| { | |
| "best_metric": 0.8947485685348511, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.11860637509266123, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0005930318754633062, | |
| "grad_norm": 0.9128352403640747, | |
| "learning_rate": 1e-05, | |
| "loss": 1.3096, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0005930318754633062, | |
| "eval_loss": 1.4674732685089111, | |
| "eval_runtime": 54.584, | |
| "eval_samples_per_second": 52.03, | |
| "eval_steps_per_second": 13.007, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0011860637509266123, | |
| "grad_norm": 0.917138397693634, | |
| "learning_rate": 2e-05, | |
| "loss": 1.6473, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0017790956263899185, | |
| "grad_norm": 0.8538322448730469, | |
| "learning_rate": 3e-05, | |
| "loss": 1.6559, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.0023721275018532247, | |
| "grad_norm": 0.7117102742195129, | |
| "learning_rate": 4e-05, | |
| "loss": 1.7047, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.0029651593773165306, | |
| "grad_norm": 0.7061640024185181, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5857, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.003558191252779837, | |
| "grad_norm": 0.7815567255020142, | |
| "learning_rate": 6e-05, | |
| "loss": 1.5563, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.004151223128243143, | |
| "grad_norm": 0.8941373825073242, | |
| "learning_rate": 7e-05, | |
| "loss": 1.5596, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.004744255003706449, | |
| "grad_norm": 0.7239966988563538, | |
| "learning_rate": 8e-05, | |
| "loss": 1.4816, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.005337286879169755, | |
| "grad_norm": 0.7085259556770325, | |
| "learning_rate": 9e-05, | |
| "loss": 1.5237, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.005930318754633061, | |
| "grad_norm": 0.6864994168281555, | |
| "learning_rate": 0.0001, | |
| "loss": 1.4989, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.006523350630096367, | |
| "grad_norm": 0.6545014381408691, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 1.509, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.007116382505559674, | |
| "grad_norm": 0.700894296169281, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 1.3554, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.00770941438102298, | |
| "grad_norm": 0.6438685059547424, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 1.3176, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.008302446256486287, | |
| "grad_norm": 0.6856859922409058, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 1.3627, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.008895478131949592, | |
| "grad_norm": 0.6096616983413696, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 1.3077, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.009488510007412899, | |
| "grad_norm": 0.6056028008460999, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 1.3192, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.010081541882876204, | |
| "grad_norm": 0.674043595790863, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 1.3199, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.01067457375833951, | |
| "grad_norm": 0.6460352540016174, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 1.275, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.011267605633802818, | |
| "grad_norm": 0.5855716466903687, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 1.1831, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.011860637509266123, | |
| "grad_norm": 0.6667662262916565, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 1.1451, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.01245366938472943, | |
| "grad_norm": 0.6556470394134521, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 1.1135, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.013046701260192735, | |
| "grad_norm": 0.8012140393257141, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 1.2346, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.013639733135656041, | |
| "grad_norm": 0.6600927710533142, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 1.1783, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.014232765011119348, | |
| "grad_norm": 0.6758884191513062, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 1.0886, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.014825796886582653, | |
| "grad_norm": 0.6599623560905457, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 1.1174, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.01541882876204596, | |
| "grad_norm": 0.6617851257324219, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 1.0537, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.016011860637509267, | |
| "grad_norm": 0.634557843208313, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 0.9785, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.016604892512972574, | |
| "grad_norm": 0.701481819152832, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 1.0195, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.017197924388435877, | |
| "grad_norm": 0.6694827079772949, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 1.0036, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.017790956263899184, | |
| "grad_norm": 0.6416652798652649, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 1.0097, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.01838398813936249, | |
| "grad_norm": 0.7445120811462402, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 0.9926, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.018977020014825798, | |
| "grad_norm": 0.6441786885261536, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 0.9812, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.019570051890289104, | |
| "grad_norm": 0.691388726234436, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 1.0629, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.020163083765752408, | |
| "grad_norm": 0.7649052739143372, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 0.9518, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.020756115641215715, | |
| "grad_norm": 1.0184032917022705, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 0.8317, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.02134914751667902, | |
| "grad_norm": 0.8641113042831421, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 0.8625, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.021942179392142328, | |
| "grad_norm": 0.950751006603241, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 0.9415, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.022535211267605635, | |
| "grad_norm": 0.8036394715309143, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 0.9917, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.02312824314306894, | |
| "grad_norm": 0.7395977973937988, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 0.9532, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.023721275018532245, | |
| "grad_norm": 0.8059082627296448, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 0.8402, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.024314306893995552, | |
| "grad_norm": 0.684325098991394, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 0.8513, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.02490733876945886, | |
| "grad_norm": 0.8658984899520874, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 0.9661, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.025500370644922166, | |
| "grad_norm": 1.7691848278045654, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 0.7337, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.02609340252038547, | |
| "grad_norm": 0.8865183591842651, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 0.6731, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.026686434395848776, | |
| "grad_norm": 0.7921429872512817, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 0.8241, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.027279466271312083, | |
| "grad_norm": 0.7519988417625427, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 0.8407, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.02787249814677539, | |
| "grad_norm": 1.0038756132125854, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 0.7287, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.028465530022238696, | |
| "grad_norm": 1.711969256401062, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 0.519, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.029058561897702003, | |
| "grad_norm": 1.0256000757217407, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 0.8058, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.029651593773165306, | |
| "grad_norm": 1.2814997434616089, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 0.9202, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.029651593773165306, | |
| "eval_loss": 1.0709905624389648, | |
| "eval_runtime": 54.7426, | |
| "eval_samples_per_second": 51.879, | |
| "eval_steps_per_second": 12.97, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.030244625648628613, | |
| "grad_norm": 1.2627829313278198, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 1.368, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.03083765752409192, | |
| "grad_norm": 1.0638519525527954, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 1.5607, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.03143068939955523, | |
| "grad_norm": 0.748832643032074, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 1.6571, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.032023721275018534, | |
| "grad_norm": 0.664341390132904, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 1.6136, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.03261675315048184, | |
| "grad_norm": 0.553286075592041, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 1.4656, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.03320978502594515, | |
| "grad_norm": 0.580268919467926, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 1.3559, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.03380281690140845, | |
| "grad_norm": 0.6548421382904053, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 1.3577, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.034395848776871754, | |
| "grad_norm": 0.5364956855773926, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 1.2899, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.03498888065233506, | |
| "grad_norm": 0.5661771893501282, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 1.3188, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.03558191252779837, | |
| "grad_norm": 0.5021414160728455, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 1.2907, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.036174944403261675, | |
| "grad_norm": 0.5147365927696228, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 1.2875, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.03676797627872498, | |
| "grad_norm": 0.5012564659118652, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 1.3046, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.03736100815418829, | |
| "grad_norm": 0.5347183346748352, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 1.317, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.037954040029651595, | |
| "grad_norm": 0.4898674488067627, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 1.2163, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.0385470719051149, | |
| "grad_norm": 0.4707097113132477, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 1.2045, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.03914010378057821, | |
| "grad_norm": 0.4987703859806061, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 1.187, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.039733135656041516, | |
| "grad_norm": 0.48926910758018494, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 1.1191, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.040326167531504815, | |
| "grad_norm": 0.4914247393608093, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 1.0986, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.04091919940696812, | |
| "grad_norm": 0.461151659488678, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 1.0333, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.04151223128243143, | |
| "grad_norm": 0.503083348274231, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 1.0957, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.042105263157894736, | |
| "grad_norm": 0.5450732111930847, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 0.928, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.04269829503335804, | |
| "grad_norm": 0.47438186407089233, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 0.9875, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.04329132690882135, | |
| "grad_norm": 0.5245692729949951, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 1.0826, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.043884358784284656, | |
| "grad_norm": 0.6992484927177429, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 0.9841, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.04447739065974796, | |
| "grad_norm": 0.5221351981163025, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 0.8991, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.04507042253521127, | |
| "grad_norm": 0.4883466362953186, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 0.8717, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.04566345441067458, | |
| "grad_norm": 0.5207772850990295, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 0.9564, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.04625648628613788, | |
| "grad_norm": 0.5393484234809875, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 0.8084, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.046849518161601184, | |
| "grad_norm": 0.5158237814903259, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 0.8711, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.04744255003706449, | |
| "grad_norm": 0.5125463604927063, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 0.8791, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.0480355819125278, | |
| "grad_norm": 0.5904987454414368, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 0.7699, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.048628613787991104, | |
| "grad_norm": 0.5575411915779114, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 0.8107, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.04922164566345441, | |
| "grad_norm": 0.5736557245254517, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 0.9037, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.04981467753891772, | |
| "grad_norm": 0.6238594651222229, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 0.8845, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.050407709414381024, | |
| "grad_norm": 0.6680272221565247, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 0.7974, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.05100074128984433, | |
| "grad_norm": 1.050567626953125, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 0.6855, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.05159377316530764, | |
| "grad_norm": 0.6209374666213989, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 0.961, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.05218680504077094, | |
| "grad_norm": 0.6250402927398682, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 0.8688, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.052779836916234245, | |
| "grad_norm": 0.5850189924240112, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 0.8454, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.05337286879169755, | |
| "grad_norm": 0.5783194899559021, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 0.7482, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.05396590066716086, | |
| "grad_norm": 0.6094998121261597, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 0.8129, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.054558932542624165, | |
| "grad_norm": 0.9608752727508545, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 0.6524, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.05515196441808747, | |
| "grad_norm": 0.7261383533477783, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 0.7844, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.05574499629355078, | |
| "grad_norm": 0.6840940713882446, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 0.7562, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.056338028169014086, | |
| "grad_norm": 0.6493006348609924, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 0.8178, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.05693106004447739, | |
| "grad_norm": 0.665821373462677, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 0.7347, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.0575240919199407, | |
| "grad_norm": 1.3736786842346191, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 0.2814, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.058117123795404006, | |
| "grad_norm": 1.1150555610656738, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 0.2965, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.058710155670867306, | |
| "grad_norm": 0.791911780834198, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 0.6964, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.05930318754633061, | |
| "grad_norm": 0.8627227544784546, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 0.6942, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.05930318754633061, | |
| "eval_loss": 0.9763190746307373, | |
| "eval_runtime": 54.8056, | |
| "eval_samples_per_second": 51.819, | |
| "eval_steps_per_second": 12.955, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.05989621942179392, | |
| "grad_norm": 0.8971495032310486, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 1.2981, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.06048925129725723, | |
| "grad_norm": 0.788299024105072, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 1.4664, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.06108228317272053, | |
| "grad_norm": 0.6883640289306641, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 1.6317, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.06167531504818384, | |
| "grad_norm": 0.5831127166748047, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 1.6468, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.06226834692364715, | |
| "grad_norm": 0.5455831289291382, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5163, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.06286137879911045, | |
| "grad_norm": 0.46923744678497314, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 1.3974, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.06345441067457376, | |
| "grad_norm": 0.5191442370414734, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 1.3646, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.06404744255003707, | |
| "grad_norm": 0.5451369881629944, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 1.3397, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.06464047442550037, | |
| "grad_norm": 0.4729631841182709, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 1.2549, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.06523350630096368, | |
| "grad_norm": 0.58608078956604, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 1.2826, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.06582653817642699, | |
| "grad_norm": 0.5048196315765381, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 1.3105, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.0664195700518903, | |
| "grad_norm": 0.47246071696281433, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 1.1458, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.0670126019273536, | |
| "grad_norm": 0.4854651391506195, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 1.0838, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.0676056338028169, | |
| "grad_norm": 0.4831504225730896, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 1.0581, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.0681986656782802, | |
| "grad_norm": 0.4957878887653351, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 1.1557, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.06879169755374351, | |
| "grad_norm": 0.5005252957344055, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 1.1404, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.06938472942920682, | |
| "grad_norm": 0.47843456268310547, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 1.0712, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.06997776130467012, | |
| "grad_norm": 0.49580538272857666, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 1.0859, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.07057079318013343, | |
| "grad_norm": 0.4846254289150238, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 1.0659, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.07116382505559674, | |
| "grad_norm": 0.5240991115570068, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 1.0308, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.07175685693106004, | |
| "grad_norm": 0.4797056317329407, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 1.0719, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.07234988880652335, | |
| "grad_norm": 0.5224462151527405, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 0.957, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.07294292068198666, | |
| "grad_norm": 0.5027078986167908, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 0.9855, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.07353595255744996, | |
| "grad_norm": 0.5758591890335083, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 0.891, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.07412898443291327, | |
| "grad_norm": 0.6827547550201416, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 0.8659, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.07472201630837658, | |
| "grad_norm": 0.5156536102294922, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 0.8322, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.07531504818383988, | |
| "grad_norm": 0.5152995586395264, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 0.7989, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.07590808005930319, | |
| "grad_norm": 0.5190814733505249, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 0.8023, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.0765011119347665, | |
| "grad_norm": 0.5331418514251709, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 0.7931, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.0770941438102298, | |
| "grad_norm": 0.625654399394989, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 0.8473, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.07768717568569311, | |
| "grad_norm": 0.5979682803153992, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 0.7122, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.07828020756115642, | |
| "grad_norm": 0.5621092915534973, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 0.7681, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.07887323943661972, | |
| "grad_norm": 0.6109548211097717, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 0.7551, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.07946627131208303, | |
| "grad_norm": 0.5191078782081604, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 0.7578, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.08005930318754632, | |
| "grad_norm": 0.5392255783081055, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 0.8482, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.08065233506300963, | |
| "grad_norm": 0.7676748037338257, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 0.5408, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.08124536693847294, | |
| "grad_norm": 0.6903252601623535, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 0.5913, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.08183839881393624, | |
| "grad_norm": 0.5773659348487854, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 0.817, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.08243143068939955, | |
| "grad_norm": 0.5975673794746399, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 0.8512, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.08302446256486286, | |
| "grad_norm": 0.551443874835968, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 0.7687, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.08361749444032616, | |
| "grad_norm": 0.8684813976287842, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 0.686, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.08421052631578947, | |
| "grad_norm": 0.5953558087348938, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 0.7231, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.08480355819125278, | |
| "grad_norm": 1.1803919076919556, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 0.3837, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.08539659006671609, | |
| "grad_norm": 0.5375893115997314, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 0.5268, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.08598962194217939, | |
| "grad_norm": 0.6543303728103638, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 0.6599, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.0865826538176427, | |
| "grad_norm": 0.5823112726211548, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 0.6481, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.087175685693106, | |
| "grad_norm": 0.7933789491653442, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 0.7557, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.08776871756856931, | |
| "grad_norm": 2.7837235927581787, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 0.2413, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.08836174944403262, | |
| "grad_norm": 0.9221758842468262, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 0.5467, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.08895478131949593, | |
| "grad_norm": 0.7956274151802063, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 0.6351, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.08895478131949593, | |
| "eval_loss": 0.904806911945343, | |
| "eval_runtime": 54.6116, | |
| "eval_samples_per_second": 52.004, | |
| "eval_steps_per_second": 13.001, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.08954781319495923, | |
| "grad_norm": 0.49180299043655396, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 1.2796, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.09014084507042254, | |
| "grad_norm": 0.5233712196350098, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 1.4463, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.09073387694588585, | |
| "grad_norm": 0.5484112501144409, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 1.5206, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.09132690882134915, | |
| "grad_norm": 0.6222096085548401, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 1.5653, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.09191994069681246, | |
| "grad_norm": 0.5047340393066406, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 1.2886, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.09251297257227575, | |
| "grad_norm": 0.5084788799285889, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 1.3614, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.09310600444773906, | |
| "grad_norm": 0.49157023429870605, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 1.2741, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.09369903632320237, | |
| "grad_norm": 0.45855769515037537, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 1.2554, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.09429206819866567, | |
| "grad_norm": 0.5020678043365479, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 1.169, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.09488510007412898, | |
| "grad_norm": 0.46542999148368835, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 1.2578, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.09547813194959229, | |
| "grad_norm": 0.5047861337661743, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 1.1876, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.0960711638250556, | |
| "grad_norm": 0.5085510015487671, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 1.053, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.0966641957005189, | |
| "grad_norm": 0.5247685313224792, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 1.1907, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.09725722757598221, | |
| "grad_norm": 0.4579811096191406, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 1.1396, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.09785025945144551, | |
| "grad_norm": 0.46387508511543274, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 1.1433, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.09844329132690882, | |
| "grad_norm": 0.46310198307037354, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 1.0559, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.09903632320237213, | |
| "grad_norm": 0.5656064748764038, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 0.984, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.09962935507783544, | |
| "grad_norm": 0.4853728711605072, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 1.1338, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.10022238695329874, | |
| "grad_norm": 0.5021612644195557, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 1.0033, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.10081541882876205, | |
| "grad_norm": 0.479026198387146, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 0.9687, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.10140845070422536, | |
| "grad_norm": 0.5409302711486816, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 0.9316, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.10200148257968866, | |
| "grad_norm": 0.48674267530441284, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 0.9845, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.10259451445515197, | |
| "grad_norm": 0.5077137351036072, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 0.88, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.10318754633061528, | |
| "grad_norm": 0.503619372844696, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 0.8173, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.10378057820607858, | |
| "grad_norm": 0.5211278200149536, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 0.8957, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.10437361008154188, | |
| "grad_norm": 0.5747191309928894, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 0.8104, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.10496664195700518, | |
| "grad_norm": 0.5396358370780945, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 0.8349, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.10555967383246849, | |
| "grad_norm": 0.5236896872520447, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 0.6931, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.1061527057079318, | |
| "grad_norm": 0.5552003979682922, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 0.7589, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.1067457375833951, | |
| "grad_norm": 0.5147039294242859, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 0.848, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.10733876945885841, | |
| "grad_norm": 0.5422276258468628, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 0.8282, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.10793180133432172, | |
| "grad_norm": 0.47938427329063416, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 0.7093, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.10852483320978502, | |
| "grad_norm": 0.5359724164009094, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 0.861, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.10911786508524833, | |
| "grad_norm": 0.5440419912338257, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 0.6547, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.10971089696071164, | |
| "grad_norm": 0.6765339970588684, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 0.5238, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.11030392883617494, | |
| "grad_norm": 0.6517006158828735, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 0.5315, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.11089696071163825, | |
| "grad_norm": 0.5708439350128174, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 0.7547, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.11148999258710156, | |
| "grad_norm": 0.6777458786964417, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 0.6551, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.11208302446256486, | |
| "grad_norm": 0.5782615542411804, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 0.7399, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.11267605633802817, | |
| "grad_norm": 0.5603607296943665, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 0.8177, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.11326908821349148, | |
| "grad_norm": 0.5594826340675354, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 0.7413, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.11386212008895479, | |
| "grad_norm": 0.7118792533874512, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 0.5511, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.11445515196441809, | |
| "grad_norm": 1.1664906740188599, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 0.3968, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.1150481838398814, | |
| "grad_norm": 0.5703531503677368, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 0.743, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.1156412157153447, | |
| "grad_norm": 0.6180101633071899, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 0.7577, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.11623424759080801, | |
| "grad_norm": 0.6108247637748718, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 0.7962, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.1168272794662713, | |
| "grad_norm": 0.5928218960762024, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 0.4661, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.11742031134173461, | |
| "grad_norm": 0.7264745235443115, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 0.2437, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.11801334321719792, | |
| "grad_norm": 0.6450815200805664, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 0.6115, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.11860637509266123, | |
| "grad_norm": 0.8737947940826416, | |
| "learning_rate": 0.0, | |
| "loss": 0.6112, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.11860637509266123, | |
| "eval_loss": 0.8947485685348511, | |
| "eval_runtime": 54.972, | |
| "eval_samples_per_second": 51.663, | |
| "eval_steps_per_second": 12.916, | |
| "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": 1.6944697731710976e+16, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |