| { | |
| "best_metric": 0.29025572538375854, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.25698682942499196, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.00128493414712496, | |
| "grad_norm": 0.24772338569164276, | |
| "learning_rate": 1e-05, | |
| "loss": 0.2298, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.00128493414712496, | |
| "eval_loss": 0.6178669929504395, | |
| "eval_runtime": 27.2325, | |
| "eval_samples_per_second": 48.141, | |
| "eval_steps_per_second": 12.044, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.00256986829424992, | |
| "grad_norm": 0.37560346722602844, | |
| "learning_rate": 2e-05, | |
| "loss": 0.3782, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0038548024413748794, | |
| "grad_norm": 0.3819539248943329, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2852, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.00513973658849984, | |
| "grad_norm": 0.4002714157104492, | |
| "learning_rate": 4e-05, | |
| "loss": 0.317, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.006424670735624799, | |
| "grad_norm": 0.4359646439552307, | |
| "learning_rate": 5e-05, | |
| "loss": 0.344, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.007709604882749759, | |
| "grad_norm": 0.46338069438934326, | |
| "learning_rate": 6e-05, | |
| "loss": 0.3828, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.008994539029874718, | |
| "grad_norm": 0.43742719292640686, | |
| "learning_rate": 7e-05, | |
| "loss": 0.4176, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.01027947317699968, | |
| "grad_norm": 0.44323965907096863, | |
| "learning_rate": 8e-05, | |
| "loss": 0.4309, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.011564407324124639, | |
| "grad_norm": 0.4445992708206177, | |
| "learning_rate": 9e-05, | |
| "loss": 0.3407, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.012849341471249599, | |
| "grad_norm": 0.4301189184188843, | |
| "learning_rate": 0.0001, | |
| "loss": 0.3721, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.014134275618374558, | |
| "grad_norm": 0.5747378468513489, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 0.4371, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.015419209765499518, | |
| "grad_norm": 0.43010470271110535, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 0.3008, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.016704143912624477, | |
| "grad_norm": 0.5371198058128357, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 0.3961, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.017989078059749437, | |
| "grad_norm": 0.45776090025901794, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 0.3741, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.019274012206874396, | |
| "grad_norm": 0.39459675550460815, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 0.3035, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.02055894635399936, | |
| "grad_norm": 0.441237211227417, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 0.2919, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.02184388050112432, | |
| "grad_norm": 0.44288119673728943, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 0.3969, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.023128814648249278, | |
| "grad_norm": 0.3865252435207367, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 0.2969, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.024413748795374238, | |
| "grad_norm": 0.41208988428115845, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 0.2706, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.025698682942499197, | |
| "grad_norm": 0.44389140605926514, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 0.2946, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.026983617089624157, | |
| "grad_norm": 0.5243967175483704, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 0.444, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.028268551236749116, | |
| "grad_norm": 0.494181752204895, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 0.3479, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.029553485383874076, | |
| "grad_norm": 0.41096654534339905, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 0.2939, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.030838419530999035, | |
| "grad_norm": 0.4182928502559662, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 0.2773, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.032123353678123995, | |
| "grad_norm": 0.4379871189594269, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 0.2607, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.033408287825248954, | |
| "grad_norm": 0.4218382239341736, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 0.3267, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.034693221972373914, | |
| "grad_norm": 0.5452669262886047, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 0.3846, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.03597815611949887, | |
| "grad_norm": 0.45285865664482117, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 0.2792, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.03726309026662383, | |
| "grad_norm": 0.4787646234035492, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 0.3211, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.03854802441374879, | |
| "grad_norm": 0.4644937217235565, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 0.3649, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.03983295856087375, | |
| "grad_norm": 0.5910214781761169, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 0.3457, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.04111789270799872, | |
| "grad_norm": 0.430152028799057, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 0.2264, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.04240282685512368, | |
| "grad_norm": 0.6709097623825073, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 0.45, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.04368776100224864, | |
| "grad_norm": 0.6646703481674194, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 0.4226, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.0449726951493736, | |
| "grad_norm": 0.5327663421630859, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 0.3341, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.046257629296498556, | |
| "grad_norm": 0.570952832698822, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 0.2926, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.047542563443623516, | |
| "grad_norm": 0.5564446449279785, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 0.3393, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.048827497590748475, | |
| "grad_norm": 0.5784032940864563, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 0.294, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.050112431737873435, | |
| "grad_norm": 0.5784353613853455, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 0.3646, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.051397365884998394, | |
| "grad_norm": 0.629645824432373, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 0.4474, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.052682300032123354, | |
| "grad_norm": 0.7099590301513672, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 0.3404, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.05396723417924831, | |
| "grad_norm": 0.6699565052986145, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 0.4504, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.05525216832637327, | |
| "grad_norm": 0.6679425239562988, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 0.4113, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.05653710247349823, | |
| "grad_norm": 0.6381205320358276, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 0.3744, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.05782203662062319, | |
| "grad_norm": 0.7222647070884705, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 0.4072, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.05910697076774815, | |
| "grad_norm": 0.6967664957046509, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 0.3006, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.06039190491487311, | |
| "grad_norm": 0.7783026695251465, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 0.3466, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.06167683906199807, | |
| "grad_norm": 1.0430123805999756, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 0.4422, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.06296177320912304, | |
| "grad_norm": 1.042852759361267, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 0.4402, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.06424670735624799, | |
| "grad_norm": 1.4085546731948853, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 0.5011, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.06424670735624799, | |
| "eval_loss": 0.34552425146102905, | |
| "eval_runtime": 27.3814, | |
| "eval_samples_per_second": 47.879, | |
| "eval_steps_per_second": 11.979, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.06553164150337296, | |
| "grad_norm": 0.4816097021102905, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 0.2166, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.06681657565049791, | |
| "grad_norm": 0.3567050099372864, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 0.2362, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.06810150979762288, | |
| "grad_norm": 0.3814481496810913, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 0.3255, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.06938644394474783, | |
| "grad_norm": 0.34487542510032654, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 0.2829, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.0706713780918728, | |
| "grad_norm": 0.30658677220344543, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 0.2159, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.07195631223899775, | |
| "grad_norm": 0.3005298376083374, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 0.375, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.07324124638612271, | |
| "grad_norm": 0.3102109134197235, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 0.2701, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.07452618053324767, | |
| "grad_norm": 0.2959745228290558, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 0.2854, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.07581111468037263, | |
| "grad_norm": 0.32974228262901306, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 0.3145, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.07709604882749758, | |
| "grad_norm": 0.3541993200778961, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 0.2751, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.07838098297462255, | |
| "grad_norm": 0.3677872121334076, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 0.3665, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.0796659171217475, | |
| "grad_norm": 0.30598706007003784, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 0.2788, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.08095085126887247, | |
| "grad_norm": 0.3018777072429657, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 0.2859, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.08223578541599744, | |
| "grad_norm": 0.3187599778175354, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 0.3103, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.08352071956312239, | |
| "grad_norm": 0.3353744447231293, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 0.3096, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.08480565371024736, | |
| "grad_norm": 0.34449559450149536, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 0.378, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.08609058785737231, | |
| "grad_norm": 0.32587167620658875, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 0.2551, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.08737552200449727, | |
| "grad_norm": 0.3395215570926666, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 0.3138, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.08866045615162223, | |
| "grad_norm": 0.3422412872314453, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 0.3302, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.0899453902987472, | |
| "grad_norm": 0.300110787153244, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 0.1943, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.09123032444587215, | |
| "grad_norm": 0.32049864530563354, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 0.2709, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.09251525859299711, | |
| "grad_norm": 0.26257699728012085, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 0.2068, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.09380019274012207, | |
| "grad_norm": 0.34572461247444153, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 0.3182, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.09508512688724703, | |
| "grad_norm": 0.32622313499450684, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 0.2391, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.09637006103437198, | |
| "grad_norm": 0.39312589168548584, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 0.3182, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.09765499518149695, | |
| "grad_norm": 0.3653692603111267, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 0.2729, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.0989399293286219, | |
| "grad_norm": 0.34862855076789856, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 0.2486, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.10022486347574687, | |
| "grad_norm": 0.4172123372554779, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 0.315, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.10150979762287182, | |
| "grad_norm": 0.4427936375141144, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 0.4112, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.10279473176999679, | |
| "grad_norm": 0.391338050365448, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 0.3037, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.10407966591712174, | |
| "grad_norm": 0.3853769302368164, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 0.2344, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.10536460006424671, | |
| "grad_norm": 0.5132508873939514, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 0.3576, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.10664953421137166, | |
| "grad_norm": 0.4463823437690735, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 0.2919, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.10793446835849663, | |
| "grad_norm": 0.5203205943107605, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 0.4128, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.1092194025056216, | |
| "grad_norm": 0.4778902232646942, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 0.3301, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.11050433665274655, | |
| "grad_norm": 0.5111919045448303, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 0.3669, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.11178927079987151, | |
| "grad_norm": 0.49184751510620117, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 0.2851, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.11307420494699646, | |
| "grad_norm": 0.482742041349411, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 0.3143, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.11435913909412143, | |
| "grad_norm": 0.4509274661540985, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 0.2703, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.11564407324124638, | |
| "grad_norm": 0.4560715854167938, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 0.2802, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.11692900738837135, | |
| "grad_norm": 0.6458951830863953, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 0.4162, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.1182139415354963, | |
| "grad_norm": 0.6800999641418457, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 0.4535, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.11949887568262127, | |
| "grad_norm": 0.5842026472091675, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 0.3058, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.12078380982974622, | |
| "grad_norm": 0.5768910050392151, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 0.3561, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.12206874397687119, | |
| "grad_norm": 0.45313623547554016, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 0.2558, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.12335367812399614, | |
| "grad_norm": 0.5546783804893494, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 0.2944, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.12463861227112111, | |
| "grad_norm": 0.5335418581962585, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 0.3012, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.12592354641824607, | |
| "grad_norm": 0.5577960014343262, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 0.3267, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.127208480565371, | |
| "grad_norm": 0.6567111611366272, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 0.342, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.12849341471249598, | |
| "grad_norm": 1.1290212869644165, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 0.5319, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.12849341471249598, | |
| "eval_loss": 0.3060326874256134, | |
| "eval_runtime": 27.3696, | |
| "eval_samples_per_second": 47.9, | |
| "eval_steps_per_second": 11.984, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.12977834885962095, | |
| "grad_norm": 0.17409755289554596, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 0.2513, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.1310632830067459, | |
| "grad_norm": 0.22197651863098145, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 0.2484, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.13234821715387085, | |
| "grad_norm": 0.2127583622932434, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 0.2062, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.13363315130099582, | |
| "grad_norm": 0.26016566157341003, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 0.272, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.13491808544812078, | |
| "grad_norm": 0.302839457988739, | |
| "learning_rate": 5e-05, | |
| "loss": 0.3538, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.13620301959524575, | |
| "grad_norm": 0.29175865650177, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 0.3136, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.13748795374237072, | |
| "grad_norm": 0.3259955942630768, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 0.3537, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.13877288788949566, | |
| "grad_norm": 0.27290788292884827, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 0.2663, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.14005782203662062, | |
| "grad_norm": 0.3092767298221588, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 0.32, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.1413427561837456, | |
| "grad_norm": 0.2668243646621704, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 0.2338, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.14262769033087055, | |
| "grad_norm": 0.30612391233444214, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 0.3283, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.1439126244779955, | |
| "grad_norm": 0.3122018873691559, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 0.3281, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.14519755862512046, | |
| "grad_norm": 0.3077618479728699, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 0.2676, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.14648249277224543, | |
| "grad_norm": 0.27045148611068726, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 0.2056, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.1477674269193704, | |
| "grad_norm": 0.30312228202819824, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 0.25, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.14905236106649533, | |
| "grad_norm": 0.3513820469379425, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 0.3131, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.1503372952136203, | |
| "grad_norm": 0.350780725479126, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 0.3772, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.15162222936074526, | |
| "grad_norm": 0.3204064667224884, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 0.2903, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.15290716350787023, | |
| "grad_norm": 0.34363776445388794, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 0.334, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.15419209765499517, | |
| "grad_norm": 0.25390133261680603, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 0.19, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.15547703180212014, | |
| "grad_norm": 0.34550103545188904, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 0.3181, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.1567619659492451, | |
| "grad_norm": 0.32693663239479065, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 0.2553, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.15804690009637007, | |
| "grad_norm": 0.334131121635437, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 0.2929, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.159331834243495, | |
| "grad_norm": 0.3320024013519287, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 0.26, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.16061676839061997, | |
| "grad_norm": 0.38280177116394043, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 0.3032, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.16190170253774494, | |
| "grad_norm": 0.378211110830307, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 0.2936, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.1631866366848699, | |
| "grad_norm": 0.2850527763366699, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 0.1929, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.16447157083199487, | |
| "grad_norm": 0.3229641914367676, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 0.2422, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.1657565049791198, | |
| "grad_norm": 0.367938369512558, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 0.2275, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.16704143912624478, | |
| "grad_norm": 0.38579079508781433, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 0.2995, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.16832637327336974, | |
| "grad_norm": 0.41520774364471436, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 0.2944, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.1696113074204947, | |
| "grad_norm": 0.3886098265647888, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 0.2747, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.17089624156761965, | |
| "grad_norm": 0.38928064703941345, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 0.2648, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.17218117571474462, | |
| "grad_norm": 0.5069098472595215, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 0.3584, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.17346610986186958, | |
| "grad_norm": 0.44694310426712036, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 0.2786, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.17475104400899455, | |
| "grad_norm": 0.4142614006996155, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 0.3115, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.1760359781561195, | |
| "grad_norm": 0.5080338716506958, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 0.3324, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.17732091230324445, | |
| "grad_norm": 0.48226046562194824, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 0.3045, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.17860584645036942, | |
| "grad_norm": 0.4620155096054077, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 0.2643, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.1798907805974944, | |
| "grad_norm": 0.5789238214492798, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 0.2967, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.18117571474461933, | |
| "grad_norm": 0.5143536925315857, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 0.336, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.1824606488917443, | |
| "grad_norm": 0.5276119112968445, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 0.3374, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.18374558303886926, | |
| "grad_norm": 0.5345407128334045, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 0.3394, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.18503051718599423, | |
| "grad_norm": 0.5086403489112854, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 0.3448, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.18631545133311916, | |
| "grad_norm": 0.5218998789787292, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 0.3096, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.18760038548024413, | |
| "grad_norm": 0.4385205805301666, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 0.2426, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.1888853196273691, | |
| "grad_norm": 0.4644688367843628, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 0.2295, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.19017025377449406, | |
| "grad_norm": 0.8642386198043823, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 0.4106, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.19145518792161903, | |
| "grad_norm": 0.7614267468452454, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 0.4495, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.19274012206874397, | |
| "grad_norm": 0.9899398684501648, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 0.4741, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.19274012206874397, | |
| "eval_loss": 0.2946407198905945, | |
| "eval_runtime": 27.4027, | |
| "eval_samples_per_second": 47.842, | |
| "eval_steps_per_second": 11.97, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.19402505621586894, | |
| "grad_norm": 0.15465191006660461, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 0.2205, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.1953099903629939, | |
| "grad_norm": 0.1653699278831482, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 0.1757, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.19659492451011887, | |
| "grad_norm": 0.19132116436958313, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 0.2579, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.1978798586572438, | |
| "grad_norm": 0.21742530167102814, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 0.2048, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.19916479280436877, | |
| "grad_norm": 0.23674198985099792, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 0.2792, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.20044972695149374, | |
| "grad_norm": 0.22005398571491241, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 0.2516, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.2017346610986187, | |
| "grad_norm": 0.23764051496982574, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 0.2899, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.20301959524574364, | |
| "grad_norm": 0.26167425513267517, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 0.3056, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.2043045293928686, | |
| "grad_norm": 0.2839110493659973, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 0.3365, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.20558946353999358, | |
| "grad_norm": 0.26760736107826233, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 0.2827, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.20687439768711854, | |
| "grad_norm": 0.2700652778148651, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 0.2887, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.20815933183424348, | |
| "grad_norm": 0.24393315613269806, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 0.2084, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.20944426598136845, | |
| "grad_norm": 0.2970396876335144, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 0.2717, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.21072920012849342, | |
| "grad_norm": 0.2823368012905121, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 0.2763, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.21201413427561838, | |
| "grad_norm": 0.34385547041893005, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 0.3361, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.21329906842274332, | |
| "grad_norm": 0.29794496297836304, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 0.2337, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.2145840025698683, | |
| "grad_norm": 0.3338766396045685, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 0.2916, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.21586893671699325, | |
| "grad_norm": 0.3520536720752716, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 0.3273, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.21715387086411822, | |
| "grad_norm": 0.3323323428630829, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 0.289, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.2184388050112432, | |
| "grad_norm": 0.33143672347068787, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 0.3219, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.21972373915836813, | |
| "grad_norm": 0.3776671290397644, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 0.3678, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.2210086733054931, | |
| "grad_norm": 0.3340603709220886, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 0.2966, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.22229360745261806, | |
| "grad_norm": 0.38343366980552673, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 0.3105, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.22357854159974302, | |
| "grad_norm": 0.36424487829208374, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 0.2137, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.22486347574686796, | |
| "grad_norm": 0.4065510928630829, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 0.2981, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.22614840989399293, | |
| "grad_norm": 0.3683447539806366, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 0.2831, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.2274333440411179, | |
| "grad_norm": 0.4579888880252838, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 0.4177, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.22871827818824286, | |
| "grad_norm": 0.360261470079422, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 0.2714, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.2300032123353678, | |
| "grad_norm": 0.46423205733299255, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 0.312, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.23128814648249277, | |
| "grad_norm": 0.4079225957393646, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 0.2842, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.23257308062961773, | |
| "grad_norm": 0.3761390149593353, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 0.2413, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.2338580147767427, | |
| "grad_norm": 0.370999813079834, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 0.2574, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.23514294892386764, | |
| "grad_norm": 0.39108723402023315, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 0.2654, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.2364278830709926, | |
| "grad_norm": 0.4329393208026886, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 0.323, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.23771281721811757, | |
| "grad_norm": 0.42127084732055664, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 0.3325, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.23899775136524254, | |
| "grad_norm": 0.44995808601379395, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 0.3381, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.24028268551236748, | |
| "grad_norm": 0.4085216522216797, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 0.2399, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.24156761965949244, | |
| "grad_norm": 0.45036596059799194, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 0.2437, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.2428525538066174, | |
| "grad_norm": 0.4397791028022766, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 0.252, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.24413748795374238, | |
| "grad_norm": 0.481174111366272, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 0.2857, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.24542242210086734, | |
| "grad_norm": 0.4796846807003021, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 0.3293, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.24670735624799228, | |
| "grad_norm": 0.5143867135047913, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 0.3064, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.24799229039511725, | |
| "grad_norm": 0.5709989666938782, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 0.3118, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.24927722454224222, | |
| "grad_norm": 0.7193058729171753, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 0.3419, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.25056215868936715, | |
| "grad_norm": 0.5157161951065063, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 0.2546, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.25184709283649215, | |
| "grad_norm": 0.5038528442382812, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 0.263, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.2531320269836171, | |
| "grad_norm": 0.6067041754722595, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 0.3171, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.254416961130742, | |
| "grad_norm": 0.891067385673523, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 0.4093, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.255701895277867, | |
| "grad_norm": 0.6529649496078491, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 0.3524, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.25698682942499196, | |
| "grad_norm": 0.9069101214408875, | |
| "learning_rate": 0.0, | |
| "loss": 0.3883, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.25698682942499196, | |
| "eval_loss": 0.29025572538375854, | |
| "eval_runtime": 27.2736, | |
| "eval_samples_per_second": 48.068, | |
| "eval_steps_per_second": 12.026, | |
| "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": 4.378285967199437e+16, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |