| { | |
| "best_metric": 1.6074018478393555, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.02050020500205002, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0001025010250102501, | |
| "grad_norm": 0.705419659614563, | |
| "learning_rate": 1.001e-05, | |
| "loss": 1.8146, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0001025010250102501, | |
| "eval_loss": 2.0756940841674805, | |
| "eval_runtime": 214.3165, | |
| "eval_samples_per_second": 19.168, | |
| "eval_steps_per_second": 4.792, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0002050020500205002, | |
| "grad_norm": 0.9089563488960266, | |
| "learning_rate": 2.002e-05, | |
| "loss": 1.5119, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0003075030750307503, | |
| "grad_norm": 0.9483783841133118, | |
| "learning_rate": 3.0029999999999995e-05, | |
| "loss": 1.6709, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.0004100041000410004, | |
| "grad_norm": 1.0198091268539429, | |
| "learning_rate": 4.004e-05, | |
| "loss": 1.9187, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.0005125051250512505, | |
| "grad_norm": 1.091034173965454, | |
| "learning_rate": 5.005e-05, | |
| "loss": 1.8457, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0006150061500615006, | |
| "grad_norm": 1.2312097549438477, | |
| "learning_rate": 6.005999999999999e-05, | |
| "loss": 2.1334, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.0007175071750717508, | |
| "grad_norm": 1.246527910232544, | |
| "learning_rate": 7.006999999999998e-05, | |
| "loss": 1.562, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.0008200082000820008, | |
| "grad_norm": 1.4241694211959839, | |
| "learning_rate": 8.008e-05, | |
| "loss": 1.7474, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.0009225092250922509, | |
| "grad_norm": 1.28750479221344, | |
| "learning_rate": 9.009e-05, | |
| "loss": 2.0525, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.001025010250102501, | |
| "grad_norm": 1.2573869228363037, | |
| "learning_rate": 0.0001001, | |
| "loss": 2.0545, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.001127511275112751, | |
| "grad_norm": 1.154963731765747, | |
| "learning_rate": 9.957315789473684e-05, | |
| "loss": 1.7098, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.0012300123001230013, | |
| "grad_norm": 1.1779276132583618, | |
| "learning_rate": 9.904631578947367e-05, | |
| "loss": 1.9339, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.0013325133251332513, | |
| "grad_norm": 1.2296435832977295, | |
| "learning_rate": 9.851947368421052e-05, | |
| "loss": 1.8391, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.0014350143501435015, | |
| "grad_norm": 1.2227979898452759, | |
| "learning_rate": 9.799263157894736e-05, | |
| "loss": 1.8549, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.0015375153751537515, | |
| "grad_norm": 1.230628252029419, | |
| "learning_rate": 9.746578947368421e-05, | |
| "loss": 1.6791, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.0016400164001640015, | |
| "grad_norm": 1.2459157705307007, | |
| "learning_rate": 9.693894736842104e-05, | |
| "loss": 1.8406, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.0017425174251742518, | |
| "grad_norm": 1.397838830947876, | |
| "learning_rate": 9.641210526315789e-05, | |
| "loss": 1.796, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.0018450184501845018, | |
| "grad_norm": 1.3623087406158447, | |
| "learning_rate": 9.588526315789473e-05, | |
| "loss": 1.6451, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.001947519475194752, | |
| "grad_norm": 1.386612057685852, | |
| "learning_rate": 9.535842105263157e-05, | |
| "loss": 1.8315, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.002050020500205002, | |
| "grad_norm": 1.4097331762313843, | |
| "learning_rate": 9.483157894736841e-05, | |
| "loss": 1.8867, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.002152521525215252, | |
| "grad_norm": 1.2772889137268066, | |
| "learning_rate": 9.430473684210526e-05, | |
| "loss": 1.8137, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.002255022550225502, | |
| "grad_norm": 1.3513169288635254, | |
| "learning_rate": 9.37778947368421e-05, | |
| "loss": 1.6449, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.0023575235752357525, | |
| "grad_norm": 1.5203684568405151, | |
| "learning_rate": 9.325105263157894e-05, | |
| "loss": 1.7515, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.0024600246002460025, | |
| "grad_norm": 1.2085375785827637, | |
| "learning_rate": 9.272421052631578e-05, | |
| "loss": 1.4646, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.0025625256252562525, | |
| "grad_norm": 1.4566919803619385, | |
| "learning_rate": 9.219736842105263e-05, | |
| "loss": 1.8844, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.0026650266502665026, | |
| "grad_norm": 1.5748794078826904, | |
| "learning_rate": 9.167052631578946e-05, | |
| "loss": 1.7089, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.0027675276752767526, | |
| "grad_norm": 1.3820806741714478, | |
| "learning_rate": 9.114368421052632e-05, | |
| "loss": 1.7397, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.002870028700287003, | |
| "grad_norm": 1.47676420211792, | |
| "learning_rate": 9.061684210526315e-05, | |
| "loss": 1.6333, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.002972529725297253, | |
| "grad_norm": 1.3956615924835205, | |
| "learning_rate": 9.009e-05, | |
| "loss": 1.5381, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.003075030750307503, | |
| "grad_norm": 1.4320456981658936, | |
| "learning_rate": 8.956315789473683e-05, | |
| "loss": 1.5204, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.003177531775317753, | |
| "grad_norm": 1.564431071281433, | |
| "learning_rate": 8.903631578947368e-05, | |
| "loss": 1.6053, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.003280032800328003, | |
| "grad_norm": 1.604783296585083, | |
| "learning_rate": 8.850947368421052e-05, | |
| "loss": 1.4296, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.0033825338253382535, | |
| "grad_norm": 2.1952967643737793, | |
| "learning_rate": 8.798263157894736e-05, | |
| "loss": 2.3028, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.0034850348503485036, | |
| "grad_norm": 1.7859286069869995, | |
| "learning_rate": 8.745578947368422e-05, | |
| "loss": 1.0892, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.0035875358753587536, | |
| "grad_norm": 2.099475145339966, | |
| "learning_rate": 8.692894736842105e-05, | |
| "loss": 1.9027, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.0036900369003690036, | |
| "grad_norm": 2.05716609954834, | |
| "learning_rate": 8.64021052631579e-05, | |
| "loss": 1.8596, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.0037925379253792536, | |
| "grad_norm": 3.045736074447632, | |
| "learning_rate": 8.587526315789473e-05, | |
| "loss": 1.6003, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.003895038950389504, | |
| "grad_norm": 3.077202320098877, | |
| "learning_rate": 8.534842105263157e-05, | |
| "loss": 1.7161, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.003997539975399754, | |
| "grad_norm": 2.8727805614471436, | |
| "learning_rate": 8.482157894736842e-05, | |
| "loss": 1.2155, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.004100041000410004, | |
| "grad_norm": 3.2088754177093506, | |
| "learning_rate": 8.429473684210525e-05, | |
| "loss": 2.3325, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.0042025420254202545, | |
| "grad_norm": 3.467123031616211, | |
| "learning_rate": 8.376789473684211e-05, | |
| "loss": 1.7372, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.004305043050430504, | |
| "grad_norm": 3.2959988117218018, | |
| "learning_rate": 8.324105263157894e-05, | |
| "loss": 2.0588, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.0044075440754407546, | |
| "grad_norm": 3.0905165672302246, | |
| "learning_rate": 8.271421052631579e-05, | |
| "loss": 1.793, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.004510045100451004, | |
| "grad_norm": 4.93464469909668, | |
| "learning_rate": 8.218736842105262e-05, | |
| "loss": 1.8903, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.004612546125461255, | |
| "grad_norm": 2.7704639434814453, | |
| "learning_rate": 8.166052631578947e-05, | |
| "loss": 1.1567, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.004715047150471505, | |
| "grad_norm": 3.1440534591674805, | |
| "learning_rate": 8.113368421052631e-05, | |
| "loss": 1.568, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.004817548175481755, | |
| "grad_norm": 4.074300289154053, | |
| "learning_rate": 8.060684210526315e-05, | |
| "loss": 1.9814, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.004920049200492005, | |
| "grad_norm": 5.697752475738525, | |
| "learning_rate": 8.008e-05, | |
| "loss": 1.862, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.005022550225502255, | |
| "grad_norm": 6.887198448181152, | |
| "learning_rate": 7.955315789473684e-05, | |
| "loss": 1.7794, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.005125051250512505, | |
| "grad_norm": 6.863124847412109, | |
| "learning_rate": 7.902631578947368e-05, | |
| "loss": 1.5001, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.005125051250512505, | |
| "eval_loss": 1.7083282470703125, | |
| "eval_runtime": 213.8938, | |
| "eval_samples_per_second": 19.206, | |
| "eval_steps_per_second": 4.801, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.0052275522755227555, | |
| "grad_norm": 0.7324660420417786, | |
| "learning_rate": 7.849947368421052e-05, | |
| "loss": 1.7833, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.005330053300533005, | |
| "grad_norm": 1.0079008340835571, | |
| "learning_rate": 7.797263157894736e-05, | |
| "loss": 2.0611, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.005432554325543256, | |
| "grad_norm": 1.0478756427764893, | |
| "learning_rate": 7.744578947368421e-05, | |
| "loss": 1.5655, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.005535055350553505, | |
| "grad_norm": 1.1489282846450806, | |
| "learning_rate": 7.691894736842104e-05, | |
| "loss": 2.0112, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.005637556375563756, | |
| "grad_norm": 1.0685741901397705, | |
| "learning_rate": 7.63921052631579e-05, | |
| "loss": 1.837, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.005740057400574006, | |
| "grad_norm": 0.9752334952354431, | |
| "learning_rate": 7.586526315789473e-05, | |
| "loss": 1.8408, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.005842558425584256, | |
| "grad_norm": 1.0988349914550781, | |
| "learning_rate": 7.533842105263158e-05, | |
| "loss": 1.6708, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.005945059450594506, | |
| "grad_norm": 1.059479832649231, | |
| "learning_rate": 7.481157894736841e-05, | |
| "loss": 1.4579, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.006047560475604756, | |
| "grad_norm": 1.0964906215667725, | |
| "learning_rate": 7.428473684210526e-05, | |
| "loss": 1.944, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.006150061500615006, | |
| "grad_norm": 1.0522820949554443, | |
| "learning_rate": 7.375789473684209e-05, | |
| "loss": 1.762, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.0062525625256252566, | |
| "grad_norm": 1.0768394470214844, | |
| "learning_rate": 7.323105263157895e-05, | |
| "loss": 1.7185, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.006355063550635506, | |
| "grad_norm": 1.1058900356292725, | |
| "learning_rate": 7.270421052631578e-05, | |
| "loss": 1.6855, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.006457564575645757, | |
| "grad_norm": 1.0793980360031128, | |
| "learning_rate": 7.217736842105263e-05, | |
| "loss": 1.6553, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.006560065600656006, | |
| "grad_norm": 1.257716417312622, | |
| "learning_rate": 7.165052631578947e-05, | |
| "loss": 1.5379, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.006662566625666257, | |
| "grad_norm": 1.116843581199646, | |
| "learning_rate": 7.11236842105263e-05, | |
| "loss": 1.6653, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.006765067650676507, | |
| "grad_norm": 1.016861081123352, | |
| "learning_rate": 7.059684210526315e-05, | |
| "loss": 1.3905, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.006867568675686757, | |
| "grad_norm": 1.1021226644515991, | |
| "learning_rate": 7.006999999999998e-05, | |
| "loss": 1.6295, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.006970069700697007, | |
| "grad_norm": 1.1197590827941895, | |
| "learning_rate": 6.954315789473684e-05, | |
| "loss": 1.6643, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.007072570725707257, | |
| "grad_norm": 1.242423415184021, | |
| "learning_rate": 6.901631578947368e-05, | |
| "loss": 1.7367, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.007175071750717507, | |
| "grad_norm": 1.1498782634735107, | |
| "learning_rate": 6.848947368421052e-05, | |
| "loss": 1.6476, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.007277572775727758, | |
| "grad_norm": 1.174621820449829, | |
| "learning_rate": 6.796263157894737e-05, | |
| "loss": 2.0002, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.007380073800738007, | |
| "grad_norm": 1.1680294275283813, | |
| "learning_rate": 6.74357894736842e-05, | |
| "loss": 1.4689, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.007482574825748258, | |
| "grad_norm": 1.3079551458358765, | |
| "learning_rate": 6.690894736842105e-05, | |
| "loss": 1.5757, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.007585075850758507, | |
| "grad_norm": 1.2481780052185059, | |
| "learning_rate": 6.638210526315788e-05, | |
| "loss": 1.6238, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.007687576875768758, | |
| "grad_norm": 1.3879331350326538, | |
| "learning_rate": 6.585526315789474e-05, | |
| "loss": 1.9082, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.007790077900779008, | |
| "grad_norm": 1.6329386234283447, | |
| "learning_rate": 6.532842105263157e-05, | |
| "loss": 1.739, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.007892578925789259, | |
| "grad_norm": 1.4674111604690552, | |
| "learning_rate": 6.480157894736842e-05, | |
| "loss": 1.6045, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.007995079950799507, | |
| "grad_norm": 1.4382115602493286, | |
| "learning_rate": 6.427473684210526e-05, | |
| "loss": 1.4885, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.008097580975809758, | |
| "grad_norm": 1.4571962356567383, | |
| "learning_rate": 6.37478947368421e-05, | |
| "loss": 1.3451, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.008200082000820008, | |
| "grad_norm": 1.6193132400512695, | |
| "learning_rate": 6.322105263157894e-05, | |
| "loss": 1.8344, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.008302583025830259, | |
| "grad_norm": 1.673788070678711, | |
| "learning_rate": 6.269421052631577e-05, | |
| "loss": 1.4983, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.008405084050840509, | |
| "grad_norm": 1.7791383266448975, | |
| "learning_rate": 6.216736842105263e-05, | |
| "loss": 1.8181, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.008507585075850758, | |
| "grad_norm": 1.6379783153533936, | |
| "learning_rate": 6.164052631578947e-05, | |
| "loss": 1.5085, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.008610086100861008, | |
| "grad_norm": 2.0746712684631348, | |
| "learning_rate": 6.111368421052631e-05, | |
| "loss": 1.8331, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.008712587125871259, | |
| "grad_norm": 1.9016090631484985, | |
| "learning_rate": 6.058684210526315e-05, | |
| "loss": 1.6955, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.008815088150881509, | |
| "grad_norm": 2.0374410152435303, | |
| "learning_rate": 6.005999999999999e-05, | |
| "loss": 1.6262, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.00891758917589176, | |
| "grad_norm": 2.458918809890747, | |
| "learning_rate": 5.953315789473684e-05, | |
| "loss": 1.782, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.009020090200902008, | |
| "grad_norm": 2.420497179031372, | |
| "learning_rate": 5.9006315789473676e-05, | |
| "loss": 1.4256, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.009122591225912259, | |
| "grad_norm": 3.1010406017303467, | |
| "learning_rate": 5.847947368421053e-05, | |
| "loss": 1.5861, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.00922509225092251, | |
| "grad_norm": 2.9462192058563232, | |
| "learning_rate": 5.795263157894737e-05, | |
| "loss": 1.7643, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.00932759327593276, | |
| "grad_norm": 3.0921950340270996, | |
| "learning_rate": 5.742578947368421e-05, | |
| "loss": 1.7482, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.00943009430094301, | |
| "grad_norm": 3.2740588188171387, | |
| "learning_rate": 5.6898947368421046e-05, | |
| "loss": 1.7103, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.009532595325953259, | |
| "grad_norm": 3.282179594039917, | |
| "learning_rate": 5.6372105263157886e-05, | |
| "loss": 1.3958, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.00963509635096351, | |
| "grad_norm": 2.8050053119659424, | |
| "learning_rate": 5.584526315789473e-05, | |
| "loss": 1.0987, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.00973759737597376, | |
| "grad_norm": 3.167931079864502, | |
| "learning_rate": 5.531842105263158e-05, | |
| "loss": 1.0239, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.00984009840098401, | |
| "grad_norm": 4.01984167098999, | |
| "learning_rate": 5.4791578947368424e-05, | |
| "loss": 2.1924, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.00994259942599426, | |
| "grad_norm": 7.1500983238220215, | |
| "learning_rate": 5.426473684210526e-05, | |
| "loss": 1.6619, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.01004510045100451, | |
| "grad_norm": 3.5555317401885986, | |
| "learning_rate": 5.37378947368421e-05, | |
| "loss": 1.0794, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.01014760147601476, | |
| "grad_norm": 3.8048603534698486, | |
| "learning_rate": 5.321105263157894e-05, | |
| "loss": 1.4438, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.01025010250102501, | |
| "grad_norm": 6.511986255645752, | |
| "learning_rate": 5.268421052631578e-05, | |
| "loss": 2.3968, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.01025010250102501, | |
| "eval_loss": 1.6530330181121826, | |
| "eval_runtime": 213.675, | |
| "eval_samples_per_second": 19.225, | |
| "eval_steps_per_second": 4.806, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.01035260352603526, | |
| "grad_norm": 0.74406498670578, | |
| "learning_rate": 5.2157368421052626e-05, | |
| "loss": 2.0527, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.010455104551045511, | |
| "grad_norm": 0.8070314526557922, | |
| "learning_rate": 5.163052631578947e-05, | |
| "loss": 1.6314, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.01055760557605576, | |
| "grad_norm": 0.8939439058303833, | |
| "learning_rate": 5.110368421052632e-05, | |
| "loss": 1.8166, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.01066010660106601, | |
| "grad_norm": 0.9275736808776855, | |
| "learning_rate": 5.057684210526316e-05, | |
| "loss": 1.7837, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.01076260762607626, | |
| "grad_norm": 1.0159188508987427, | |
| "learning_rate": 5.005e-05, | |
| "loss": 1.8655, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.010865108651086511, | |
| "grad_norm": 0.9382534027099609, | |
| "learning_rate": 4.9523157894736836e-05, | |
| "loss": 1.634, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.010967609676096762, | |
| "grad_norm": 0.8507430553436279, | |
| "learning_rate": 4.899631578947368e-05, | |
| "loss": 1.6201, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.01107011070110701, | |
| "grad_norm": 0.9986007809638977, | |
| "learning_rate": 4.846947368421052e-05, | |
| "loss": 1.7325, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.01117261172611726, | |
| "grad_norm": 0.955406904220581, | |
| "learning_rate": 4.794263157894737e-05, | |
| "loss": 1.6368, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.011275112751127511, | |
| "grad_norm": 0.9369739294052124, | |
| "learning_rate": 4.7415789473684206e-05, | |
| "loss": 1.5455, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.011377613776137762, | |
| "grad_norm": 1.0813074111938477, | |
| "learning_rate": 4.688894736842105e-05, | |
| "loss": 1.6737, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.011480114801148012, | |
| "grad_norm": 1.034050703048706, | |
| "learning_rate": 4.636210526315789e-05, | |
| "loss": 1.5688, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.01158261582615826, | |
| "grad_norm": 1.0678492784500122, | |
| "learning_rate": 4.583526315789473e-05, | |
| "loss": 1.6165, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.011685116851168511, | |
| "grad_norm": 1.0256990194320679, | |
| "learning_rate": 4.530842105263158e-05, | |
| "loss": 1.3301, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.011787617876178762, | |
| "grad_norm": 1.0190774202346802, | |
| "learning_rate": 4.4781578947368416e-05, | |
| "loss": 1.6662, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.011890118901189012, | |
| "grad_norm": 1.0290199518203735, | |
| "learning_rate": 4.425473684210526e-05, | |
| "loss": 1.5645, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.011992619926199263, | |
| "grad_norm": 1.0103486776351929, | |
| "learning_rate": 4.372789473684211e-05, | |
| "loss": 1.2945, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.012095120951209511, | |
| "grad_norm": 1.2646342515945435, | |
| "learning_rate": 4.320105263157895e-05, | |
| "loss": 1.735, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.012197621976219762, | |
| "grad_norm": 1.2448196411132812, | |
| "learning_rate": 4.2674210526315786e-05, | |
| "loss": 1.5587, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.012300123001230012, | |
| "grad_norm": 1.3261544704437256, | |
| "learning_rate": 4.2147368421052626e-05, | |
| "loss": 1.7949, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.012402624026240263, | |
| "grad_norm": 1.2770674228668213, | |
| "learning_rate": 4.162052631578947e-05, | |
| "loss": 1.4959, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.012505125051250513, | |
| "grad_norm": 1.4941719770431519, | |
| "learning_rate": 4.109368421052631e-05, | |
| "loss": 1.5308, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.012607626076260762, | |
| "grad_norm": 1.2907934188842773, | |
| "learning_rate": 4.056684210526316e-05, | |
| "loss": 1.7877, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.012710127101271012, | |
| "grad_norm": 1.3961200714111328, | |
| "learning_rate": 4.004e-05, | |
| "loss": 1.7712, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.012812628126281263, | |
| "grad_norm": 1.4105072021484375, | |
| "learning_rate": 3.951315789473684e-05, | |
| "loss": 1.6403, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.012915129151291513, | |
| "grad_norm": 1.323488473892212, | |
| "learning_rate": 3.898631578947368e-05, | |
| "loss": 1.5076, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.013017630176301764, | |
| "grad_norm": 1.4488778114318848, | |
| "learning_rate": 3.845947368421052e-05, | |
| "loss": 1.655, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.013120131201312012, | |
| "grad_norm": 1.50008225440979, | |
| "learning_rate": 3.7932631578947367e-05, | |
| "loss": 1.7404, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.013222632226322263, | |
| "grad_norm": 1.4806251525878906, | |
| "learning_rate": 3.7405789473684206e-05, | |
| "loss": 1.4379, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.013325133251332513, | |
| "grad_norm": 1.5019195079803467, | |
| "learning_rate": 3.6878947368421045e-05, | |
| "loss": 1.5715, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.013427634276342764, | |
| "grad_norm": 1.620043158531189, | |
| "learning_rate": 3.635210526315789e-05, | |
| "loss": 1.0703, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.013530135301353014, | |
| "grad_norm": 1.8016197681427002, | |
| "learning_rate": 3.582526315789474e-05, | |
| "loss": 1.3446, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.013632636326363263, | |
| "grad_norm": 2.6074109077453613, | |
| "learning_rate": 3.5298421052631576e-05, | |
| "loss": 1.3707, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.013735137351373513, | |
| "grad_norm": 2.0289924144744873, | |
| "learning_rate": 3.477157894736842e-05, | |
| "loss": 1.836, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.013837638376383764, | |
| "grad_norm": 2.613877058029175, | |
| "learning_rate": 3.424473684210526e-05, | |
| "loss": 1.9227, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.013940139401394014, | |
| "grad_norm": 2.5930979251861572, | |
| "learning_rate": 3.37178947368421e-05, | |
| "loss": 1.5137, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.014042640426404265, | |
| "grad_norm": 2.3788273334503174, | |
| "learning_rate": 3.319105263157894e-05, | |
| "loss": 1.5968, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.014145141451414513, | |
| "grad_norm": 2.544044256210327, | |
| "learning_rate": 3.2664210526315786e-05, | |
| "loss": 1.8297, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.014247642476424764, | |
| "grad_norm": 1.9537235498428345, | |
| "learning_rate": 3.213736842105263e-05, | |
| "loss": 1.19, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.014350143501435014, | |
| "grad_norm": 2.7683722972869873, | |
| "learning_rate": 3.161052631578947e-05, | |
| "loss": 1.753, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.014452644526445265, | |
| "grad_norm": 2.9598679542541504, | |
| "learning_rate": 3.108368421052632e-05, | |
| "loss": 1.6878, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.014555145551455515, | |
| "grad_norm": 2.162545919418335, | |
| "learning_rate": 3.0556842105263156e-05, | |
| "loss": 1.0099, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.014657646576465764, | |
| "grad_norm": 2.538316249847412, | |
| "learning_rate": 3.0029999999999995e-05, | |
| "loss": 1.57, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.014760147601476014, | |
| "grad_norm": 2.4566807746887207, | |
| "learning_rate": 2.9503157894736838e-05, | |
| "loss": 1.2957, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.014862648626486265, | |
| "grad_norm": 2.836510181427002, | |
| "learning_rate": 2.8976315789473684e-05, | |
| "loss": 1.7585, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.014965149651496515, | |
| "grad_norm": 2.739304780960083, | |
| "learning_rate": 2.8449473684210523e-05, | |
| "loss": 1.0675, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.015067650676506766, | |
| "grad_norm": 3.267035484313965, | |
| "learning_rate": 2.7922631578947366e-05, | |
| "loss": 1.3077, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.015170151701517014, | |
| "grad_norm": 3.737492084503174, | |
| "learning_rate": 2.7395789473684212e-05, | |
| "loss": 1.5623, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.015272652726527265, | |
| "grad_norm": 3.9221222400665283, | |
| "learning_rate": 2.686894736842105e-05, | |
| "loss": 1.4467, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.015375153751537515, | |
| "grad_norm": 9.252695083618164, | |
| "learning_rate": 2.634210526315789e-05, | |
| "loss": 2.2941, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.015375153751537515, | |
| "eval_loss": 1.6176789999008179, | |
| "eval_runtime": 213.9116, | |
| "eval_samples_per_second": 19.204, | |
| "eval_steps_per_second": 4.801, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.015477654776547766, | |
| "grad_norm": 0.5801679491996765, | |
| "learning_rate": 2.5815263157894736e-05, | |
| "loss": 1.5924, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.015580155801558016, | |
| "grad_norm": 0.7414329051971436, | |
| "learning_rate": 2.528842105263158e-05, | |
| "loss": 1.8859, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.015682656826568265, | |
| "grad_norm": 0.7194873690605164, | |
| "learning_rate": 2.4761578947368418e-05, | |
| "loss": 1.7569, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.015785157851578517, | |
| "grad_norm": 0.7468145489692688, | |
| "learning_rate": 2.423473684210526e-05, | |
| "loss": 1.5206, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.015887658876588766, | |
| "grad_norm": 0.9180918335914612, | |
| "learning_rate": 2.3707894736842103e-05, | |
| "loss": 1.6067, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.015990159901599015, | |
| "grad_norm": 1.052054762840271, | |
| "learning_rate": 2.3181052631578946e-05, | |
| "loss": 1.5043, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.016092660926609267, | |
| "grad_norm": 0.8969054818153381, | |
| "learning_rate": 2.265421052631579e-05, | |
| "loss": 1.5507, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.016195161951619515, | |
| "grad_norm": 0.9579923152923584, | |
| "learning_rate": 2.212736842105263e-05, | |
| "loss": 1.5959, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.016297662976629768, | |
| "grad_norm": 1.0006402730941772, | |
| "learning_rate": 2.1600526315789474e-05, | |
| "loss": 1.6553, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.016400164001640016, | |
| "grad_norm": 1.090065598487854, | |
| "learning_rate": 2.1073684210526313e-05, | |
| "loss": 1.7441, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.016502665026650265, | |
| "grad_norm": 1.0853620767593384, | |
| "learning_rate": 2.0546842105263155e-05, | |
| "loss": 1.8341, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.016605166051660517, | |
| "grad_norm": 0.9956846833229065, | |
| "learning_rate": 2.002e-05, | |
| "loss": 1.7975, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.016707667076670766, | |
| "grad_norm": 1.2386821508407593, | |
| "learning_rate": 1.949315789473684e-05, | |
| "loss": 1.7063, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.016810168101681018, | |
| "grad_norm": 0.9561980962753296, | |
| "learning_rate": 1.8966315789473683e-05, | |
| "loss": 1.4661, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.016912669126691267, | |
| "grad_norm": 1.2328252792358398, | |
| "learning_rate": 1.8439473684210522e-05, | |
| "loss": 1.7836, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.017015170151701516, | |
| "grad_norm": 1.1177650690078735, | |
| "learning_rate": 1.791263157894737e-05, | |
| "loss": 1.7525, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.017117671176711768, | |
| "grad_norm": 1.012744426727295, | |
| "learning_rate": 1.738578947368421e-05, | |
| "loss": 1.5806, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.017220172201722016, | |
| "grad_norm": 1.1991872787475586, | |
| "learning_rate": 1.685894736842105e-05, | |
| "loss": 1.9107, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.01732267322673227, | |
| "grad_norm": 1.1519800424575806, | |
| "learning_rate": 1.6332105263157893e-05, | |
| "loss": 1.7668, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.017425174251742517, | |
| "grad_norm": 1.0659619569778442, | |
| "learning_rate": 1.5805263157894735e-05, | |
| "loss": 1.5949, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.017527675276752766, | |
| "grad_norm": 1.173683524131775, | |
| "learning_rate": 1.5278421052631578e-05, | |
| "loss": 1.4432, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.017630176301763018, | |
| "grad_norm": 1.2720458507537842, | |
| "learning_rate": 1.4751578947368419e-05, | |
| "loss": 1.6366, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.017732677326773267, | |
| "grad_norm": 1.412047266960144, | |
| "learning_rate": 1.4224736842105262e-05, | |
| "loss": 1.3457, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.01783517835178352, | |
| "grad_norm": 1.2512191534042358, | |
| "learning_rate": 1.3697894736842106e-05, | |
| "loss": 1.4677, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.017937679376793768, | |
| "grad_norm": 1.4642128944396973, | |
| "learning_rate": 1.3171052631578945e-05, | |
| "loss": 1.6189, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.018040180401804017, | |
| "grad_norm": 1.37020742893219, | |
| "learning_rate": 1.264421052631579e-05, | |
| "loss": 1.669, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.01814268142681427, | |
| "grad_norm": 1.3239974975585938, | |
| "learning_rate": 1.211736842105263e-05, | |
| "loss": 1.5652, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.018245182451824517, | |
| "grad_norm": 1.3516684770584106, | |
| "learning_rate": 1.1590526315789473e-05, | |
| "loss": 1.5116, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.01834768347683477, | |
| "grad_norm": 1.7171400785446167, | |
| "learning_rate": 1.1063684210526316e-05, | |
| "loss": 2.1205, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.01845018450184502, | |
| "grad_norm": 1.3495593070983887, | |
| "learning_rate": 1.0536842105263156e-05, | |
| "loss": 1.1325, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.018552685526855267, | |
| "grad_norm": 1.9317636489868164, | |
| "learning_rate": 1.001e-05, | |
| "loss": 1.5652, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.01865518655186552, | |
| "grad_norm": 2.422668695449829, | |
| "learning_rate": 9.483157894736842e-06, | |
| "loss": 1.5383, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.018757687576875768, | |
| "grad_norm": 2.1638026237487793, | |
| "learning_rate": 8.956315789473684e-06, | |
| "loss": 1.8706, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.01886018860188602, | |
| "grad_norm": 2.067598342895508, | |
| "learning_rate": 8.429473684210525e-06, | |
| "loss": 1.867, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.01896268962689627, | |
| "grad_norm": 2.594118118286133, | |
| "learning_rate": 7.902631578947368e-06, | |
| "loss": 1.5805, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.019065190651906518, | |
| "grad_norm": 2.4905567169189453, | |
| "learning_rate": 7.3757894736842095e-06, | |
| "loss": 1.0738, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.01916769167691677, | |
| "grad_norm": 2.5180282592773438, | |
| "learning_rate": 6.848947368421053e-06, | |
| "loss": 1.7173, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.01927019270192702, | |
| "grad_norm": 2.4175117015838623, | |
| "learning_rate": 6.322105263157895e-06, | |
| "loss": 1.5599, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.01937269372693727, | |
| "grad_norm": 2.6707475185394287, | |
| "learning_rate": 5.7952631578947365e-06, | |
| "loss": 1.3726, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.01947519475194752, | |
| "grad_norm": 3.7667629718780518, | |
| "learning_rate": 5.268421052631578e-06, | |
| "loss": 0.8815, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.019577695776957768, | |
| "grad_norm": 3.1217727661132812, | |
| "learning_rate": 4.741578947368421e-06, | |
| "loss": 1.5793, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.01968019680196802, | |
| "grad_norm": 3.0410215854644775, | |
| "learning_rate": 4.2147368421052626e-06, | |
| "loss": 1.5182, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.01978269782697827, | |
| "grad_norm": 2.907959222793579, | |
| "learning_rate": 3.6878947368421047e-06, | |
| "loss": 1.2825, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.01988519885198852, | |
| "grad_norm": 3.6714532375335693, | |
| "learning_rate": 3.1610526315789474e-06, | |
| "loss": 1.2199, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.01998769987699877, | |
| "grad_norm": 3.782942295074463, | |
| "learning_rate": 2.634210526315789e-06, | |
| "loss": 1.436, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.02009020090200902, | |
| "grad_norm": 4.33630895614624, | |
| "learning_rate": 2.1073684210526313e-06, | |
| "loss": 1.2589, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.02019270192701927, | |
| "grad_norm": 3.1009058952331543, | |
| "learning_rate": 1.5805263157894737e-06, | |
| "loss": 1.0389, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.02029520295202952, | |
| "grad_norm": 3.7881686687469482, | |
| "learning_rate": 1.0536842105263156e-06, | |
| "loss": 1.0837, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.02039770397703977, | |
| "grad_norm": 5.994607925415039, | |
| "learning_rate": 5.268421052631578e-07, | |
| "loss": 1.7627, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.02050020500205002, | |
| "grad_norm": 6.927132606506348, | |
| "learning_rate": 0.0, | |
| "loss": 1.7932, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.02050020500205002, | |
| "eval_loss": 1.6074018478393555, | |
| "eval_runtime": 214.0108, | |
| "eval_samples_per_second": 19.195, | |
| "eval_steps_per_second": 4.799, | |
| "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": 7.127217035280384e+16, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |