| { | |
| "best_metric": 0.05274021625518799, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.08663634394628547, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.00043318171973142733, | |
| "grad_norm": 3.9565112590789795, | |
| "learning_rate": 5e-06, | |
| "loss": 0.7524, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.00043318171973142733, | |
| "eval_loss": 4.976785659790039, | |
| "eval_runtime": 204.1866, | |
| "eval_samples_per_second": 19.041, | |
| "eval_steps_per_second": 9.521, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0008663634394628547, | |
| "grad_norm": 7.407309532165527, | |
| "learning_rate": 1e-05, | |
| "loss": 1.3786, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.001299545159194282, | |
| "grad_norm": 9.137940406799316, | |
| "learning_rate": 1.5e-05, | |
| "loss": 1.6639, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.0017327268789257093, | |
| "grad_norm": 9.636576652526855, | |
| "learning_rate": 2e-05, | |
| "loss": 1.6429, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.0021659085986571367, | |
| "grad_norm": 7.883628845214844, | |
| "learning_rate": 2.5e-05, | |
| "loss": 1.5626, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.002599090318388564, | |
| "grad_norm": 6.31871223449707, | |
| "learning_rate": 3e-05, | |
| "loss": 1.5388, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.0030322720381199915, | |
| "grad_norm": 4.32879638671875, | |
| "learning_rate": 3.5e-05, | |
| "loss": 1.4263, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.0034654537578514186, | |
| "grad_norm": 3.1516454219818115, | |
| "learning_rate": 4e-05, | |
| "loss": 1.3707, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.003898635477582846, | |
| "grad_norm": 3.7077221870422363, | |
| "learning_rate": 4.5e-05, | |
| "loss": 1.3346, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.004331817197314273, | |
| "grad_norm": 3.2946524620056152, | |
| "learning_rate": 5e-05, | |
| "loss": 1.3252, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.004764998917045701, | |
| "grad_norm": 3.5760467052459717, | |
| "learning_rate": 5.500000000000001e-05, | |
| "loss": 1.159, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.005198180636777128, | |
| "grad_norm": 4.902008056640625, | |
| "learning_rate": 6e-05, | |
| "loss": 0.9312, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.005631362356508555, | |
| "grad_norm": 6.108403205871582, | |
| "learning_rate": 6.500000000000001e-05, | |
| "loss": 0.8111, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.006064544076239983, | |
| "grad_norm": 6.699007511138916, | |
| "learning_rate": 7e-05, | |
| "loss": 0.4968, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.00649772579597141, | |
| "grad_norm": 6.076148986816406, | |
| "learning_rate": 7.500000000000001e-05, | |
| "loss": 0.6053, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.006930907515702837, | |
| "grad_norm": 4.3749098777771, | |
| "learning_rate": 8e-05, | |
| "loss": 0.496, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.007364089235434265, | |
| "grad_norm": 4.691214561462402, | |
| "learning_rate": 8.5e-05, | |
| "loss": 0.3746, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.007797270955165692, | |
| "grad_norm": 1.961999535560608, | |
| "learning_rate": 9e-05, | |
| "loss": 0.2746, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.00823045267489712, | |
| "grad_norm": 2.699904203414917, | |
| "learning_rate": 9.5e-05, | |
| "loss": 0.308, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.008663634394628547, | |
| "grad_norm": 2.5429675579071045, | |
| "learning_rate": 0.0001, | |
| "loss": 0.2816, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.009096816114359974, | |
| "grad_norm": 5.605649471282959, | |
| "learning_rate": 9.999238475781957e-05, | |
| "loss": 0.2493, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.009529997834091402, | |
| "grad_norm": 2.444078207015991, | |
| "learning_rate": 9.99695413509548e-05, | |
| "loss": 0.2686, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.009963179553822829, | |
| "grad_norm": 2.162829637527466, | |
| "learning_rate": 9.99314767377287e-05, | |
| "loss": 0.1468, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.010396361273554255, | |
| "grad_norm": 1.8878003358840942, | |
| "learning_rate": 9.987820251299122e-05, | |
| "loss": 0.1481, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.010829542993285684, | |
| "grad_norm": 1.073910117149353, | |
| "learning_rate": 9.980973490458728e-05, | |
| "loss": 0.1747, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.01126272471301711, | |
| "grad_norm": 4.712179660797119, | |
| "learning_rate": 9.972609476841367e-05, | |
| "loss": 0.1783, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.011695906432748537, | |
| "grad_norm": 3.8842508792877197, | |
| "learning_rate": 9.962730758206611e-05, | |
| "loss": 0.1553, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.012129088152479966, | |
| "grad_norm": 1.4134128093719482, | |
| "learning_rate": 9.951340343707852e-05, | |
| "loss": 0.1073, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.012562269872211393, | |
| "grad_norm": 1.8056902885437012, | |
| "learning_rate": 9.938441702975689e-05, | |
| "loss": 0.1946, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.01299545159194282, | |
| "grad_norm": 1.0871001482009888, | |
| "learning_rate": 9.924038765061042e-05, | |
| "loss": 0.0863, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.013428633311674248, | |
| "grad_norm": 1.1687226295471191, | |
| "learning_rate": 9.908135917238321e-05, | |
| "loss": 0.0939, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.013861815031405674, | |
| "grad_norm": 0.7801273465156555, | |
| "learning_rate": 9.890738003669029e-05, | |
| "loss": 0.067, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.014294996751137101, | |
| "grad_norm": 1.7727514505386353, | |
| "learning_rate": 9.871850323926177e-05, | |
| "loss": 0.1051, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.01472817847086853, | |
| "grad_norm": 1.272719144821167, | |
| "learning_rate": 9.851478631379982e-05, | |
| "loss": 0.0542, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.015161360190599956, | |
| "grad_norm": 0.9942130446434021, | |
| "learning_rate": 9.829629131445342e-05, | |
| "loss": 0.0976, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.015594541910331383, | |
| "grad_norm": 1.0211302042007446, | |
| "learning_rate": 9.806308479691595e-05, | |
| "loss": 0.051, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.01602772363006281, | |
| "grad_norm": 0.9490691423416138, | |
| "learning_rate": 9.781523779815179e-05, | |
| "loss": 0.0635, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.01646090534979424, | |
| "grad_norm": 0.638464629650116, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 0.0638, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.016894087069525665, | |
| "grad_norm": 0.5109328031539917, | |
| "learning_rate": 9.727592877996585e-05, | |
| "loss": 0.0233, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.017327268789257094, | |
| "grad_norm": 0.8561784029006958, | |
| "learning_rate": 9.698463103929542e-05, | |
| "loss": 0.0479, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.017760450508988522, | |
| "grad_norm": 1.437472939491272, | |
| "learning_rate": 9.667902132486009e-05, | |
| "loss": 0.0846, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.018193632228719947, | |
| "grad_norm": 0.5984365344047546, | |
| "learning_rate": 9.635919272833938e-05, | |
| "loss": 0.0423, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.018626813948451375, | |
| "grad_norm": 1.3377480506896973, | |
| "learning_rate": 9.602524267262203e-05, | |
| "loss": 0.0807, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.019059995668182804, | |
| "grad_norm": 0.24663518369197845, | |
| "learning_rate": 9.567727288213005e-05, | |
| "loss": 0.0189, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.01949317738791423, | |
| "grad_norm": 0.4021802544593811, | |
| "learning_rate": 9.53153893518325e-05, | |
| "loss": 0.0141, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.019926359107645657, | |
| "grad_norm": 0.26074135303497314, | |
| "learning_rate": 9.493970231495835e-05, | |
| "loss": 0.0108, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.020359540827377086, | |
| "grad_norm": 0.31161004304885864, | |
| "learning_rate": 9.45503262094184e-05, | |
| "loss": 0.0197, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.02079272254710851, | |
| "grad_norm": 0.357147216796875, | |
| "learning_rate": 9.414737964294636e-05, | |
| "loss": 0.0062, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.02122590426683994, | |
| "grad_norm": 0.22662493586540222, | |
| "learning_rate": 9.373098535696979e-05, | |
| "loss": 0.008, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.021659085986571368, | |
| "grad_norm": 0.31289562582969666, | |
| "learning_rate": 9.330127018922194e-05, | |
| "loss": 0.008, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.021659085986571368, | |
| "eval_loss": 0.2094399482011795, | |
| "eval_runtime": 204.5839, | |
| "eval_samples_per_second": 19.004, | |
| "eval_steps_per_second": 9.502, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.022092267706302793, | |
| "grad_norm": 2.441636323928833, | |
| "learning_rate": 9.285836503510562e-05, | |
| "loss": 0.2969, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.02252544942603422, | |
| "grad_norm": 3.2028415203094482, | |
| "learning_rate": 9.24024048078213e-05, | |
| "loss": 0.2594, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.02295863114576565, | |
| "grad_norm": 1.429338812828064, | |
| "learning_rate": 9.193352839727121e-05, | |
| "loss": 0.21, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.023391812865497075, | |
| "grad_norm": 0.7525247931480408, | |
| "learning_rate": 9.145187862775209e-05, | |
| "loss": 0.1651, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.023824994585228503, | |
| "grad_norm": 2.9362690448760986, | |
| "learning_rate": 9.09576022144496e-05, | |
| "loss": 0.2147, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.02425817630495993, | |
| "grad_norm": 2.608370780944824, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 0.2761, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.024691358024691357, | |
| "grad_norm": 2.0152244567871094, | |
| "learning_rate": 8.993177550236464e-05, | |
| "loss": 0.1976, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.025124539744422785, | |
| "grad_norm": 0.9376019835472107, | |
| "learning_rate": 8.940053768033609e-05, | |
| "loss": 0.1674, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.025557721464154214, | |
| "grad_norm": 0.7567312121391296, | |
| "learning_rate": 8.885729807284856e-05, | |
| "loss": 0.1707, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.02599090318388564, | |
| "grad_norm": 0.4066583812236786, | |
| "learning_rate": 8.83022221559489e-05, | |
| "loss": 0.1036, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.026424084903617067, | |
| "grad_norm": 0.9996243119239807, | |
| "learning_rate": 8.773547901113862e-05, | |
| "loss": 0.1479, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.026857266623348496, | |
| "grad_norm": 1.146002173423767, | |
| "learning_rate": 8.715724127386972e-05, | |
| "loss": 0.1592, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.02729044834307992, | |
| "grad_norm": 0.6335452198982239, | |
| "learning_rate": 8.656768508095853e-05, | |
| "loss": 0.1147, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.02772363006281135, | |
| "grad_norm": 0.46830275654792786, | |
| "learning_rate": 8.596699001693255e-05, | |
| "loss": 0.1149, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.028156811782542777, | |
| "grad_norm": 0.5706634521484375, | |
| "learning_rate": 8.535533905932738e-05, | |
| "loss": 0.1181, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.028589993502274202, | |
| "grad_norm": 0.4780918061733246, | |
| "learning_rate": 8.473291852294987e-05, | |
| "loss": 0.0775, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.02902317522200563, | |
| "grad_norm": 0.45546865463256836, | |
| "learning_rate": 8.409991800312493e-05, | |
| "loss": 0.0855, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.02945635694173706, | |
| "grad_norm": 0.8334485292434692, | |
| "learning_rate": 8.345653031794292e-05, | |
| "loss": 0.0958, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.029889538661468484, | |
| "grad_norm": 0.6183558702468872, | |
| "learning_rate": 8.280295144952536e-05, | |
| "loss": 0.0958, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.030322720381199913, | |
| "grad_norm": 0.7801946997642517, | |
| "learning_rate": 8.213938048432697e-05, | |
| "loss": 0.1081, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.03075590210093134, | |
| "grad_norm": 0.5213829278945923, | |
| "learning_rate": 8.146601955249188e-05, | |
| "loss": 0.0898, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.031189083820662766, | |
| "grad_norm": 0.6553680300712585, | |
| "learning_rate": 8.07830737662829e-05, | |
| "loss": 0.1158, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.0316222655403942, | |
| "grad_norm": 0.6341337561607361, | |
| "learning_rate": 8.009075115760243e-05, | |
| "loss": 0.0805, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.03205544726012562, | |
| "grad_norm": 0.47776395082473755, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 0.0671, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.03248862897985705, | |
| "grad_norm": 0.6857931017875671, | |
| "learning_rate": 7.86788218175523e-05, | |
| "loss": 0.0757, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.03292181069958848, | |
| "grad_norm": 0.7055604457855225, | |
| "learning_rate": 7.795964517353735e-05, | |
| "loss": 0.0995, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.033354992419319905, | |
| "grad_norm": 0.8226217031478882, | |
| "learning_rate": 7.723195175075136e-05, | |
| "loss": 0.0734, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.03378817413905133, | |
| "grad_norm": 0.6480997204780579, | |
| "learning_rate": 7.649596321166024e-05, | |
| "loss": 0.0708, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.03422135585878276, | |
| "grad_norm": 0.5634680390357971, | |
| "learning_rate": 7.575190374550272e-05, | |
| "loss": 0.06, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.03465453757851419, | |
| "grad_norm": 0.5001603364944458, | |
| "learning_rate": 7.500000000000001e-05, | |
| "loss": 0.0499, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.03508771929824561, | |
| "grad_norm": 0.3587808609008789, | |
| "learning_rate": 7.424048101231686e-05, | |
| "loss": 0.0339, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.035520901017977044, | |
| "grad_norm": 0.5878409147262573, | |
| "learning_rate": 7.347357813929454e-05, | |
| "loss": 0.0634, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.03595408273770847, | |
| "grad_norm": 0.5627532005310059, | |
| "learning_rate": 7.269952498697734e-05, | |
| "loss": 0.0498, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.036387264457439894, | |
| "grad_norm": 0.522148847579956, | |
| "learning_rate": 7.191855733945387e-05, | |
| "loss": 0.0239, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.036820446177171326, | |
| "grad_norm": 0.6190194487571716, | |
| "learning_rate": 7.113091308703498e-05, | |
| "loss": 0.0461, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.03725362789690275, | |
| "grad_norm": 0.5469635128974915, | |
| "learning_rate": 7.033683215379002e-05, | |
| "loss": 0.0413, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.037686809616634176, | |
| "grad_norm": 0.3395079970359802, | |
| "learning_rate": 6.953655642446368e-05, | |
| "loss": 0.0338, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.03811999133636561, | |
| "grad_norm": 0.8697502017021179, | |
| "learning_rate": 6.873032967079561e-05, | |
| "loss": 0.0772, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.03855317305609703, | |
| "grad_norm": 0.6907822489738464, | |
| "learning_rate": 6.7918397477265e-05, | |
| "loss": 0.0291, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.03898635477582846, | |
| "grad_norm": 0.47123125195503235, | |
| "learning_rate": 6.710100716628344e-05, | |
| "loss": 0.0268, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.03941953649555989, | |
| "grad_norm": 0.2734449803829193, | |
| "learning_rate": 6.627840772285784e-05, | |
| "loss": 0.012, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.039852718215291315, | |
| "grad_norm": 0.27939316630363464, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 0.0145, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.04028589993502274, | |
| "grad_norm": 0.2553471028804779, | |
| "learning_rate": 6.461858523613684e-05, | |
| "loss": 0.0174, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.04071908165475417, | |
| "grad_norm": 0.15355971455574036, | |
| "learning_rate": 6.378186779084995e-05, | |
| "loss": 0.006, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.0411522633744856, | |
| "grad_norm": 0.2645378112792969, | |
| "learning_rate": 6.294095225512603e-05, | |
| "loss": 0.0155, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.04158544509421702, | |
| "grad_norm": 0.36685898900032043, | |
| "learning_rate": 6.209609477998338e-05, | |
| "loss": 0.0247, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.042018626813948454, | |
| "grad_norm": 0.24984949827194214, | |
| "learning_rate": 6.124755271719325e-05, | |
| "loss": 0.0127, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.04245180853367988, | |
| "grad_norm": 0.42770883440971375, | |
| "learning_rate": 6.0395584540887963e-05, | |
| "loss": 0.0153, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.042884990253411304, | |
| "grad_norm": 0.01884501427412033, | |
| "learning_rate": 5.9540449768827246e-05, | |
| "loss": 0.0004, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.043318171973142736, | |
| "grad_norm": 0.1676054745912552, | |
| "learning_rate": 5.868240888334653e-05, | |
| "loss": 0.0012, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.043318171973142736, | |
| "eval_loss": 0.08056939393281937, | |
| "eval_runtime": 205.2456, | |
| "eval_samples_per_second": 18.943, | |
| "eval_steps_per_second": 9.472, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.04375135369287416, | |
| "grad_norm": 0.9752314686775208, | |
| "learning_rate": 5.782172325201155e-05, | |
| "loss": 0.1654, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.044184535412605586, | |
| "grad_norm": 0.903446614742279, | |
| "learning_rate": 5.695865504800327e-05, | |
| "loss": 0.1832, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.04461771713233702, | |
| "grad_norm": 0.6539014577865601, | |
| "learning_rate": 5.6093467170257374e-05, | |
| "loss": 0.1592, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.04505089885206844, | |
| "grad_norm": 0.4455128014087677, | |
| "learning_rate": 5.522642316338268e-05, | |
| "loss": 0.1276, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.04548408057179987, | |
| "grad_norm": 0.5171334743499756, | |
| "learning_rate": 5.435778713738292e-05, | |
| "loss": 0.1779, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.0459172622915313, | |
| "grad_norm": 0.6053957343101501, | |
| "learning_rate": 5.348782368720626e-05, | |
| "loss": 0.1559, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.046350444011262724, | |
| "grad_norm": 0.5147733092308044, | |
| "learning_rate": 5.26167978121472e-05, | |
| "loss": 0.1247, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.04678362573099415, | |
| "grad_norm": 0.6188340783119202, | |
| "learning_rate": 5.174497483512506e-05, | |
| "loss": 0.1215, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.04721680745072558, | |
| "grad_norm": 0.4769625663757324, | |
| "learning_rate": 5.0872620321864185e-05, | |
| "loss": 0.1144, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.047649989170457006, | |
| "grad_norm": 0.5022192597389221, | |
| "learning_rate": 5e-05, | |
| "loss": 0.1214, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.04808317089018843, | |
| "grad_norm": 0.3280904293060303, | |
| "learning_rate": 4.912737967813583e-05, | |
| "loss": 0.0796, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.04851635260991986, | |
| "grad_norm": 0.5279192328453064, | |
| "learning_rate": 4.825502516487497e-05, | |
| "loss": 0.1151, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.04894953432965129, | |
| "grad_norm": 0.47400593757629395, | |
| "learning_rate": 4.738320218785281e-05, | |
| "loss": 0.1186, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.04938271604938271, | |
| "grad_norm": 0.6582249999046326, | |
| "learning_rate": 4.6512176312793736e-05, | |
| "loss": 0.0961, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.049815897769114145, | |
| "grad_norm": 0.40480566024780273, | |
| "learning_rate": 4.564221286261709e-05, | |
| "loss": 0.0765, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.05024907948884557, | |
| "grad_norm": 0.3763730823993683, | |
| "learning_rate": 4.477357683661734e-05, | |
| "loss": 0.0739, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.050682261208576995, | |
| "grad_norm": 0.40720927715301514, | |
| "learning_rate": 4.390653282974264e-05, | |
| "loss": 0.0798, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.05111544292830843, | |
| "grad_norm": 0.7141808271408081, | |
| "learning_rate": 4.3041344951996746e-05, | |
| "loss": 0.0874, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.05154862464803985, | |
| "grad_norm": 0.5295721888542175, | |
| "learning_rate": 4.2178276747988446e-05, | |
| "loss": 0.0861, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.05198180636777128, | |
| "grad_norm": 0.3550686240196228, | |
| "learning_rate": 4.131759111665349e-05, | |
| "loss": 0.0749, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.05241498808750271, | |
| "grad_norm": 0.5598456859588623, | |
| "learning_rate": 4.045955023117276e-05, | |
| "loss": 0.139, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.052848169807234134, | |
| "grad_norm": 0.3951553702354431, | |
| "learning_rate": 3.960441545911204e-05, | |
| "loss": 0.0669, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.05328135152696556, | |
| "grad_norm": 0.5975047945976257, | |
| "learning_rate": 3.875244728280676e-05, | |
| "loss": 0.1032, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.05371453324669699, | |
| "grad_norm": 0.49492087960243225, | |
| "learning_rate": 3.790390522001662e-05, | |
| "loss": 0.0682, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.054147714966428416, | |
| "grad_norm": 0.540847897529602, | |
| "learning_rate": 3.705904774487396e-05, | |
| "loss": 0.0871, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.05458089668615984, | |
| "grad_norm": 0.44237419962882996, | |
| "learning_rate": 3.6218132209150045e-05, | |
| "loss": 0.0571, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.05501407840589127, | |
| "grad_norm": 0.5057970881462097, | |
| "learning_rate": 3.5381414763863166e-05, | |
| "loss": 0.0885, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.0554472601256227, | |
| "grad_norm": 0.3728044629096985, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 0.0395, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.05588044184535412, | |
| "grad_norm": 0.39620381593704224, | |
| "learning_rate": 3.372159227714218e-05, | |
| "loss": 0.048, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.056313623565085555, | |
| "grad_norm": 0.5850706696510315, | |
| "learning_rate": 3.289899283371657e-05, | |
| "loss": 0.0673, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.05674680528481698, | |
| "grad_norm": 0.4705985486507416, | |
| "learning_rate": 3.2081602522734986e-05, | |
| "loss": 0.0457, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.057179987004548405, | |
| "grad_norm": 0.39884859323501587, | |
| "learning_rate": 3.12696703292044e-05, | |
| "loss": 0.0609, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.05761316872427984, | |
| "grad_norm": 0.34650862216949463, | |
| "learning_rate": 3.046344357553632e-05, | |
| "loss": 0.0439, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.05804635044401126, | |
| "grad_norm": 0.5266785621643066, | |
| "learning_rate": 2.9663167846209998e-05, | |
| "loss": 0.0665, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.05847953216374269, | |
| "grad_norm": 0.23444664478302002, | |
| "learning_rate": 2.886908691296504e-05, | |
| "loss": 0.0308, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.05891271388347412, | |
| "grad_norm": 0.25425028800964355, | |
| "learning_rate": 2.8081442660546125e-05, | |
| "loss": 0.0171, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.059345895603205544, | |
| "grad_norm": 0.40159934759140015, | |
| "learning_rate": 2.7300475013022663e-05, | |
| "loss": 0.0369, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.05977907732293697, | |
| "grad_norm": 0.32667964696884155, | |
| "learning_rate": 2.6526421860705473e-05, | |
| "loss": 0.0294, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.0602122590426684, | |
| "grad_norm": 0.40988171100616455, | |
| "learning_rate": 2.575951898768315e-05, | |
| "loss": 0.039, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.060645440762399826, | |
| "grad_norm": 0.23806576430797577, | |
| "learning_rate": 2.500000000000001e-05, | |
| "loss": 0.0233, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.06107862248213125, | |
| "grad_norm": 0.8812161087989807, | |
| "learning_rate": 2.4248096254497288e-05, | |
| "loss": 0.043, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.06151180420186268, | |
| "grad_norm": 0.38497522473335266, | |
| "learning_rate": 2.350403678833976e-05, | |
| "loss": 0.0168, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.06194498592159411, | |
| "grad_norm": 0.2601085901260376, | |
| "learning_rate": 2.2768048249248648e-05, | |
| "loss": 0.0048, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.06237816764132553, | |
| "grad_norm": 0.2766932249069214, | |
| "learning_rate": 2.2040354826462668e-05, | |
| "loss": 0.0226, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.06281134936105696, | |
| "grad_norm": 0.09686120599508286, | |
| "learning_rate": 2.132117818244771e-05, | |
| "loss": 0.0024, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.0632445310807884, | |
| "grad_norm": 0.08909226208925247, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 0.0031, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.06367771280051981, | |
| "grad_norm": 0.2203822135925293, | |
| "learning_rate": 1.9909248842397584e-05, | |
| "loss": 0.0149, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.06411089452025125, | |
| "grad_norm": 0.12339666485786438, | |
| "learning_rate": 1.9216926233717085e-05, | |
| "loss": 0.0041, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.06454407623998268, | |
| "grad_norm": 0.2815093696117401, | |
| "learning_rate": 1.8533980447508137e-05, | |
| "loss": 0.0132, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.0649772579597141, | |
| "grad_norm": 0.0794215053319931, | |
| "learning_rate": 1.7860619515673033e-05, | |
| "loss": 0.0021, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.0649772579597141, | |
| "eval_loss": 0.05704093724489212, | |
| "eval_runtime": 204.8642, | |
| "eval_samples_per_second": 18.978, | |
| "eval_steps_per_second": 9.489, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.06541043967944553, | |
| "grad_norm": 0.3460334539413452, | |
| "learning_rate": 1.7197048550474643e-05, | |
| "loss": 0.1264, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.06584362139917696, | |
| "grad_norm": 0.30614346265792847, | |
| "learning_rate": 1.6543469682057106e-05, | |
| "loss": 0.0801, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.06627680311890838, | |
| "grad_norm": 0.39302000403404236, | |
| "learning_rate": 1.5900081996875083e-05, | |
| "loss": 0.1032, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.06670998483863981, | |
| "grad_norm": 0.45320767164230347, | |
| "learning_rate": 1.526708147705013e-05, | |
| "loss": 0.1144, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.06714316655837124, | |
| "grad_norm": 0.4174154996871948, | |
| "learning_rate": 1.4644660940672627e-05, | |
| "loss": 0.1109, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.06757634827810266, | |
| "grad_norm": 0.3307155966758728, | |
| "learning_rate": 1.4033009983067452e-05, | |
| "loss": 0.09, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.06800952999783409, | |
| "grad_norm": 0.35494112968444824, | |
| "learning_rate": 1.3432314919041478e-05, | |
| "loss": 0.0815, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.06844271171756552, | |
| "grad_norm": 0.39967793226242065, | |
| "learning_rate": 1.2842758726130283e-05, | |
| "loss": 0.1042, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.06887589343729694, | |
| "grad_norm": 0.5634881258010864, | |
| "learning_rate": 1.22645209888614e-05, | |
| "loss": 0.1295, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.06930907515702837, | |
| "grad_norm": 0.371140718460083, | |
| "learning_rate": 1.1697777844051105e-05, | |
| "loss": 0.0926, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.0697422568767598, | |
| "grad_norm": 0.35622888803482056, | |
| "learning_rate": 1.1142701927151456e-05, | |
| "loss": 0.0781, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.07017543859649122, | |
| "grad_norm": 0.43054744601249695, | |
| "learning_rate": 1.0599462319663905e-05, | |
| "loss": 0.0843, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.07060862031622266, | |
| "grad_norm": 0.3432992100715637, | |
| "learning_rate": 1.006822449763537e-05, | |
| "loss": 0.0529, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.07104180203595409, | |
| "grad_norm": 0.2856929898262024, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 0.0602, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.0714749837556855, | |
| "grad_norm": 0.3701120615005493, | |
| "learning_rate": 9.042397785550405e-06, | |
| "loss": 0.0895, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.07190816547541694, | |
| "grad_norm": 0.37228187918663025, | |
| "learning_rate": 8.548121372247918e-06, | |
| "loss": 0.0793, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.07234134719514837, | |
| "grad_norm": 0.3235107958316803, | |
| "learning_rate": 8.066471602728803e-06, | |
| "loss": 0.0589, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.07277452891487979, | |
| "grad_norm": 0.4608737528324127, | |
| "learning_rate": 7.597595192178702e-06, | |
| "loss": 0.0802, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.07320771063461122, | |
| "grad_norm": 0.3900294005870819, | |
| "learning_rate": 7.1416349648943894e-06, | |
| "loss": 0.0829, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.07364089235434265, | |
| "grad_norm": 0.4117908775806427, | |
| "learning_rate": 6.698729810778065e-06, | |
| "loss": 0.0706, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.07407407407407407, | |
| "grad_norm": 0.41260048747062683, | |
| "learning_rate": 6.269014643030213e-06, | |
| "loss": 0.0834, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.0745072557938055, | |
| "grad_norm": 0.3703679144382477, | |
| "learning_rate": 5.852620357053651e-06, | |
| "loss": 0.056, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.07494043751353693, | |
| "grad_norm": 0.382755309343338, | |
| "learning_rate": 5.449673790581611e-06, | |
| "loss": 0.0484, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.07537361923326835, | |
| "grad_norm": 0.5890815854072571, | |
| "learning_rate": 5.060297685041659e-06, | |
| "loss": 0.1126, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.07580680095299978, | |
| "grad_norm": 0.3182831406593323, | |
| "learning_rate": 4.684610648167503e-06, | |
| "loss": 0.0543, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.07623998267273122, | |
| "grad_norm": 0.44664430618286133, | |
| "learning_rate": 4.322727117869951e-06, | |
| "loss": 0.0824, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.07667316439246263, | |
| "grad_norm": 0.7139434814453125, | |
| "learning_rate": 3.974757327377981e-06, | |
| "loss": 0.0688, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.07710634611219407, | |
| "grad_norm": 0.374163955450058, | |
| "learning_rate": 3.6408072716606346e-06, | |
| "loss": 0.0508, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.0775395278319255, | |
| "grad_norm": 0.35698890686035156, | |
| "learning_rate": 3.3209786751399187e-06, | |
| "loss": 0.042, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.07797270955165692, | |
| "grad_norm": 0.38510996103286743, | |
| "learning_rate": 3.0153689607045845e-06, | |
| "loss": 0.0394, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.07840589127138835, | |
| "grad_norm": 0.5943405628204346, | |
| "learning_rate": 2.724071220034158e-06, | |
| "loss": 0.0716, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.07883907299111978, | |
| "grad_norm": 0.5582049489021301, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 0.0563, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.0792722547108512, | |
| "grad_norm": 0.2697021961212158, | |
| "learning_rate": 2.1847622018482283e-06, | |
| "loss": 0.0311, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.07970543643058263, | |
| "grad_norm": 0.3521104156970978, | |
| "learning_rate": 1.9369152030840556e-06, | |
| "loss": 0.048, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.08013861815031406, | |
| "grad_norm": 0.301751971244812, | |
| "learning_rate": 1.70370868554659e-06, | |
| "loss": 0.0218, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.08057179987004548, | |
| "grad_norm": 0.4299044609069824, | |
| "learning_rate": 1.4852136862001764e-06, | |
| "loss": 0.0366, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.08100498158977691, | |
| "grad_norm": 0.48658519983291626, | |
| "learning_rate": 1.2814967607382432e-06, | |
| "loss": 0.0139, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.08143816330950834, | |
| "grad_norm": 0.08318212628364563, | |
| "learning_rate": 1.0926199633097157e-06, | |
| "loss": 0.003, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.08187134502923976, | |
| "grad_norm": 0.34329327940940857, | |
| "learning_rate": 9.186408276168013e-07, | |
| "loss": 0.0281, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.0823045267489712, | |
| "grad_norm": 0.23912468552589417, | |
| "learning_rate": 7.596123493895991e-07, | |
| "loss": 0.0172, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.08273770846870263, | |
| "grad_norm": 0.19108746945858002, | |
| "learning_rate": 6.15582970243117e-07, | |
| "loss": 0.012, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.08317089018843404, | |
| "grad_norm": 0.673153281211853, | |
| "learning_rate": 4.865965629214819e-07, | |
| "loss": 0.0185, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.08360407190816548, | |
| "grad_norm": 0.28818559646606445, | |
| "learning_rate": 3.7269241793390085e-07, | |
| "loss": 0.0166, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.08403725362789691, | |
| "grad_norm": 0.23559705913066864, | |
| "learning_rate": 2.7390523158633554e-07, | |
| "loss": 0.0113, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.08447043534762833, | |
| "grad_norm": 0.02800111472606659, | |
| "learning_rate": 1.9026509541272275e-07, | |
| "loss": 0.001, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.08490361706735976, | |
| "grad_norm": 0.16618917882442474, | |
| "learning_rate": 1.2179748700879012e-07, | |
| "loss": 0.006, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.08533679878709119, | |
| "grad_norm": 0.30197080969810486, | |
| "learning_rate": 6.852326227130834e-08, | |
| "loss": 0.0095, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.08576998050682261, | |
| "grad_norm": 0.1305372416973114, | |
| "learning_rate": 3.04586490452119e-08, | |
| "loss": 0.0032, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.08620316222655404, | |
| "grad_norm": 0.37086614966392517, | |
| "learning_rate": 7.615242180436522e-09, | |
| "loss": 0.0082, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.08663634394628547, | |
| "grad_norm": 0.013217504136264324, | |
| "learning_rate": 0.0, | |
| "loss": 0.0003, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.08663634394628547, | |
| "eval_loss": 0.05274021625518799, | |
| "eval_runtime": 204.9157, | |
| "eval_samples_per_second": 18.974, | |
| "eval_steps_per_second": 9.487, | |
| "step": 200 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 200, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 50, | |
| "stateful_callbacks": { | |
| "EarlyStoppingCallback": { | |
| "args": { | |
| "early_stopping_patience": 5, | |
| "early_stopping_threshold": 0.0 | |
| }, | |
| "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": 1.1410122874866893e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |