3.5609989627383714e-05, | |
| "loss": 0.0397, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.0825075398535113, | |
| "grad_norm": 0.05218919366598129, | |
| "learning_rate": 3.5510252932258834e-05, | |
| "loss": 0.1636, | |
| "step": 5025 | |
| }, | |
| { | |
| "epoch": 1.0878931495045239, | |
| "grad_norm": 0.13161726295948029, | |
| "learning_rate": 3.541051623713397e-05, | |
| "loss": 0.1354, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 1.0932787591555364, | |
| "grad_norm": 0.2851119041442871, | |
| "learning_rate": 3.5310779542009095e-05, | |
| "loss": 0.0301, | |
| "step": 5075 | |
| }, | |
| { | |
| "epoch": 1.098664368806549, | |
| "grad_norm": 0.03602718561887741, | |
| "learning_rate": 3.521104284688423e-05, | |
| "loss": 0.0037, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 1.1040499784575615, | |
| "grad_norm": 20.054950714111328, | |
| "learning_rate": 3.5111306151759356e-05, | |
| "loss": 0.1972, | |
| "step": 5125 | |
| }, | |
| { | |
| "epoch": 1.1094355881085738, | |
| "grad_norm": 100.69612121582031, | |
| "learning_rate": 3.501156945663448e-05, | |
| "loss": 0.0889, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 1.1148211977595863, | |
| "grad_norm": 11.003225326538086, | |
| "learning_rate": 3.491183276150962e-05, | |
| "loss": 0.1155, | |
| "step": 5175 | |
| }, | |
| { | |
| "epoch": 1.1202068074105989, | |
| "grad_norm": 8.291934967041016, | |
| "learning_rate": 3.4812096066384744e-05, | |
| "loss": 0.2039, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 1.1255924170616114, | |
| "grad_norm": 0.06259515136480331, | |
| "learning_rate": 3.471235937125988e-05, | |
| "loss": 0.1264, | |
| "step": 5225 | |
| }, | |
| { | |
| "epoch": 1.130978026712624, | |
| "grad_norm": 1.9804569482803345, | |
| "learning_rate": 3.4612622676135005e-05, | |
| "loss": 0.1603, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 1.1363636363636362, | |
| "grad_norm": 13.11237621307373, | |
| "learning_rate": 3.451288598101013e-05, | |
| "loss": 0.0895, | |
| "step": 5275 | |
| }, | |
| { | |
| "epoch": 1.1417492460146488, | |
| "grad_norm": 0.14050540328025818, | |
| "learning_rate": 3.4413149285885266e-05, | |
| "loss": 0.1879, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 1.1471348556656613, | |
| "grad_norm": 0.08870512247085571, | |
| "learning_rate": 3.4313412590760393e-05, | |
| "loss": 0.0668, | |
| "step": 5325 | |
| }, | |
| { | |
| "epoch": 1.1525204653166738, | |
| "grad_norm": 1.545466661453247, | |
| "learning_rate": 3.421367589563552e-05, | |
| "loss": 0.0765, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 1.1579060749676864, | |
| "grad_norm": 0.2768205404281616, | |
| "learning_rate": 3.4113939200510654e-05, | |
| "loss": 0.1389, | |
| "step": 5375 | |
| }, | |
| { | |
| "epoch": 1.163291684618699, | |
| "grad_norm": 15.847858428955078, | |
| "learning_rate": 3.401420250538578e-05, | |
| "loss": 0.1629, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 1.1686772942697115, | |
| "grad_norm": 0.10181305557489395, | |
| "learning_rate": 3.3914465810260916e-05, | |
| "loss": 0.2408, | |
| "step": 5425 | |
| }, | |
| { | |
| "epoch": 1.1740629039207238, | |
| "grad_norm": 1.0575032234191895, | |
| "learning_rate": 3.381472911513604e-05, | |
| "loss": 0.1404, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 1.1794485135717363, | |
| "grad_norm": 0.14517611265182495, | |
| "learning_rate": 3.371499242001117e-05, | |
| "loss": 0.1526, | |
| "step": 5475 | |
| }, | |
| { | |
| "epoch": 1.1848341232227488, | |
| "grad_norm": 3.418147087097168, | |
| "learning_rate": 3.3615255724886304e-05, | |
| "loss": 0.204, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 1.1902197328737614, | |
| "grad_norm": 0.12968988716602325, | |
| "learning_rate": 3.351551902976143e-05, | |
| "loss": 0.1302, | |
| "step": 5525 | |
| }, | |
| { | |
| "epoch": 1.195605342524774, | |
| "grad_norm": 0.49698755145072937, | |
| "learning_rate": 3.341578233463656e-05, | |
| "loss": 0.1656, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 1.2009909521757862, | |
| "grad_norm": 0.4227307140827179, | |
| "learning_rate": 3.331604563951169e-05, | |
| "loss": 0.1857, | |
| "step": 5575 | |
| }, | |
| { | |
| "epoch": 1.2063765618267988, | |
| "grad_norm": 1.2486361265182495, | |
| "learning_rate": 3.321630894438682e-05, | |
| "loss": 0.0902, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 1.2117621714778113, | |
| "grad_norm": 2.5114333629608154, | |
| "learning_rate": 3.311657224926195e-05, | |
| "loss": 0.1306, | |
| "step": 5625 | |
| }, | |
| { | |
| "epoch": 1.2171477811288238, | |
| "grad_norm": 0.07333961874246597, | |
| "learning_rate": 3.301683555413708e-05, | |
| "loss": 0.0746, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 1.2225333907798364, | |
| "grad_norm": 0.26353907585144043, | |
| "learning_rate": 3.291709885901221e-05, | |
| "loss": 0.0902, | |
| "step": 5675 | |
| }, | |
| { | |
| "epoch": 1.2279190004308487, | |
| "grad_norm": 4.065846920013428, | |
| "learning_rate": 3.281736216388734e-05, | |
| "loss": 0.0805, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 1.2333046100818612, | |
| "grad_norm": 0.06470991671085358, | |
| "learning_rate": 3.271762546876247e-05, | |
| "loss": 0.1123, | |
| "step": 5725 | |
| }, | |
| { | |
| "epoch": 1.2386902197328737, | |
| "grad_norm": 0.8340930938720703, | |
| "learning_rate": 3.26178887736376e-05, | |
| "loss": 0.1006, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 1.2440758293838863, | |
| "grad_norm": 1.6223039627075195, | |
| "learning_rate": 3.251815207851273e-05, | |
| "loss": 0.1494, | |
| "step": 5775 | |
| }, | |
| { | |
| "epoch": 1.2494614390348988, | |
| "grad_norm": 5.608697891235352, | |
| "learning_rate": 3.2418415383387856e-05, | |
| "loss": 0.1074, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 1.2548470486859111, | |
| "grad_norm": 0.0557849146425724, | |
| "learning_rate": 3.231867868826299e-05, | |
| "loss": 0.084, | |
| "step": 5825 | |
| }, | |
| { | |
| "epoch": 1.2602326583369237, | |
| "grad_norm": 27.92970085144043, | |
| "learning_rate": 3.221894199313812e-05, | |
| "loss": 0.0981, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 1.2656182679879362, | |
| "grad_norm": 0.06425247341394424, | |
| "learning_rate": 3.2119205298013244e-05, | |
| "loss": 0.0708, | |
| "step": 5875 | |
| }, | |
| { | |
| "epoch": 1.2710038776389487, | |
| "grad_norm": 0.17037852108478546, | |
| "learning_rate": 3.201946860288838e-05, | |
| "loss": 0.0508, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 1.2763894872899613, | |
| "grad_norm": 71.36393737792969, | |
| "learning_rate": 3.1919731907763505e-05, | |
| "loss": 0.0935, | |
| "step": 5925 | |
| }, | |
| { | |
| "epoch": 1.2817750969409736, | |
| "grad_norm": 0.062485672533512115, | |
| "learning_rate": 3.181999521263864e-05, | |
| "loss": 0.0126, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 1.2871607065919863, | |
| "grad_norm": 2.009347438812256, | |
| "learning_rate": 3.1720258517513767e-05, | |
| "loss": 0.1319, | |
| "step": 5975 | |
| }, | |
| { | |
| "epoch": 1.2925463162429986, | |
| "grad_norm": 0.06675919890403748, | |
| "learning_rate": 3.1620521822388894e-05, | |
| "loss": 0.0746, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 1.2979319258940112, | |
| "grad_norm": 8.066916465759277, | |
| "learning_rate": 3.152078512726403e-05, | |
| "loss": 0.1249, | |
| "step": 6025 | |
| }, | |
| { | |
| "epoch": 1.3033175355450237, | |
| "grad_norm": 0.03880688175559044, | |
| "learning_rate": 3.1421048432139155e-05, | |
| "loss": 0.1348, | |
| "step": 6050 | |
| }, | |
| { | |
| "epoch": 1.3087031451960363, | |
| "grad_norm": 0.30525916814804077, | |
| "learning_rate": 3.132131173701429e-05, | |
| "loss": 0.0379, | |
| "step": 6075 | |
| }, | |
| { | |
| "epoch": 1.3140887548470488, | |
| "grad_norm": 19.745574951171875, | |
| "learning_rate": 3.122157504188941e-05, | |
| "loss": 0.2353, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 1.319474364498061, | |
| "grad_norm": 0.17969800531864166, | |
| "learning_rate": 3.112183834676454e-05, | |
| "loss": 0.0887, | |
| "step": 6125 | |
| }, | |
| { | |
| "epoch": 1.3248599741490736, | |
| "grad_norm": 0.11662442237138748, | |
| "learning_rate": 3.102210165163968e-05, | |
| "loss": 0.0777, | |
| "step": 6150 | |
| }, | |
| { | |
| "epoch": 1.3302455838000862, | |
| "grad_norm": 29.151994705200195, | |
| "learning_rate": 3.0922364956514804e-05, | |
| "loss": 0.1288, | |
| "step": 6175 | |
| }, | |
| { | |
| "epoch": 1.3356311934510987, | |
| "grad_norm": 0.1103227511048317, | |
| "learning_rate": 3.082262826138993e-05, | |
| "loss": 0.1441, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 1.3410168031021112, | |
| "grad_norm": 0.055622648447752, | |
| "learning_rate": 3.072289156626506e-05, | |
| "loss": 0.0355, | |
| "step": 6225 | |
| }, | |
| { | |
| "epoch": 1.3464024127531236, | |
| "grad_norm": 0.05069245770573616, | |
| "learning_rate": 3.062315487114019e-05, | |
| "loss": 0.1014, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 1.351788022404136, | |
| "grad_norm": 2.9234323501586914, | |
| "learning_rate": 3.0523418176015326e-05, | |
| "loss": 0.1902, | |
| "step": 6275 | |
| }, | |
| { | |
| "epoch": 1.3571736320551486, | |
| "grad_norm": 0.10473991930484772, | |
| "learning_rate": 3.042368148089045e-05, | |
| "loss": 0.1063, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 1.3625592417061612, | |
| "grad_norm": 0.2278774380683899, | |
| "learning_rate": 3.032394478576558e-05, | |
| "loss": 0.1619, | |
| "step": 6325 | |
| }, | |
| { | |
| "epoch": 1.3679448513571737, | |
| "grad_norm": 0.31000664830207825, | |
| "learning_rate": 3.022420809064071e-05, | |
| "loss": 0.1219, | |
| "step": 6350 | |
| }, | |
| { | |
| "epoch": 1.373330461008186, | |
| "grad_norm": 0.3597787022590637, | |
| "learning_rate": 3.012447139551584e-05, | |
| "loss": 0.1056, | |
| "step": 6375 | |
| }, | |
| { | |
| "epoch": 1.3787160706591985, | |
| "grad_norm": 0.11010152846574783, | |
| "learning_rate": 3.0024734700390972e-05, | |
| "loss": 0.1596, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 1.384101680310211, | |
| "grad_norm": 1.2052332162857056, | |
| "learning_rate": 2.99249980052661e-05, | |
| "loss": 0.1476, | |
| "step": 6425 | |
| }, | |
| { | |
| "epoch": 1.3894872899612236, | |
| "grad_norm": 3.561495542526245, | |
| "learning_rate": 2.982526131014123e-05, | |
| "loss": 0.0991, | |
| "step": 6450 | |
| }, | |
| { | |
| "epoch": 1.3948728996122362, | |
| "grad_norm": 0.16460971534252167, | |
| "learning_rate": 2.972552461501636e-05, | |
| "loss": 0.0774, | |
| "step": 6475 | |
| }, | |
| { | |
| "epoch": 1.4002585092632485, | |
| "grad_norm": 0.08335640281438828, | |
| "learning_rate": 2.962578791989149e-05, | |
| "loss": 0.1055, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 1.4056441189142612, | |
| "grad_norm": 24.70379638671875, | |
| "learning_rate": 2.9526051224766614e-05, | |
| "loss": 0.0675, | |
| "step": 6525 | |
| }, | |
| { | |
| "epoch": 1.4110297285652735, | |
| "grad_norm": 0.06125853583216667, | |
| "learning_rate": 2.9426314529641748e-05, | |
| "loss": 0.0687, | |
| "step": 6550 | |
| }, | |
| { | |
| "epoch": 1.416415338216286, | |
| "grad_norm": 0.05415304750204086, | |
| "learning_rate": 2.932657783451688e-05, | |
| "loss": 0.0439, | |
| "step": 6575 | |
| }, | |
| { | |
| "epoch": 1.4218009478672986, | |
| "grad_norm": 0.06519858539104462, | |
| "learning_rate": 2.922684113939201e-05, | |
| "loss": 0.052, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 1.4271865575183111, | |
| "grad_norm": 0.05177925154566765, | |
| "learning_rate": 2.9127104444267133e-05, | |
| "loss": 0.068, | |
| "step": 6625 | |
| }, | |
| { | |
| "epoch": 1.4325721671693237, | |
| "grad_norm": 0.0696578249335289, | |
| "learning_rate": 2.9027367749142263e-05, | |
| "loss": 0.1637, | |
| "step": 6650 | |
| }, | |
| { | |
| "epoch": 1.437957776820336, | |
| "grad_norm": 0.07040733844041824, | |
| "learning_rate": 2.8927631054017397e-05, | |
| "loss": 0.0799, | |
| "step": 6675 | |
| }, | |
| { | |
| "epoch": 1.4433433864713485, | |
| "grad_norm": 0.09942576289176941, | |
| "learning_rate": 2.8827894358892528e-05, | |
| "loss": 0.1363, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 1.448728996122361, | |
| "grad_norm": 0.07192389667034149, | |
| "learning_rate": 2.8728157663767658e-05, | |
| "loss": 0.1003, | |
| "step": 6725 | |
| }, | |
| { | |
| "epoch": 1.4541146057733736, | |
| "grad_norm": 0.0946580320596695, | |
| "learning_rate": 2.8628420968642782e-05, | |
| "loss": 0.1488, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 1.4595002154243861, | |
| "grad_norm": 0.07671581953763962, | |
| "learning_rate": 2.8528684273517913e-05, | |
| "loss": 0.0701, | |
| "step": 6775 | |
| }, | |
| { | |
| "epoch": 1.4648858250753984, | |
| "grad_norm": 0.06100593879818916, | |
| "learning_rate": 2.8428947578393046e-05, | |
| "loss": 0.0232, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 1.470271434726411, | |
| "grad_norm": 0.04395158588886261, | |
| "learning_rate": 2.8329210883268177e-05, | |
| "loss": 0.0763, | |
| "step": 6825 | |
| }, | |
| { | |
| "epoch": 1.4756570443774235, | |
| "grad_norm": 0.03843090683221817, | |
| "learning_rate": 2.82294741881433e-05, | |
| "loss": 0.0832, | |
| "step": 6850 | |
| }, | |
| { | |
| "epoch": 1.481042654028436, | |
| "grad_norm": 32.23625564575195, | |
| "learning_rate": 2.812973749301843e-05, | |
| "loss": 0.117, | |
| "step": 6875 | |
| }, | |
| { | |
| "epoch": 1.4864282636794486, | |
| "grad_norm": 0.042882196605205536, | |
| "learning_rate": 2.8030000797893562e-05, | |
| "loss": 0.1001, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 1.491813873330461, | |
| "grad_norm": 0.1393657624721527, | |
| "learning_rate": 2.7930264102768696e-05, | |
| "loss": 0.1046, | |
| "step": 6925 | |
| }, | |
| { | |
| "epoch": 1.4971994829814734, | |
| "grad_norm": 30.964231491088867, | |
| "learning_rate": 2.783052740764382e-05, | |
| "loss": 0.1901, | |
| "step": 6950 | |
| }, | |
| { | |
| "epoch": 1.502585092632486, | |
| "grad_norm": 1.4588963985443115, | |
| "learning_rate": 2.773079071251895e-05, | |
| "loss": 0.0741, | |
| "step": 6975 | |
| }, | |
| { | |
| "epoch": 1.5079707022834985, | |
| "grad_norm": 0.06478077918291092, | |
| "learning_rate": 2.763105401739408e-05, | |
| "loss": 0.08, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 1.513356311934511, | |
| "grad_norm": 0.05906078219413757, | |
| "learning_rate": 2.753131732226921e-05, | |
| "loss": 0.0685, | |
| "step": 7025 | |
| }, | |
| { | |
| "epoch": 1.5187419215855233, | |
| "grad_norm": 4.263314247131348, | |
| "learning_rate": 2.7431580627144338e-05, | |
| "loss": 0.0689, | |
| "step": 7050 | |
| }, | |
| { | |
| "epoch": 1.524127531236536, | |
| "grad_norm": 0.07187110185623169, | |
| "learning_rate": 2.733184393201947e-05, | |
| "loss": 0.1388, | |
| "step": 7075 | |
| }, | |
| { | |
| "epoch": 1.5295131408875484, | |
| "grad_norm": 0.06428200751543045, | |
| "learning_rate": 2.72321072368946e-05, | |
| "loss": 0.0718, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 1.534898750538561, | |
| "grad_norm": 0.05657276138663292, | |
| "learning_rate": 2.713237054176973e-05, | |
| "loss": 0.0777, | |
| "step": 7125 | |
| }, | |
| { | |
| "epoch": 1.5402843601895735, | |
| "grad_norm": 29.23991584777832, | |
| "learning_rate": 2.703263384664486e-05, | |
| "loss": 0.1151, | |
| "step": 7150 | |
| }, | |
| { | |
| "epoch": 1.5456699698405858, | |
| "grad_norm": 13.988266944885254, | |
| "learning_rate": 2.6932897151519987e-05, | |
| "loss": 0.1451, | |
| "step": 7175 | |
| }, | |
| { | |
| "epoch": 1.5510555794915986, | |
| "grad_norm": 1.5213409662246704, | |
| "learning_rate": 2.6833160456395118e-05, | |
| "loss": 0.1063, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 1.5564411891426109, | |
| "grad_norm": 0.10175667703151703, | |
| "learning_rate": 2.6733423761270248e-05, | |
| "loss": 0.0863, | |
| "step": 7225 | |
| }, | |
| { | |
| "epoch": 1.5618267987936234, | |
| "grad_norm": 0.12581761181354523, | |
| "learning_rate": 2.663368706614538e-05, | |
| "loss": 0.2038, | |
| "step": 7250 | |
| }, | |
| { | |
| "epoch": 1.567212408444636, | |
| "grad_norm": 0.11780127137899399, | |
| "learning_rate": 2.6533950371020506e-05, | |
| "loss": 0.1042, | |
| "step": 7275 | |
| }, | |
| { | |
| "epoch": 1.5725980180956485, | |
| "grad_norm": 0.2194374054670334, | |
| "learning_rate": 2.6434213675895636e-05, | |
| "loss": 0.1805, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 1.577983627746661, | |
| "grad_norm": 0.18749874830245972, | |
| "learning_rate": 2.6334476980770767e-05, | |
| "loss": 0.1292, | |
| "step": 7325 | |
| }, | |
| { | |
| "epoch": 1.5833692373976733, | |
| "grad_norm": 0.12621381878852844, | |
| "learning_rate": 2.6234740285645897e-05, | |
| "loss": 0.0807, | |
| "step": 7350 | |
| }, | |
| { | |
| "epoch": 1.5887548470486859, | |
| "grad_norm": 0.09865190833806992, | |
| "learning_rate": 2.6135003590521025e-05, | |
| "loss": 0.0985, | |
| "step": 7375 | |
| }, | |
| { | |
| "epoch": 1.5941404566996984, | |
| "grad_norm": 0.11946694552898407, | |
| "learning_rate": 2.6035266895396155e-05, | |
| "loss": 0.0895, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 1.599526066350711, | |
| "grad_norm": 1.2805886268615723, | |
| "learning_rate": 2.5935530200271286e-05, | |
| "loss": 0.1058, | |
| "step": 7425 | |
| }, | |
| { | |
| "epoch": 1.6049116760017235, | |
| "grad_norm": 0.14271412789821625, | |
| "learning_rate": 2.5835793505146416e-05, | |
| "loss": 0.1345, | |
| "step": 7450 | |
| }, | |
| { | |
| "epoch": 1.6102972856527358, | |
| "grad_norm": 32.52022933959961, | |
| "learning_rate": 2.5736056810021547e-05, | |
| "loss": 0.1266, | |
| "step": 7475 | |
| }, | |
| { | |
| "epoch": 1.6156828953037485, | |
| "grad_norm": 0.12038301676511765, | |
| "learning_rate": 2.5636320114896674e-05, | |
| "loss": 0.1492, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 1.6210685049547608, | |
| "grad_norm": 1.8868168592453003, | |
| "learning_rate": 2.5536583419771804e-05, | |
| "loss": 0.0776, | |
| "step": 7525 | |
| }, | |
| { | |
| "epoch": 1.6264541146057734, | |
| "grad_norm": 1.943164348602295, | |
| "learning_rate": 2.5436846724646935e-05, | |
| "loss": 0.0348, | |
| "step": 7550 | |
| }, | |
| { | |
| "epoch": 1.631839724256786, | |
| "grad_norm": 0.07310531288385391, | |
| "learning_rate": 2.5337110029522065e-05, | |
| "loss": 0.0375, | |
| "step": 7575 | |
| }, | |
| { | |
| "epoch": 1.6372253339077982, | |
| "grad_norm": 0.05736231431365013, | |
| "learning_rate": 2.5237373334397192e-05, | |
| "loss": 0.0588, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 1.642610943558811, | |
| "grad_norm": 0.05780019611120224, | |
| "learning_rate": 2.5137636639272323e-05, | |
| "loss": 0.0524, | |
| "step": 7625 | |
| }, | |
| { | |
| "epoch": 1.6479965532098233, | |
| "grad_norm": 0.5779060125350952, | |
| "learning_rate": 2.5037899944147453e-05, | |
| "loss": 0.1221, | |
| "step": 7650 | |
| }, | |
| { | |
| "epoch": 1.6533821628608358, | |
| "grad_norm": 0.05291671305894852, | |
| "learning_rate": 2.493816324902258e-05, | |
| "loss": 0.0978, | |
| "step": 7675 | |
| }, | |
| { | |
| "epoch": 1.6587677725118484, | |
| "grad_norm": 0.04571978747844696, | |
| "learning_rate": 2.483842655389771e-05, | |
| "loss": 0.0798, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 1.6641533821628607, | |
| "grad_norm": 0.0946052223443985, | |
| "learning_rate": 2.473868985877284e-05, | |
| "loss": 0.0712, | |
| "step": 7725 | |
| }, | |
| { | |
| "epoch": 1.6695389918138734, | |
| "grad_norm": 0.04669623076915741, | |
| "learning_rate": 2.4638953163647972e-05, | |
| "loss": 0.1253, | |
| "step": 7750 | |
| }, | |
| { | |
| "epoch": 1.6749246014648858, | |
| "grad_norm": 0.03837985917925835, | |
| "learning_rate": 2.45392164685231e-05, | |
| "loss": 0.0281, | |
| "step": 7775 | |
| }, | |
| { | |
| "epoch": 1.6803102111158983, | |
| "grad_norm": 0.1306626796722412, | |
| "learning_rate": 2.443947977339823e-05, | |
| "loss": 0.0298, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 1.6856958207669108, | |
| "grad_norm": 0.04202219098806381, | |
| "learning_rate": 2.4339743078273357e-05, | |
| "loss": 0.1771, | |
| "step": 7825 | |
| }, | |
| { | |
| "epoch": 1.6910814304179234, | |
| "grad_norm": 9.636494636535645, | |
| "learning_rate": 2.424000638314849e-05, | |
| "loss": 0.1739, | |
| "step": 7850 | |
| }, | |
| { | |
| "epoch": 1.696467040068936, | |
| "grad_norm": 0.05588022246956825, | |
| "learning_rate": 2.414026968802362e-05, | |
| "loss": 0.1273, | |
| "step": 7875 | |
| }, | |
| { | |
| "epoch": 1.7018526497199482, | |
| "grad_norm": 0.30812951922416687, | |
| "learning_rate": 2.404053299289875e-05, | |
| "loss": 0.0813, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 1.7072382593709607, | |
| "grad_norm": 1.565081000328064, | |
| "learning_rate": 2.394079629777388e-05, | |
| "loss": 0.0759, | |
| "step": 7925 | |
| }, | |
| { | |
| "epoch": 1.7126238690219733, | |
| "grad_norm": 0.06271021068096161, | |
| "learning_rate": 2.3841059602649006e-05, | |
| "loss": 0.1473, | |
| "step": 7950 | |
| }, | |
| { | |
| "epoch": 1.7180094786729858, | |
| "grad_norm": 0.06820254772901535, | |
| "learning_rate": 2.374132290752414e-05, | |
| "loss": 0.0978, | |
| "step": 7975 | |
| }, | |
| { | |
| "epoch": 1.7233950883239983, | |
| "grad_norm": 0.0788588598370552, | |
| "learning_rate": 2.3641586212399267e-05, | |
| "loss": 0.1251, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 1.7287806979750107, | |
| "grad_norm": 0.08439130336046219, | |
| "learning_rate": 2.3541849517274398e-05, | |
| "loss": 0.0489, | |
| "step": 8025 | |
| }, | |
| { | |
| "epoch": 1.7341663076260234, | |
| "grad_norm": 0.08412740379571915, | |
| "learning_rate": 2.3442112822149525e-05, | |
| "loss": 0.0972, | |
| "step": 8050 | |
| }, | |
| { | |
| "epoch": 1.7395519172770357, | |
| "grad_norm": 0.05004223436117172, | |
| "learning_rate": 2.3342376127024655e-05, | |
| "loss": 0.0867, | |
| "step": 8075 | |
| }, | |
| { | |
| "epoch": 1.7449375269280483, | |
| "grad_norm": 0.08401107043027878, | |
| "learning_rate": 2.3242639431899786e-05, | |
| "loss": 0.1558, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 1.7503231365790608, | |
| "grad_norm": 0.1881813257932663, | |
| "learning_rate": 2.3142902736774916e-05, | |
| "loss": 0.0759, | |
| "step": 8125 | |
| }, | |
| { | |
| "epoch": 1.7557087462300731, | |
| "grad_norm": 1.4269578456878662, | |
| "learning_rate": 2.3043166041650043e-05, | |
| "loss": 0.0702, | |
| "step": 8150 | |
| }, | |
| { | |
| "epoch": 1.7610943558810859, | |
| "grad_norm": 43.845985412597656, | |
| "learning_rate": 2.2943429346525174e-05, | |
| "loss": 0.0901, | |
| "step": 8175 | |
| }, | |
| { | |
| "epoch": 1.7664799655320982, | |
| "grad_norm": 0.10664209723472595, | |
| "learning_rate": 2.2843692651400304e-05, | |
| "loss": 0.0651, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 1.7718655751831107, | |
| "grad_norm": 3.7163357734680176, | |
| "learning_rate": 2.2743955956275435e-05, | |
| "loss": 0.0471, | |
| "step": 8225 | |
| }, | |
| { | |
| "epoch": 1.7772511848341233, | |
| "grad_norm": 0.05055355280637741, | |
| "learning_rate": 2.2644219261150565e-05, | |
| "loss": 0.1338, | |
| "step": 8250 | |
| }, | |
| { | |
| "epoch": 1.7826367944851356, | |
| "grad_norm": 0.23772941529750824, | |
| "learning_rate": 2.2544482566025693e-05, | |
| "loss": 0.2019, | |
| "step": 8275 | |
| }, | |
| { | |
| "epoch": 1.7880224041361483, | |
| "grad_norm": 0.03852352499961853, | |
| "learning_rate": 2.2444745870900823e-05, | |
| "loss": 0.0408, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 1.7934080137871606, | |
| "grad_norm": 2.454216480255127, | |
| "learning_rate": 2.2345009175775954e-05, | |
| "loss": 0.1086, | |
| "step": 8325 | |
| }, | |
| { | |
| "epoch": 1.7987936234381732, | |
| "grad_norm": 0.2967223525047302, | |
| "learning_rate": 2.2245272480651084e-05, | |
| "loss": 0.0961, | |
| "step": 8350 | |
| }, | |
| { | |
| "epoch": 1.8041792330891857, | |
| "grad_norm": 0.5866406559944153, | |
| "learning_rate": 2.214553578552621e-05, | |
| "loss": 0.1074, | |
| "step": 8375 | |
| }, | |
| { | |
| "epoch": 1.8095648427401982, | |
| "grad_norm": 0.07801081985235214, | |
| "learning_rate": 2.2045799090401342e-05, | |
| "loss": 0.0628, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 1.8149504523912108, | |
| "grad_norm": 18.57970428466797, | |
| "learning_rate": 2.194606239527647e-05, | |
| "loss": 0.1468, | |
| "step": 8425 | |
| }, | |
| { | |
| "epoch": 1.820336062042223, | |
| "grad_norm": 0.5681573152542114, | |
| "learning_rate": 2.1846325700151603e-05, | |
| "loss": 0.0894, | |
| "step": 8450 | |
| }, | |
| { | |
| "epoch": 1.8257216716932356, | |
| "grad_norm": 0.07245413213968277, | |
| "learning_rate": 2.174658900502673e-05, | |
| "loss": 0.1625, | |
| "step": 8475 | |
| }, | |
| { | |
| "epoch": 1.8311072813442482, | |
| "grad_norm": 0.058218203485012054, | |
| "learning_rate": 2.164685230990186e-05, | |
| "loss": 0.0966, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 1.8364928909952607, | |
| "grad_norm": 0.7239183187484741, | |
| "learning_rate": 2.1547115614776988e-05, | |
| "loss": 0.1296, | |
| "step": 8525 | |
| }, | |
| { | |
| "epoch": 1.8418785006462732, | |
| "grad_norm": 0.0625491514801979, | |
| "learning_rate": 2.1447378919652118e-05, | |
| "loss": 0.1106, | |
| "step": 8550 | |
| }, | |
| { | |
| "epoch": 1.8472641102972855, | |
| "grad_norm": 0.11988529562950134, | |
| "learning_rate": 2.134764222452725e-05, | |
| "loss": 0.129, | |
| "step": 8575 | |
| }, | |
| { | |
| "epoch": 1.8526497199482983, | |
| "grad_norm": 23.358821868896484, | |
| "learning_rate": 2.124790552940238e-05, | |
| "loss": 0.1312, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 1.8580353295993106, | |
| "grad_norm": 1.5045102834701538, | |
| "learning_rate": 2.114816883427751e-05, | |
| "loss": 0.107, | |
| "step": 8625 | |
| }, | |
| { | |
| "epoch": 1.8634209392503231, | |
| "grad_norm": 0.06715774536132812, | |
| "learning_rate": 2.1048432139152637e-05, | |
| "loss": 0.0365, | |
| "step": 8650 | |
| }, | |
| { | |
| "epoch": 1.8688065489013357, | |
| "grad_norm": 1.9744484424591064, | |
| "learning_rate": 2.0948695444027767e-05, | |
| "loss": 0.0944, | |
| "step": 8675 | |
| }, | |
| { | |
| "epoch": 1.874192158552348, | |
| "grad_norm": 0.28752002120018005, | |
| "learning_rate": 2.0848958748902898e-05, | |
| "loss": 0.1517, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 1.8795777682033608, | |
| "grad_norm": 0.06683491915464401, | |
| "learning_rate": 2.074922205377803e-05, | |
| "loss": 0.0558, | |
| "step": 8725 | |
| }, | |
| { | |
| "epoch": 1.884963377854373, | |
| "grad_norm": 2.5277748107910156, | |
| "learning_rate": 2.0649485358653155e-05, | |
| "loss": 0.124, | |
| "step": 8750 | |
| }, | |
| { | |
| "epoch": 1.8903489875053856, | |
| "grad_norm": 1.8914086818695068, | |
| "learning_rate": 2.0549748663528286e-05, | |
| "loss": 0.0586, | |
| "step": 8775 | |
| }, | |
| { | |
| "epoch": 1.8957345971563981, | |
| "grad_norm": 0.06591325998306274, | |
| "learning_rate": 2.0450011968403416e-05, | |
| "loss": 0.0511, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 1.9011202068074105, | |
| "grad_norm": 1.5716817378997803, | |
| "learning_rate": 2.0350275273278547e-05, | |
| "loss": 0.1128, | |
| "step": 8825 | |
| }, | |
| { | |
| "epoch": 1.9065058164584232, | |
| "grad_norm": 0.05281971022486687, | |
| "learning_rate": 2.0250538578153674e-05, | |
| "loss": 0.1034, | |
| "step": 8850 | |
| }, | |
| { | |
| "epoch": 1.9118914261094355, | |
| "grad_norm": 0.04660295695066452, | |
| "learning_rate": 2.0150801883028805e-05, | |
| "loss": 0.216, | |
| "step": 8875 | |
| }, | |
| { | |
| "epoch": 1.917277035760448, | |
| "grad_norm": 0.2716023027896881, | |
| "learning_rate": 2.0051065187903935e-05, | |
| "loss": 0.189, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 1.9226626454114606, | |
| "grad_norm": 0.06641989201307297, | |
| "learning_rate": 1.9951328492779066e-05, | |
| "loss": 0.0583, | |
| "step": 8925 | |
| }, | |
| { | |
| "epoch": 1.9280482550624731, | |
| "grad_norm": 0.053841717541217804, | |
| "learning_rate": 1.9851591797654193e-05, | |
| "loss": 0.0942, | |
| "step": 8950 | |
| }, | |
| { | |
| "epoch": 1.9334338647134857, | |
| "grad_norm": 0.05274002254009247, | |
| "learning_rate": 1.9751855102529323e-05, | |
| "loss": 0.1183, | |
| "step": 8975 | |
| }, | |
| { | |
| "epoch": 1.938819474364498, | |
| "grad_norm": 0.08617954701185226, | |
| "learning_rate": 1.9652118407404454e-05, | |
| "loss": 0.0855, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 1.9442050840155105, | |
| "grad_norm": 0.06696368008852005, | |
| "learning_rate": 1.9552381712279584e-05, | |
| "loss": 0.117, | |
| "step": 9025 | |
| }, | |
| { | |
| "epoch": 1.949590693666523, | |
| "grad_norm": 7.755177974700928, | |
| "learning_rate": 1.9452645017154715e-05, | |
| "loss": 0.1239, | |
| "step": 9050 | |
| }, | |
| { | |
| "epoch": 1.9549763033175356, | |
| "grad_norm": 0.1449265480041504, | |
| "learning_rate": 1.9352908322029842e-05, | |
| "loss": 0.0605, | |
| "step": 9075 | |
| }, | |
| { | |
| "epoch": 1.9603619129685481, | |
| "grad_norm": 0.04622127115726471, | |
| "learning_rate": 1.9253171626904972e-05, | |
| "loss": 0.1222, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 1.9657475226195604, | |
| "grad_norm": 0.051821399480104446, | |
| "learning_rate": 1.91534349317801e-05, | |
| "loss": 0.0562, | |
| "step": 9125 | |
| }, | |
| { | |
| "epoch": 1.9711331322705732, | |
| "grad_norm": 0.04180682450532913, | |
| "learning_rate": 1.9053698236655234e-05, | |
| "loss": 0.0473, | |
| "step": 9150 | |
| }, | |
| { | |
| "epoch": 1.9765187419215855, | |
| "grad_norm": 0.0317608080804348, | |
| "learning_rate": 1.895396154153036e-05, | |
| "loss": 0.0576, | |
| "step": 9175 | |
| }, | |
| { | |
| "epoch": 1.981904351572598, | |
| "grad_norm": 0.026255348697304726, | |
| "learning_rate": 1.885422484640549e-05, | |
| "loss": 0.0315, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 1.9872899612236106, | |
| "grad_norm": 1.6421452760696411, | |
| "learning_rate": 1.8754488151280618e-05, | |
| "loss": 0.1244, | |
| "step": 9225 | |
| }, | |
| { | |
| "epoch": 1.9926755708746229, | |
| "grad_norm": 0.0805448517203331, | |
| "learning_rate": 1.865475145615575e-05, | |
| "loss": 0.0834, | |
| "step": 9250 | |
| }, | |
| { | |
| "epoch": 1.9980611805256356, | |
| "grad_norm": 0.03920817747712135, | |
| "learning_rate": 1.855501476103088e-05, | |
| "loss": 0.0997, | |
| "step": 9275 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.9691909942906388, | |
| "eval_auc": 0.9931809600784133, | |
| "eval_f1": 0.9694248449861022, | |
| "eval_loss": 0.1328408569097519, | |
| "eval_precision": 0.9589678510998308, | |
| "eval_recall": 0.9801124081279723, | |
| "eval_runtime": 1746.5378, | |
| "eval_samples_per_second": 5.315, | |
| "eval_steps_per_second": 0.333, | |
| "step": 9284 | |
| }, | |
| { | |
| "epoch": 2.003446790176648, | |
| "grad_norm": 0.672393262386322, | |
| "learning_rate": 1.845527806590601e-05, | |
| "loss": 0.0478, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 2.0088323998276607, | |
| "grad_norm": 0.0429680272936821, | |
| "learning_rate": 1.8355541370781137e-05, | |
| "loss": 0.0864, | |
| "step": 9325 | |
| }, | |
| { | |
| "epoch": 2.014218009478673, | |
| "grad_norm": 0.036236416548490524, | |
| "learning_rate": 1.8255804675656267e-05, | |
| "loss": 0.0491, | |
| "step": 9350 | |
| }, | |
| { | |
| "epoch": 2.0196036191296853, | |
| "grad_norm": 0.036336880177259445, | |
| "learning_rate": 1.8156067980531398e-05, | |
| "loss": 0.0581, | |
| "step": 9375 | |
| }, | |
| { | |
| "epoch": 2.024989228780698, | |
| "grad_norm": 0.032438818365335464, | |
| "learning_rate": 1.805633128540653e-05, | |
| "loss": 0.0419, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 2.0303748384317104, | |
| "grad_norm": 1.6185767650604248, | |
| "learning_rate": 1.795659459028166e-05, | |
| "loss": 0.0767, | |
| "step": 9425 | |
| }, | |
| { | |
| "epoch": 2.035760448082723, | |
| "grad_norm": 2.298982858657837, | |
| "learning_rate": 1.7856857895156786e-05, | |
| "loss": 0.0983, | |
| "step": 9450 | |
| }, | |
| { | |
| "epoch": 2.0411460577337355, | |
| "grad_norm": 8.202011108398438, | |
| "learning_rate": 1.7757121200031917e-05, | |
| "loss": 0.0845, | |
| "step": 9475 | |
| }, | |
| { | |
| "epoch": 2.046531667384748, | |
| "grad_norm": 0.029248099774122238, | |
| "learning_rate": 1.7657384504907047e-05, | |
| "loss": 0.053, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 2.0519172770357605, | |
| "grad_norm": 0.02675529569387436, | |
| "learning_rate": 1.7557647809782178e-05, | |
| "loss": 0.0053, | |
| "step": 9525 | |
| }, | |
| { | |
| "epoch": 2.057302886686773, | |
| "grad_norm": 17.323036193847656, | |
| "learning_rate": 1.7457911114657305e-05, | |
| "loss": 0.127, | |
| "step": 9550 | |
| }, | |
| { | |
| "epoch": 2.0626884963377856, | |
| "grad_norm": 0.03516780957579613, | |
| "learning_rate": 1.7358174419532435e-05, | |
| "loss": 0.1891, | |
| "step": 9575 | |
| }, | |
| { | |
| "epoch": 2.068074105988798, | |
| "grad_norm": 3.910149574279785, | |
| "learning_rate": 1.7258437724407562e-05, | |
| "loss": 0.1083, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 2.0734597156398102, | |
| "grad_norm": 0.18281258642673492, | |
| "learning_rate": 1.7158701029282696e-05, | |
| "loss": 0.0678, | |
| "step": 9625 | |
| }, | |
| { | |
| "epoch": 2.078845325290823, | |
| "grad_norm": 0.04458484426140785, | |
| "learning_rate": 1.7058964334157823e-05, | |
| "loss": 0.0904, | |
| "step": 9650 | |
| }, | |
| { | |
| "epoch": 2.0842309349418353, | |
| "grad_norm": 0.20179250836372375, | |
| "learning_rate": 1.6959227639032954e-05, | |
| "loss": 0.0388, | |
| "step": 9675 | |
| }, | |
| { | |
| "epoch": 2.089616544592848, | |
| "grad_norm": 0.02497880347073078, | |
| "learning_rate": 1.685949094390808e-05, | |
| "loss": 0.0031, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 2.0950021542438604, | |
| "grad_norm": 0.02411968819797039, | |
| "learning_rate": 1.675975424878321e-05, | |
| "loss": 0.0691, | |
| "step": 9725 | |
| }, | |
| { | |
| "epoch": 2.1003877638948727, | |
| "grad_norm": 0.035218965262174606, | |
| "learning_rate": 1.6660017553658346e-05, | |
| "loss": 0.0511, | |
| "step": 9750 | |
| }, | |
| { | |
| "epoch": 2.1057733735458855, | |
| "grad_norm": 0.037588488310575485, | |
| "learning_rate": 1.6560280858533473e-05, | |
| "loss": 0.0335, | |
| "step": 9775 | |
| }, | |
| { | |
| "epoch": 2.1111589831968978, | |
| "grad_norm": 0.03383982926607132, | |
| "learning_rate": 1.6460544163408603e-05, | |
| "loss": 0.0516, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 2.1165445928479105, | |
| "grad_norm": 0.6187468767166138, | |
| "learning_rate": 1.636080746828373e-05, | |
| "loss": 0.0983, | |
| "step": 9825 | |
| }, | |
| { | |
| "epoch": 2.121930202498923, | |
| "grad_norm": 0.23063376545906067, | |
| "learning_rate": 1.626107077315886e-05, | |
| "loss": 0.1008, | |
| "step": 9850 | |
| }, | |
| { | |
| "epoch": 2.127315812149935, | |
| "grad_norm": 0.04398832470178604, | |
| "learning_rate": 1.616133407803399e-05, | |
| "loss": 0.027, | |
| "step": 9875 | |
| }, | |
| { | |
| "epoch": 2.132701421800948, | |
| "grad_norm": 0.0390457846224308, | |
| "learning_rate": 1.6061597382909122e-05, | |
| "loss": 0.0588, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 2.13808703145196, | |
| "grad_norm": 0.8915455937385559, | |
| "learning_rate": 1.596186068778425e-05, | |
| "loss": 0.1456, | |
| "step": 9925 | |
| }, | |
| { | |
| "epoch": 2.143472641102973, | |
| "grad_norm": 68.75797271728516, | |
| "learning_rate": 1.586212399265938e-05, | |
| "loss": 0.0791, | |
| "step": 9950 | |
| }, | |
| { | |
| "epoch": 2.1488582507539853, | |
| "grad_norm": 15.131389617919922, | |
| "learning_rate": 1.576238729753451e-05, | |
| "loss": 0.1169, | |
| "step": 9975 | |
| }, | |
| { | |
| "epoch": 2.154243860404998, | |
| "grad_norm": 0.05127192661166191, | |
| "learning_rate": 1.566265060240964e-05, | |
| "loss": 0.0582, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 2.1596294700560104, | |
| "grad_norm": 0.03388744965195656, | |
| "learning_rate": 1.5562913907284768e-05, | |
| "loss": 0.0609, | |
| "step": 10025 | |
| }, | |
| { | |
| "epoch": 2.1650150797070227, | |
| "grad_norm": 0.0341382697224617, | |
| "learning_rate": 1.5463177212159898e-05, | |
| "loss": 0.1072, | |
| "step": 10050 | |
| }, | |
| { | |
| "epoch": 2.1704006893580354, | |
| "grad_norm": 0.032444145530462265, | |
| "learning_rate": 1.536344051703503e-05, | |
| "loss": 0.0447, | |
| "step": 10075 | |
| }, | |
| { | |
| "epoch": 2.1757862990090477, | |
| "grad_norm": 0.024341439828276634, | |
| "learning_rate": 1.526370382191016e-05, | |
| "loss": 0.0139, | |
| "step": 10100 | |
| }, | |
| { | |
| "epoch": 2.1811719086600605, | |
| "grad_norm": 1.856371521949768, | |
| "learning_rate": 1.5163967126785286e-05, | |
| "loss": 0.0965, | |
| "step": 10125 | |
| }, | |
| { | |
| "epoch": 2.186557518311073, | |
| "grad_norm": 3.588845729827881, | |
| "learning_rate": 1.5064230431660417e-05, | |
| "loss": 0.0047, | |
| "step": 10150 | |
| }, | |
| { | |
| "epoch": 2.191943127962085, | |
| "grad_norm": 0.025228984653949738, | |
| "learning_rate": 1.4964493736535547e-05, | |
| "loss": 0.0525, | |
| "step": 10175 | |
| }, | |
| { | |
| "epoch": 2.197328737613098, | |
| "grad_norm": 0.023060867562890053, | |
| "learning_rate": 1.4864757041410676e-05, | |
| "loss": 0.0817, | |
| "step": 10200 | |
| }, | |
| { | |
| "epoch": 2.20271434726411, | |
| "grad_norm": 0.026297984644770622, | |
| "learning_rate": 1.4765020346285807e-05, | |
| "loss": 0.0313, | |
| "step": 10225 | |
| }, | |
| { | |
| "epoch": 2.208099956915123, | |
| "grad_norm": 0.020485829561948776, | |
| "learning_rate": 1.4665283651160936e-05, | |
| "loss": 0.0022, | |
| "step": 10250 | |
| }, | |
| { | |
| "epoch": 2.2134855665661353, | |
| "grad_norm": 0.023789284750819206, | |
| "learning_rate": 1.4565546956036066e-05, | |
| "loss": 0.069, | |
| "step": 10275 | |
| }, | |
| { | |
| "epoch": 2.2188711762171476, | |
| "grad_norm": 0.02372126840054989, | |
| "learning_rate": 1.4465810260911195e-05, | |
| "loss": 0.0181, | |
| "step": 10300 | |
| }, | |
| { | |
| "epoch": 2.2242567858681603, | |
| "grad_norm": 0.02078492008149624, | |
| "learning_rate": 1.4366073565786325e-05, | |
| "loss": 0.0312, | |
| "step": 10325 | |
| }, | |
| { | |
| "epoch": 2.2296423955191726, | |
| "grad_norm": 0.09992769360542297, | |
| "learning_rate": 1.4266336870661454e-05, | |
| "loss": 0.1303, | |
| "step": 10350 | |
| }, | |
| { | |
| "epoch": 2.2350280051701854, | |
| "grad_norm": 0.14363807439804077, | |
| "learning_rate": 1.4166600175536585e-05, | |
| "loss": 0.1266, | |
| "step": 10375 | |
| }, | |
| { | |
| "epoch": 2.2404136148211977, | |
| "grad_norm": 0.047046031802892685, | |
| "learning_rate": 1.4066863480411714e-05, | |
| "loss": 0.0385, | |
| "step": 10400 | |
| }, | |
| { | |
| "epoch": 2.2457992244722105, | |
| "grad_norm": 0.045923490077257156, | |
| "learning_rate": 1.3967126785286844e-05, | |
| "loss": 0.0046, | |
| "step": 10425 | |
| }, | |
| { | |
| "epoch": 2.251184834123223, | |
| "grad_norm": 21.856792449951172, | |
| "learning_rate": 1.3867390090161971e-05, | |
| "loss": 0.1563, | |
| "step": 10450 | |
| }, | |
| { | |
| "epoch": 2.256570443774235, | |
| "grad_norm": 0.07415134459733963, | |
| "learning_rate": 1.3767653395037103e-05, | |
| "loss": 0.1118, | |
| "step": 10475 | |
| }, | |
| { | |
| "epoch": 2.261956053425248, | |
| "grad_norm": 0.06203525885939598, | |
| "learning_rate": 1.366791669991223e-05, | |
| "loss": 0.0484, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 2.26734166307626, | |
| "grad_norm": 0.03413529694080353, | |
| "learning_rate": 1.3568180004787363e-05, | |
| "loss": 0.0395, | |
| "step": 10525 | |
| }, | |
| { | |
| "epoch": 2.2727272727272725, | |
| "grad_norm": 2.930551528930664, | |
| "learning_rate": 1.3468443309662493e-05, | |
| "loss": 0.0553, | |
| "step": 10550 | |
| }, | |
| { | |
| "epoch": 2.2781128823782852, | |
| "grad_norm": 0.027626806870102882, | |
| "learning_rate": 1.336870661453762e-05, | |
| "loss": 0.0611, | |
| "step": 10575 | |
| }, | |
| { | |
| "epoch": 2.2834984920292976, | |
| "grad_norm": 100.386962890625, | |
| "learning_rate": 1.3268969919412753e-05, | |
| "loss": 0.0417, | |
| "step": 10600 | |
| }, | |
| { | |
| "epoch": 2.2888841016803103, | |
| "grad_norm": 0.02713247761130333, | |
| "learning_rate": 1.316923322428788e-05, | |
| "loss": 0.0763, | |
| "step": 10625 | |
| }, | |
| { | |
| "epoch": 2.2942697113313226, | |
| "grad_norm": 1.6741771697998047, | |
| "learning_rate": 1.3069496529163012e-05, | |
| "loss": 0.098, | |
| "step": 10650 | |
| }, | |
| { | |
| "epoch": 2.2996553209823354, | |
| "grad_norm": 6.76040506362915, | |
| "learning_rate": 1.2969759834038139e-05, | |
| "loss": 0.0836, | |
| "step": 10675 | |
| }, | |
| { | |
| "epoch": 2.3050409306333477, | |
| "grad_norm": 1.2940654754638672, | |
| "learning_rate": 1.287002313891327e-05, | |
| "loss": 0.0466, | |
| "step": 10700 | |
| }, | |
| { | |
| "epoch": 2.31042654028436, | |
| "grad_norm": 0.4345569610595703, | |
| "learning_rate": 1.2770286443788398e-05, | |
| "loss": 0.0564, | |
| "step": 10725 | |
| }, | |
| { | |
| "epoch": 2.3158121499353728, | |
| "grad_norm": 2.130342721939087, | |
| "learning_rate": 1.2670549748663529e-05, | |
| "loss": 0.0852, | |
| "step": 10750 | |
| }, | |
| { | |
| "epoch": 2.321197759586385, | |
| "grad_norm": 0.11710493266582489, | |
| "learning_rate": 1.2570813053538658e-05, | |
| "loss": 0.0509, | |
| "step": 10775 | |
| }, | |
| { | |
| "epoch": 2.326583369237398, | |
| "grad_norm": 0.026548050343990326, | |
| "learning_rate": 1.2471076358413788e-05, | |
| "loss": 0.004, | |
| "step": 10800 | |
| }, | |
| { | |
| "epoch": 2.33196897888841, | |
| "grad_norm": 0.03190811350941658, | |
| "learning_rate": 1.2371339663288919e-05, | |
| "loss": 0.0522, | |
| "step": 10825 | |
| }, | |
| { | |
| "epoch": 2.337354588539423, | |
| "grad_norm": 81.03248596191406, | |
| "learning_rate": 1.2271602968164048e-05, | |
| "loss": 0.0468, | |
| "step": 10850 | |
| }, | |
| { | |
| "epoch": 2.342740198190435, | |
| "grad_norm": 0.02296995371580124, | |
| "learning_rate": 1.2171866273039176e-05, | |
| "loss": 0.0255, | |
| "step": 10875 | |
| }, | |
| { | |
| "epoch": 2.3481258078414475, | |
| "grad_norm": 0.020512187853455544, | |
| "learning_rate": 1.2072129577914307e-05, | |
| "loss": 0.0499, | |
| "step": 10900 | |
| }, | |
| { | |
| "epoch": 2.3535114174924603, | |
| "grad_norm": 0.2346896231174469, | |
| "learning_rate": 1.1972392882789436e-05, | |
| "loss": 0.0904, | |
| "step": 10925 | |
| }, | |
| { | |
| "epoch": 2.3588970271434726, | |
| "grad_norm": 0.03325672820210457, | |
| "learning_rate": 1.1872656187664566e-05, | |
| "loss": 0.1062, | |
| "step": 10950 | |
| }, | |
| { | |
| "epoch": 2.364282636794485, | |
| "grad_norm": 0.0365658663213253, | |
| "learning_rate": 1.1772919492539695e-05, | |
| "loss": 0.0542, | |
| "step": 10975 | |
| }, | |
| { | |
| "epoch": 2.3696682464454977, | |
| "grad_norm": 0.04546808823943138, | |
| "learning_rate": 1.1673182797414826e-05, | |
| "loss": 0.0828, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 2.37505385609651, | |
| "grad_norm": 0.037818703800439835, | |
| "learning_rate": 1.1573446102289954e-05, | |
| "loss": 0.0034, | |
| "step": 11025 | |
| }, | |
| { | |
| "epoch": 2.3804394657475227, | |
| "grad_norm": 0.025843236595392227, | |
| "learning_rate": 1.1473709407165085e-05, | |
| "loss": 0.0419, | |
| "step": 11050 | |
| }, | |
| { | |
| "epoch": 2.385825075398535, | |
| "grad_norm": 0.030073661357164383, | |
| "learning_rate": 1.1373972712040214e-05, | |
| "loss": 0.1035, | |
| "step": 11075 | |
| }, | |
| { | |
| "epoch": 2.391210685049548, | |
| "grad_norm": 1.6878631114959717, | |
| "learning_rate": 1.1274236016915344e-05, | |
| "loss": 0.0982, | |
| "step": 11100 | |
| }, | |
| { | |
| "epoch": 2.39659629470056, | |
| "grad_norm": 0.03798234090209007, | |
| "learning_rate": 1.1174499321790475e-05, | |
| "loss": 0.057, | |
| "step": 11125 | |
| }, | |
| { | |
| "epoch": 2.4019819043515724, | |
| "grad_norm": 0.030130743980407715, | |
| "learning_rate": 1.1074762626665604e-05, | |
| "loss": 0.0424, | |
| "step": 11150 | |
| }, | |
| { | |
| "epoch": 2.407367514002585, | |
| "grad_norm": 0.029741058126091957, | |
| "learning_rate": 1.0975025931540734e-05, | |
| "loss": 0.0183, | |
| "step": 11175 | |
| }, | |
| { | |
| "epoch": 2.4127531236535975, | |
| "grad_norm": 0.025296270847320557, | |
| "learning_rate": 1.0875289236415863e-05, | |
| "loss": 0.0099, | |
| "step": 11200 | |
| }, | |
| { | |
| "epoch": 2.4181387333046103, | |
| "grad_norm": 0.020655043423175812, | |
| "learning_rate": 1.0775552541290992e-05, | |
| "loss": 0.0521, | |
| "step": 11225 | |
| }, | |
| { | |
| "epoch": 2.4235243429556226, | |
| "grad_norm": 0.26498308777809143, | |
| "learning_rate": 1.0675815846166122e-05, | |
| "loss": 0.1073, | |
| "step": 11250 | |
| }, | |
| { | |
| "epoch": 2.428909952606635, | |
| "grad_norm": 0.023832648992538452, | |
| "learning_rate": 1.0576079151041251e-05, | |
| "loss": 0.042, | |
| "step": 11275 | |
| }, | |
| { | |
| "epoch": 2.4342955622576476, | |
| "grad_norm": 0.025238890200853348, | |
| "learning_rate": 1.0476342455916382e-05, | |
| "loss": 0.0765, | |
| "step": 11300 | |
| }, | |
| { | |
| "epoch": 2.43968117190866, | |
| "grad_norm": 0.029484113678336143, | |
| "learning_rate": 1.037660576079151e-05, | |
| "loss": 0.1135, | |
| "step": 11325 | |
| }, | |
| { | |
| "epoch": 2.4450667815596727, | |
| "grad_norm": 4.394412994384766, | |
| "learning_rate": 1.0276869065666641e-05, | |
| "loss": 0.033, | |
| "step": 11350 | |
| }, | |
| { | |
| "epoch": 2.450452391210685, | |
| "grad_norm": 0.02717486396431923, | |
| "learning_rate": 1.017713237054177e-05, | |
| "loss": 0.0756, | |
| "step": 11375 | |
| }, | |
| { | |
| "epoch": 2.4558380008616973, | |
| "grad_norm": 0.03741078078746796, | |
| "learning_rate": 1.0077395675416899e-05, | |
| "loss": 0.0959, | |
| "step": 11400 | |
| }, | |
| { | |
| "epoch": 2.46122361051271, | |
| "grad_norm": 0.025637760758399963, | |
| "learning_rate": 9.977658980292029e-06, | |
| "loss": 0.0421, | |
| "step": 11425 | |
| }, | |
| { | |
| "epoch": 2.4666092201637224, | |
| "grad_norm": 0.04782960191369057, | |
| "learning_rate": 9.877922285167158e-06, | |
| "loss": 0.1026, | |
| "step": 11450 | |
| }, | |
| { | |
| "epoch": 2.471994829814735, | |
| "grad_norm": 0.04921013489365578, | |
| "learning_rate": 9.77818559004229e-06, | |
| "loss": 0.0067, | |
| "step": 11475 | |
| }, | |
| { | |
| "epoch": 2.4773804394657475, | |
| "grad_norm": 0.04560457542538643, | |
| "learning_rate": 9.678448894917419e-06, | |
| "loss": 0.089, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 2.4827660491167602, | |
| "grad_norm": 0.02379678376019001, | |
| "learning_rate": 9.578712199792548e-06, | |
| "loss": 0.0625, | |
| "step": 11525 | |
| }, | |
| { | |
| "epoch": 2.4881516587677726, | |
| "grad_norm": 0.02355758473277092, | |
| "learning_rate": 9.478975504667678e-06, | |
| "loss": 0.0375, | |
| "step": 11550 | |
| }, | |
| { | |
| "epoch": 2.493537268418785, | |
| "grad_norm": 0.025686856359243393, | |
| "learning_rate": 9.379238809542807e-06, | |
| "loss": 0.0407, | |
| "step": 11575 | |
| }, | |
| { | |
| "epoch": 2.4989228780697976, | |
| "grad_norm": 0.028306877240538597, | |
| "learning_rate": 9.279502114417938e-06, | |
| "loss": 0.0277, | |
| "step": 11600 | |
| }, | |
| { | |
| "epoch": 2.50430848772081, | |
| "grad_norm": 0.02451007254421711, | |
| "learning_rate": 9.179765419293066e-06, | |
| "loss": 0.0431, | |
| "step": 11625 | |
| }, | |
| { | |
| "epoch": 2.5096940973718223, | |
| "grad_norm": 3.4917030334472656, | |
| "learning_rate": 9.080028724168197e-06, | |
| "loss": 0.0778, | |
| "step": 11650 | |
| }, | |
| { | |
| "epoch": 2.515079707022835, | |
| "grad_norm": 0.03555026650428772, | |
| "learning_rate": 8.980292029043326e-06, | |
| "loss": 0.0597, | |
| "step": 11675 | |
| }, | |
| { | |
| "epoch": 2.5204653166738473, | |
| "grad_norm": 0.05606693774461746, | |
| "learning_rate": 8.880555333918456e-06, | |
| "loss": 0.0789, | |
| "step": 11700 | |
| }, | |
| { | |
| "epoch": 2.52585092632486, | |
| "grad_norm": 0.03267287090420723, | |
| "learning_rate": 8.780818638793585e-06, | |
| "loss": 0.0212, | |
| "step": 11725 | |
| }, | |
| { | |
| "epoch": 2.5312365359758724, | |
| "grad_norm": 0.028950348496437073, | |
| "learning_rate": 8.681081943668714e-06, | |
| "loss": 0.0142, | |
| "step": 11750 | |
| }, | |
| { | |
| "epoch": 2.536622145626885, | |
| "grad_norm": 0.02338344044983387, | |
| "learning_rate": 8.581345248543844e-06, | |
| "loss": 0.0022, | |
| "step": 11775 | |
| }, | |
| { | |
| "epoch": 2.5420077552778975, | |
| "grad_norm": 0.042633187025785446, | |
| "learning_rate": 8.481608553418973e-06, | |
| "loss": 0.1394, | |
| "step": 11800 | |
| }, | |
| { | |
| "epoch": 2.5473933649289098, | |
| "grad_norm": 0.1146039292216301, | |
| "learning_rate": 8.381871858294104e-06, | |
| "loss": 0.1077, | |
| "step": 11825 | |
| }, | |
| { | |
| "epoch": 2.5527789745799225, | |
| "grad_norm": 0.03126833215355873, | |
| "learning_rate": 8.282135163169234e-06, | |
| "loss": 0.0714, | |
| "step": 11850 | |
| }, | |
| { | |
| "epoch": 2.558164584230935, | |
| "grad_norm": 0.032733842730522156, | |
| "learning_rate": 8.182398468044363e-06, | |
| "loss": 0.1485, | |
| "step": 11875 | |
| }, | |
| { | |
| "epoch": 2.563550193881947, | |
| "grad_norm": 0.5381880402565002, | |
| "learning_rate": 8.082661772919494e-06, | |
| "loss": 0.061, | |
| "step": 11900 | |
| }, | |
| { | |
| "epoch": 2.56893580353296, | |
| "grad_norm": 0.031703583896160126, | |
| "learning_rate": 7.982925077794622e-06, | |
| "loss": 0.0555, | |
| "step": 11925 | |
| }, | |
| { | |
| "epoch": 2.5743214131839727, | |
| "grad_norm": 0.10838180035352707, | |
| "learning_rate": 7.883188382669753e-06, | |
| "loss": 0.0063, | |
| "step": 11950 | |
| }, | |
| { | |
| "epoch": 2.579707022834985, | |
| "grad_norm": 0.05079588666558266, | |
| "learning_rate": 7.783451687544882e-06, | |
| "loss": 0.0251, | |
| "step": 11975 | |
| }, | |
| { | |
| "epoch": 2.5850926324859973, | |
| "grad_norm": 0.023224713280797005, | |
| "learning_rate": 7.683714992420012e-06, | |
| "loss": 0.0351, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 2.59047824213701, | |
| "grad_norm": 0.02427135780453682, | |
| "learning_rate": 7.583978297295141e-06, | |
| "loss": 0.0435, | |
| "step": 12025 | |
| }, | |
| { | |
| "epoch": 2.5958638517880224, | |
| "grad_norm": 10.110872268676758, | |
| "learning_rate": 7.484241602170271e-06, | |
| "loss": 0.0877, | |
| "step": 12050 | |
| }, | |
| { | |
| "epoch": 2.6012494614390347, | |
| "grad_norm": 0.018899202346801758, | |
| "learning_rate": 7.3845049070454004e-06, | |
| "loss": 0.0266, | |
| "step": 12075 | |
| }, | |
| { | |
| "epoch": 2.6066350710900474, | |
| "grad_norm": 2.7654731273651123, | |
| "learning_rate": 7.28476821192053e-06, | |
| "loss": 0.0766, | |
| "step": 12100 | |
| }, | |
| { | |
| "epoch": 2.6120206807410598, | |
| "grad_norm": 0.02472410537302494, | |
| "learning_rate": 7.18503151679566e-06, | |
| "loss": 0.0775, | |
| "step": 12125 | |
| }, | |
| { | |
| "epoch": 2.6174062903920725, | |
| "grad_norm": 0.08125100284814835, | |
| "learning_rate": 7.085294821670789e-06, | |
| "loss": 0.033, | |
| "step": 12150 | |
| }, | |
| { | |
| "epoch": 2.622791900043085, | |
| "grad_norm": 0.01764783076941967, | |
| "learning_rate": 6.985558126545918e-06, | |
| "loss": 0.0032, | |
| "step": 12175 | |
| }, | |
| { | |
| "epoch": 2.6281775096940976, | |
| "grad_norm": 0.019586993381381035, | |
| "learning_rate": 6.885821431421048e-06, | |
| "loss": 0.0463, | |
| "step": 12200 | |
| }, | |
| { | |
| "epoch": 2.63356311934511, | |
| "grad_norm": 1.3871238231658936, | |
| "learning_rate": 6.786084736296179e-06, | |
| "loss": 0.0267, | |
| "step": 12225 | |
| }, | |
| { | |
| "epoch": 2.638948728996122, | |
| "grad_norm": 0.018140500411391258, | |
| "learning_rate": 6.686348041171309e-06, | |
| "loss": 0.0696, | |
| "step": 12250 | |
| }, | |
| { | |
| "epoch": 2.644334338647135, | |
| "grad_norm": 0.0250703152269125, | |
| "learning_rate": 6.586611346046439e-06, | |
| "loss": 0.0742, | |
| "step": 12275 | |
| }, | |
| { | |
| "epoch": 2.6497199482981473, | |
| "grad_norm": 0.020942581817507744, | |
| "learning_rate": 6.4868746509215674e-06, | |
| "loss": 0.0302, | |
| "step": 12300 | |
| }, | |
| { | |
| "epoch": 2.6551055579491596, | |
| "grad_norm": 0.018071915954351425, | |
| "learning_rate": 6.387137955796697e-06, | |
| "loss": 0.0367, | |
| "step": 12325 | |
| }, | |
| { | |
| "epoch": 2.6604911676001723, | |
| "grad_norm": 19.54036521911621, | |
| "learning_rate": 6.287401260671827e-06, | |
| "loss": 0.0825, | |
| "step": 12350 | |
| }, | |
| { | |
| "epoch": 2.665876777251185, | |
| "grad_norm": 0.016232356429100037, | |
| "learning_rate": 6.1876645655469564e-06, | |
| "loss": 0.0359, | |
| "step": 12375 | |
| }, | |
| { | |
| "epoch": 2.6712623869021974, | |
| "grad_norm": 0.024337278679013252, | |
| "learning_rate": 6.087927870422086e-06, | |
| "loss": 0.0273, | |
| "step": 12400 | |
| }, | |
| { | |
| "epoch": 2.6766479965532097, | |
| "grad_norm": 0.28807416558265686, | |
| "learning_rate": 5.988191175297216e-06, | |
| "loss": 0.0266, | |
| "step": 12425 | |
| }, | |
| { | |
| "epoch": 2.6820336062042225, | |
| "grad_norm": 0.01607998088002205, | |
| "learning_rate": 5.8884544801723455e-06, | |
| "loss": 0.0666, | |
| "step": 12450 | |
| }, | |
| { | |
| "epoch": 2.687419215855235, | |
| "grad_norm": 0.021535271778702736, | |
| "learning_rate": 5.788717785047474e-06, | |
| "loss": 0.0611, | |
| "step": 12475 | |
| }, | |
| { | |
| "epoch": 2.692804825506247, | |
| "grad_norm": 12.746896743774414, | |
| "learning_rate": 5.688981089922604e-06, | |
| "loss": 0.101, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 2.69819043515726, | |
| "grad_norm": 1.019801378250122, | |
| "learning_rate": 5.5892443947977345e-06, | |
| "loss": 0.0415, | |
| "step": 12525 | |
| }, | |
| { | |
| "epoch": 2.703576044808272, | |
| "grad_norm": 0.092778280377388, | |
| "learning_rate": 5.489507699672864e-06, | |
| "loss": 0.0365, | |
| "step": 12550 | |
| }, | |
| { | |
| "epoch": 2.708961654459285, | |
| "grad_norm": 0.5808451771736145, | |
| "learning_rate": 5.389771004547994e-06, | |
| "loss": 0.0983, | |
| "step": 12575 | |
| }, | |
| { | |
| "epoch": 2.7143472641102973, | |
| "grad_norm": 0.17807550728321075, | |
| "learning_rate": 5.2900343094231235e-06, | |
| "loss": 0.0214, | |
| "step": 12600 | |
| }, | |
| { | |
| "epoch": 2.71973287376131, | |
| "grad_norm": 0.031068023294210434, | |
| "learning_rate": 5.190297614298253e-06, | |
| "loss": 0.0623, | |
| "step": 12625 | |
| }, | |
| { | |
| "epoch": 2.7251184834123223, | |
| "grad_norm": 0.023084493353962898, | |
| "learning_rate": 5.090560919173382e-06, | |
| "loss": 0.0154, | |
| "step": 12650 | |
| }, | |
| { | |
| "epoch": 2.7305040930633346, | |
| "grad_norm": 0.0403926819562912, | |
| "learning_rate": 4.990824224048512e-06, | |
| "loss": 0.0358, | |
| "step": 12675 | |
| }, | |
| { | |
| "epoch": 2.7358897027143474, | |
| "grad_norm": 0.43667861819267273, | |
| "learning_rate": 4.891087528923642e-06, | |
| "loss": 0.0016, | |
| "step": 12700 | |
| }, | |
| { | |
| "epoch": 2.7412753123653597, | |
| "grad_norm": 0.07555987685918808, | |
| "learning_rate": 4.791350833798772e-06, | |
| "loss": 0.0721, | |
| "step": 12725 | |
| }, | |
| { | |
| "epoch": 2.746660922016372, | |
| "grad_norm": 0.022048979997634888, | |
| "learning_rate": 4.6916141386739015e-06, | |
| "loss": 0.0565, | |
| "step": 12750 | |
| }, | |
| { | |
| "epoch": 2.7520465316673848, | |
| "grad_norm": 0.020737633109092712, | |
| "learning_rate": 4.591877443549031e-06, | |
| "loss": 0.038, | |
| "step": 12775 | |
| }, | |
| { | |
| "epoch": 2.757432141318397, | |
| "grad_norm": 2.780998945236206, | |
| "learning_rate": 4.492140748424161e-06, | |
| "loss": 0.1345, | |
| "step": 12800 | |
| }, | |
| { | |
| "epoch": 2.76281775096941, | |
| "grad_norm": 0.0388825386762619, | |
| "learning_rate": 4.39240405329929e-06, | |
| "loss": 0.0194, | |
| "step": 12825 | |
| }, | |
| { | |
| "epoch": 2.768203360620422, | |
| "grad_norm": 7.417761325836182, | |
| "learning_rate": 4.292667358174419e-06, | |
| "loss": 0.0343, | |
| "step": 12850 | |
| }, | |
| { | |
| "epoch": 2.773588970271435, | |
| "grad_norm": 0.02559277042746544, | |
| "learning_rate": 4.192930663049549e-06, | |
| "loss": 0.002, | |
| "step": 12875 | |
| }, | |
| { | |
| "epoch": 2.7789745799224472, | |
| "grad_norm": 0.019186072051525116, | |
| "learning_rate": 4.0931939679246795e-06, | |
| "loss": 0.0732, | |
| "step": 12900 | |
| }, | |
| { | |
| "epoch": 2.7843601895734595, | |
| "grad_norm": 0.034625016152858734, | |
| "learning_rate": 3.993457272799809e-06, | |
| "loss": 0.0631, | |
| "step": 12925 | |
| }, | |
| { | |
| "epoch": 2.7897457992244723, | |
| "grad_norm": 0.07090263068675995, | |
| "learning_rate": 3.893720577674939e-06, | |
| "loss": 0.0705, | |
| "step": 12950 | |
| }, | |
| { | |
| "epoch": 2.7951314088754846, | |
| "grad_norm": 0.01826942153275013, | |
| "learning_rate": 3.793983882550068e-06, | |
| "loss": 0.0212, | |
| "step": 12975 | |
| }, | |
| { | |
| "epoch": 2.800517018526497, | |
| "grad_norm": 7.512089729309082, | |
| "learning_rate": 3.6942471874251977e-06, | |
| "loss": 0.0609, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 2.8059026281775097, | |
| "grad_norm": 92.8465805053711, | |
| "learning_rate": 3.594510492300327e-06, | |
| "loss": 0.0397, | |
| "step": 13025 | |
| }, | |
| { | |
| "epoch": 2.8112882378285224, | |
| "grad_norm": 2.2542502880096436, | |
| "learning_rate": 3.4947737971754566e-06, | |
| "loss": 0.0542, | |
| "step": 13050 | |
| }, | |
| { | |
| "epoch": 2.8166738474795348, | |
| "grad_norm": 0.026438845321536064, | |
| "learning_rate": 3.395037102050587e-06, | |
| "loss": 0.0257, | |
| "step": 13075 | |
| }, | |
| { | |
| "epoch": 2.822059457130547, | |
| "grad_norm": 0.027188275009393692, | |
| "learning_rate": 3.2953004069257164e-06, | |
| "loss": 0.0357, | |
| "step": 13100 | |
| }, | |
| { | |
| "epoch": 2.82744506678156, | |
| "grad_norm": 0.023801114410161972, | |
| "learning_rate": 3.195563711800846e-06, | |
| "loss": 0.0453, | |
| "step": 13125 | |
| }, | |
| { | |
| "epoch": 2.832830676432572, | |
| "grad_norm": 0.10319499671459198, | |
| "learning_rate": 3.0958270166759757e-06, | |
| "loss": 0.0796, | |
| "step": 13150 | |
| }, | |
| { | |
| "epoch": 2.8382162860835844, | |
| "grad_norm": 0.031992778182029724, | |
| "learning_rate": 2.9960903215511054e-06, | |
| "loss": 0.0505, | |
| "step": 13175 | |
| }, | |
| { | |
| "epoch": 2.843601895734597, | |
| "grad_norm": 0.12323871999979019, | |
| "learning_rate": 2.8963536264262346e-06, | |
| "loss": 0.0366, | |
| "step": 13200 | |
| }, | |
| { | |
| "epoch": 2.8489875053856095, | |
| "grad_norm": 0.019458066672086716, | |
| "learning_rate": 2.7966169313013647e-06, | |
| "loss": 0.0232, | |
| "step": 13225 | |
| }, | |
| { | |
| "epoch": 2.8543731150366223, | |
| "grad_norm": 0.024557072669267654, | |
| "learning_rate": 2.6968802361764944e-06, | |
| "loss": 0.0745, | |
| "step": 13250 | |
| }, | |
| { | |
| "epoch": 2.8597587246876346, | |
| "grad_norm": 5.838003635406494, | |
| "learning_rate": 2.5971435410516236e-06, | |
| "loss": 0.0553, | |
| "step": 13275 | |
| }, | |
| { | |
| "epoch": 2.8651443343386473, | |
| "grad_norm": 1.81170654296875, | |
| "learning_rate": 2.4974068459267533e-06, | |
| "loss": 0.0341, | |
| "step": 13300 | |
| }, | |
| { | |
| "epoch": 2.8705299439896597, | |
| "grad_norm": 4.6604180335998535, | |
| "learning_rate": 2.3976701508018834e-06, | |
| "loss": 0.0166, | |
| "step": 13325 | |
| }, | |
| { | |
| "epoch": 2.875915553640672, | |
| "grad_norm": 0.9680020809173584, | |
| "learning_rate": 2.2979334556770127e-06, | |
| "loss": 0.0575, | |
| "step": 13350 | |
| }, | |
| { | |
| "epoch": 2.8813011632916847, | |
| "grad_norm": 0.044519729912281036, | |
| "learning_rate": 2.1981967605521423e-06, | |
| "loss": 0.0795, | |
| "step": 13375 | |
| }, | |
| { | |
| "epoch": 2.886686772942697, | |
| "grad_norm": 0.019457995891571045, | |
| "learning_rate": 2.098460065427272e-06, | |
| "loss": 0.0278, | |
| "step": 13400 | |
| }, | |
| { | |
| "epoch": 2.8920723825937094, | |
| "grad_norm": 0.017885301262140274, | |
| "learning_rate": 1.9987233703024017e-06, | |
| "loss": 0.0149, | |
| "step": 13425 | |
| }, | |
| { | |
| "epoch": 2.897457992244722, | |
| "grad_norm": 37.20232009887695, | |
| "learning_rate": 1.8989866751775313e-06, | |
| "loss": 0.0252, | |
| "step": 13450 | |
| }, | |
| { | |
| "epoch": 2.902843601895735, | |
| "grad_norm": 0.03464394435286522, | |
| "learning_rate": 1.799249980052661e-06, | |
| "loss": 0.0956, | |
| "step": 13475 | |
| }, | |
| { | |
| "epoch": 2.908229211546747, | |
| "grad_norm": 0.08744242042303085, | |
| "learning_rate": 1.6995132849277909e-06, | |
| "loss": 0.0632, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 2.9136148211977595, | |
| "grad_norm": 0.014135221019387245, | |
| "learning_rate": 1.5997765898029205e-06, | |
| "loss": 0.0276, | |
| "step": 13525 | |
| }, | |
| { | |
| "epoch": 2.9190004308487723, | |
| "grad_norm": 0.06255125254392624, | |
| "learning_rate": 1.50003989467805e-06, | |
| "loss": 0.0825, | |
| "step": 13550 | |
| }, | |
| { | |
| "epoch": 2.9243860404997846, | |
| "grad_norm": 10.135443687438965, | |
| "learning_rate": 1.4003031995531797e-06, | |
| "loss": 0.0442, | |
| "step": 13575 | |
| }, | |
| { | |
| "epoch": 2.929771650150797, | |
| "grad_norm": 0.01875634863972664, | |
| "learning_rate": 1.3005665044283093e-06, | |
| "loss": 0.0304, | |
| "step": 13600 | |
| }, | |
| { | |
| "epoch": 2.9351572598018096, | |
| "grad_norm": 0.13877597451210022, | |
| "learning_rate": 1.200829809303439e-06, | |
| "loss": 0.0367, | |
| "step": 13625 | |
| }, | |
| { | |
| "epoch": 2.940542869452822, | |
| "grad_norm": 0.016782578080892563, | |
| "learning_rate": 1.1010931141785687e-06, | |
| "loss": 0.1222, | |
| "step": 13650 | |
| }, | |
| { | |
| "epoch": 2.9459284791038347, | |
| "grad_norm": 0.03902614489197731, | |
| "learning_rate": 1.0013564190536983e-06, | |
| "loss": 0.1027, | |
| "step": 13675 | |
| }, | |
| { | |
| "epoch": 2.951314088754847, | |
| "grad_norm": 0.017981288954615593, | |
| "learning_rate": 9.016197239288279e-07, | |
| "loss": 0.014, | |
| "step": 13700 | |
| }, | |
| { | |
| "epoch": 2.9566996984058598, | |
| "grad_norm": 0.016431229189038277, | |
| "learning_rate": 8.018830288039577e-07, | |
| "loss": 0.016, | |
| "step": 13725 | |
| }, | |
| { | |
| "epoch": 2.962085308056872, | |
| "grad_norm": 0.031093724071979523, | |
| "learning_rate": 7.021463336790872e-07, | |
| "loss": 0.0563, | |
| "step": 13750 | |
| }, | |
| { | |
| "epoch": 2.9674709177078844, | |
| "grad_norm": 0.06026161089539528, | |
| "learning_rate": 6.024096385542169e-07, | |
| "loss": 0.0381, | |
| "step": 13775 | |
| }, | |
| { | |
| "epoch": 2.972856527358897, | |
| "grad_norm": 7.449411869049072, | |
| "learning_rate": 5.026729434293466e-07, | |
| "loss": 0.0633, | |
| "step": 13800 | |
| }, | |
| { | |
| "epoch": 2.9782421370099095, | |
| "grad_norm": 0.022474773228168488, | |
| "learning_rate": 4.0293624830447624e-07, | |
| "loss": 0.0601, | |
| "step": 13825 | |
| }, | |
| { | |
| "epoch": 2.983627746660922, | |
| "grad_norm": 0.020873403176665306, | |
| "learning_rate": 3.0319955317960585e-07, | |
| "loss": 0.0036, | |
| "step": 13850 | |
| }, | |
| { | |
| "epoch": 2.9890133563119345, | |
| "grad_norm": 0.0238470621407032, | |
| "learning_rate": 2.0346285805473552e-07, | |
| "loss": 0.0519, | |
| "step": 13875 | |
| }, | |
| { | |
| "epoch": 2.994398965962947, | |
| "grad_norm": 0.03307591751217842, | |
| "learning_rate": 1.0372616292986516e-07, | |
| "loss": 0.002, | |
| "step": 13900 | |
| }, | |
| { | |
| "epoch": 2.9997845756139596, | |
| "grad_norm": 0.9745203852653503, | |
| "learning_rate": 3.989467804994814e-09, | |
| "loss": 0.058, | |
| "step": 13925 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.9739308413228482, | |
| "eval_auc": 0.9959614324316973, | |
| "eval_f1": 0.9739448751076658, | |
| "eval_loss": 0.12016157805919647, | |
| "eval_precision": 0.9701844701844702, | |
| "eval_recall": 0.9777345438824038, | |
| "eval_runtime": 1664.4298, | |
| "eval_samples_per_second": 5.577, | |
| "eval_steps_per_second": 0.349, | |
| "step": 13926 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 13926, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "EarlyStoppingCallback": { | |
| "args": { | |
| "early_stopping_patience": 5, | |
| "early_stopping_threshold": 0.01 | |
| }, | |
| "attributes": { | |
| "early_stopping_patience_counter": 0 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 7327379780720640.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |