| { | |
| "best_metric": 0.555241584777832, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-150", | |
| "epoch": 3.007518796992481, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.015037593984962405, | |
| "grad_norm": 0.11457784473896027, | |
| "learning_rate": 1e-05, | |
| "loss": 0.6089, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.015037593984962405, | |
| "eval_loss": 0.8226256966590881, | |
| "eval_runtime": 3.38, | |
| "eval_samples_per_second": 33.136, | |
| "eval_steps_per_second": 8.284, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.03007518796992481, | |
| "grad_norm": 0.1543322652578354, | |
| "learning_rate": 2e-05, | |
| "loss": 0.8686, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.045112781954887216, | |
| "grad_norm": 0.16240979731082916, | |
| "learning_rate": 3e-05, | |
| "loss": 0.8141, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.06015037593984962, | |
| "grad_norm": 0.17936722934246063, | |
| "learning_rate": 4e-05, | |
| "loss": 0.7206, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.07518796992481203, | |
| "grad_norm": 0.1907089203596115, | |
| "learning_rate": 5e-05, | |
| "loss": 0.784, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.09022556390977443, | |
| "grad_norm": 0.1974342316389084, | |
| "learning_rate": 6e-05, | |
| "loss": 0.75, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.10526315789473684, | |
| "grad_norm": 0.23462049663066864, | |
| "learning_rate": 7e-05, | |
| "loss": 0.8442, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.12030075187969924, | |
| "grad_norm": 0.2285780906677246, | |
| "learning_rate": 8e-05, | |
| "loss": 0.7375, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.13533834586466165, | |
| "grad_norm": 0.23215308785438538, | |
| "learning_rate": 9e-05, | |
| "loss": 0.7081, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.15037593984962405, | |
| "grad_norm": 0.25710222125053406, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6915, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.16541353383458646, | |
| "grad_norm": 0.23554426431655884, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 0.6363, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.18045112781954886, | |
| "grad_norm": 0.26785606145858765, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 0.5488, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.19548872180451127, | |
| "grad_norm": 0.300786554813385, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 0.6614, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.21052631578947367, | |
| "grad_norm": 0.34388741850852966, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 0.6807, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.22556390977443608, | |
| "grad_norm": 0.43668121099472046, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 0.6145, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.24060150375939848, | |
| "grad_norm": 0.674685001373291, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 0.7951, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.2556390977443609, | |
| "grad_norm": 0.3028257191181183, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 0.7573, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.2706766917293233, | |
| "grad_norm": 0.284080445766449, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 0.7908, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.2857142857142857, | |
| "grad_norm": 0.26206010580062866, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 0.7506, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.3007518796992481, | |
| "grad_norm": 0.22683000564575195, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 0.6703, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.3157894736842105, | |
| "grad_norm": 0.20807921886444092, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 0.6345, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.3308270676691729, | |
| "grad_norm": 0.2084735780954361, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 0.604, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.3458646616541353, | |
| "grad_norm": 0.20663465559482574, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 0.6068, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.3609022556390977, | |
| "grad_norm": 0.2244303822517395, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 0.6732, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.37593984962406013, | |
| "grad_norm": 0.2055593729019165, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 0.5607, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.39097744360902253, | |
| "grad_norm": 0.2082129716873169, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 0.5831, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.40601503759398494, | |
| "grad_norm": 0.1980503648519516, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 0.5023, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.42105263157894735, | |
| "grad_norm": 0.2241385579109192, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 0.5327, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.43609022556390975, | |
| "grad_norm": 0.23611101508140564, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 0.5463, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.45112781954887216, | |
| "grad_norm": 0.2858544588088989, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 0.632, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.46616541353383456, | |
| "grad_norm": 0.3600847125053406, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 0.5857, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.48120300751879697, | |
| "grad_norm": 0.634533703327179, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 0.7072, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.49624060150375937, | |
| "grad_norm": 0.11360874772071838, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 0.6829, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.5112781954887218, | |
| "grad_norm": 0.12521837651729584, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 0.6568, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.5263157894736842, | |
| "grad_norm": 0.1402267962694168, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 0.6213, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.5413533834586466, | |
| "grad_norm": 0.14699652791023254, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 0.6715, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.556390977443609, | |
| "grad_norm": 0.154379740357399, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 0.6248, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.5714285714285714, | |
| "grad_norm": 0.18539367616176605, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 0.6054, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.5864661654135338, | |
| "grad_norm": 0.19980503618717194, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 0.5943, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.6015037593984962, | |
| "grad_norm": 0.20120373368263245, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 0.6088, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.6165413533834586, | |
| "grad_norm": 0.20035739243030548, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 0.5723, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.631578947368421, | |
| "grad_norm": 0.20934194326400757, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 0.5627, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.6466165413533834, | |
| "grad_norm": 0.2100801318883896, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 0.5499, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.6616541353383458, | |
| "grad_norm": 0.2176390141248703, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 0.5382, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.6766917293233082, | |
| "grad_norm": 0.22677041590213776, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 0.5968, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.6917293233082706, | |
| "grad_norm": 0.24823607504367828, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 0.4171, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.706766917293233, | |
| "grad_norm": 0.29296374320983887, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 0.5258, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.7218045112781954, | |
| "grad_norm": 0.4760672152042389, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 0.5269, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.7368421052631579, | |
| "grad_norm": 0.1132049560546875, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 0.6381, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.7518796992481203, | |
| "grad_norm": 0.12979930639266968, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 0.6685, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.7518796992481203, | |
| "eval_loss": 0.5862260460853577, | |
| "eval_runtime": 3.3821, | |
| "eval_samples_per_second": 33.116, | |
| "eval_steps_per_second": 8.279, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.7669172932330827, | |
| "grad_norm": 0.14075040817260742, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 0.6873, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.7819548872180451, | |
| "grad_norm": 0.1557828038930893, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 0.6984, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.7969924812030075, | |
| "grad_norm": 0.1478954255580902, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 0.5438, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.8120300751879699, | |
| "grad_norm": 0.15712638199329376, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 0.4774, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.8270676691729323, | |
| "grad_norm": 0.17720134556293488, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 0.507, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.8421052631578947, | |
| "grad_norm": 0.1685304194688797, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 0.4597, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.8571428571428571, | |
| "grad_norm": 0.21216347813606262, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 0.6305, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.8721804511278195, | |
| "grad_norm": 0.21171170473098755, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 0.5459, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.8872180451127819, | |
| "grad_norm": 0.22091779112815857, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 0.5401, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.9022556390977443, | |
| "grad_norm": 0.23896510899066925, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 0.5378, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.9172932330827067, | |
| "grad_norm": 0.24744990468025208, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 0.477, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.9323308270676691, | |
| "grad_norm": 0.31011462211608887, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 0.514, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.9473684210526315, | |
| "grad_norm": 0.38164022564888, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 0.4897, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.9624060150375939, | |
| "grad_norm": 0.7132401466369629, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 0.4955, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.9774436090225563, | |
| "grad_norm": 0.14969220757484436, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 0.5485, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.9924812030075187, | |
| "grad_norm": 0.22407089173793793, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 0.4904, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 1.0075187969924813, | |
| "grad_norm": 0.8124484419822693, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 1.0379, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 1.0225563909774436, | |
| "grad_norm": 0.12400493770837784, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 0.6444, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 1.037593984962406, | |
| "grad_norm": 0.14174407720565796, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 0.6231, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 1.0526315789473684, | |
| "grad_norm": 0.14666052162647247, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 0.5752, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 1.0676691729323309, | |
| "grad_norm": 0.16379950940608978, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 0.6372, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 1.0827067669172932, | |
| "grad_norm": 0.17941914498806, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 0.6128, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 1.0977443609022557, | |
| "grad_norm": 0.17348399758338928, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 0.4972, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 1.112781954887218, | |
| "grad_norm": 0.19245783984661102, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 0.5319, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 1.1278195488721805, | |
| "grad_norm": 0.1955268234014511, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 0.4335, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 1.1428571428571428, | |
| "grad_norm": 0.22424666583538055, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 0.5399, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 1.1578947368421053, | |
| "grad_norm": 0.20298990607261658, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 0.4551, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 1.1729323308270676, | |
| "grad_norm": 0.21570706367492676, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 0.4494, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 1.1879699248120301, | |
| "grad_norm": 0.242059126496315, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 0.4621, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 1.2030075187969924, | |
| "grad_norm": 0.22083941102027893, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 0.3466, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 1.218045112781955, | |
| "grad_norm": 0.29217585921287537, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 0.4276, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 1.2330827067669172, | |
| "grad_norm": 0.3911180794239044, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 0.3831, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 1.2481203007518797, | |
| "grad_norm": 0.2208462655544281, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 0.4069, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 1.263157894736842, | |
| "grad_norm": 0.14193277060985565, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 0.6454, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 1.2781954887218046, | |
| "grad_norm": 0.17254970967769623, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 0.6492, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 1.2932330827067668, | |
| "grad_norm": 0.17795664072036743, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 0.5842, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 1.3082706766917294, | |
| "grad_norm": 0.21064266562461853, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 0.6568, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 1.3233082706766917, | |
| "grad_norm": 0.21046878397464752, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 0.564, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 1.3383458646616542, | |
| "grad_norm": 0.2297947257757187, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 0.5719, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 1.3533834586466165, | |
| "grad_norm": 0.21308641135692596, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 0.4958, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 1.368421052631579, | |
| "grad_norm": 0.21259456872940063, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 0.4928, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 1.3834586466165413, | |
| "grad_norm": 0.23126700520515442, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 0.4984, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 1.3984962406015038, | |
| "grad_norm": 0.2473018318414688, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 0.502, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 1.413533834586466, | |
| "grad_norm": 0.22947290539741516, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 0.4036, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 1.4285714285714286, | |
| "grad_norm": 0.2785811126232147, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 0.4428, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 1.443609022556391, | |
| "grad_norm": 0.28426018357276917, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 0.3525, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 1.4586466165413534, | |
| "grad_norm": 0.3277798295021057, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 0.4164, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 1.4736842105263157, | |
| "grad_norm": 0.5964440703392029, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 0.5171, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 1.4887218045112782, | |
| "grad_norm": 0.28781628608703613, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 0.455, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 1.5037593984962405, | |
| "grad_norm": 0.13746999204158783, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 0.6356, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.5037593984962405, | |
| "eval_loss": 0.5638276934623718, | |
| "eval_runtime": 3.3731, | |
| "eval_samples_per_second": 33.204, | |
| "eval_steps_per_second": 8.301, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.518796992481203, | |
| "grad_norm": 0.16840910911560059, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 0.7206, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 1.5338345864661656, | |
| "grad_norm": 0.19142265617847443, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 0.6365, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 1.5488721804511278, | |
| "grad_norm": 0.19640135765075684, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 0.5281, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 1.5639097744360901, | |
| "grad_norm": 0.20392058789730072, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 0.5404, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 1.5789473684210527, | |
| "grad_norm": 0.21644632518291473, | |
| "learning_rate": 5e-05, | |
| "loss": 0.5099, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 1.5939849624060152, | |
| "grad_norm": 0.24281011521816254, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 0.5477, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 1.6090225563909775, | |
| "grad_norm": 0.24991512298583984, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 0.4954, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 1.6240601503759398, | |
| "grad_norm": 0.25631803274154663, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 0.4676, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 1.6390977443609023, | |
| "grad_norm": 0.25062280893325806, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 0.4063, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 1.6541353383458648, | |
| "grad_norm": 0.26639077067375183, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 0.4253, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 1.669172932330827, | |
| "grad_norm": 0.2973855137825012, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 0.5089, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 1.6842105263157894, | |
| "grad_norm": 0.2878214120864868, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 0.3851, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 1.699248120300752, | |
| "grad_norm": 0.3631870746612549, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 0.3705, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 1.7142857142857144, | |
| "grad_norm": 0.48780307173728943, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 0.327, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 1.7293233082706767, | |
| "grad_norm": 0.319959819316864, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 0.4293, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 1.744360902255639, | |
| "grad_norm": 0.14660124480724335, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 0.628, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 1.7593984962406015, | |
| "grad_norm": 0.17272670567035675, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 0.5909, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 1.774436090225564, | |
| "grad_norm": 0.18323040008544922, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 0.6245, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 1.7894736842105263, | |
| "grad_norm": 0.19948799908161163, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 0.5076, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 1.8045112781954886, | |
| "grad_norm": 0.21641798317432404, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 0.492, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 1.8195488721804511, | |
| "grad_norm": 0.21880875527858734, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 0.4592, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 1.8345864661654137, | |
| "grad_norm": 0.24364233016967773, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 0.5003, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 1.849624060150376, | |
| "grad_norm": 0.259734570980072, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 0.511, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 1.8646616541353382, | |
| "grad_norm": 0.26867565512657166, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 0.4958, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 1.8796992481203008, | |
| "grad_norm": 0.2885945737361908, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 0.4937, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 1.8947368421052633, | |
| "grad_norm": 0.2841680943965912, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 0.3929, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 1.9097744360902256, | |
| "grad_norm": 0.33479347825050354, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 0.4029, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 1.9248120300751879, | |
| "grad_norm": 0.3607374131679535, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 0.4149, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 1.9398496240601504, | |
| "grad_norm": 0.4023395776748657, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 0.3412, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 1.954887218045113, | |
| "grad_norm": 0.5105811357498169, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 0.3745, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 1.9699248120300752, | |
| "grad_norm": 0.3421552777290344, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 0.5052, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 1.9849624060150375, | |
| "grad_norm": 0.20374922454357147, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 0.4647, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "grad_norm": 0.6035916209220886, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 0.4439, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 2.0150375939849625, | |
| "grad_norm": 0.1350356936454773, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 0.6368, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 2.030075187969925, | |
| "grad_norm": 0.15917502343654633, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 0.6403, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 2.045112781954887, | |
| "grad_norm": 0.1806429624557495, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 0.6467, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 2.0601503759398496, | |
| "grad_norm": 0.19085431098937988, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 0.5759, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 2.075187969924812, | |
| "grad_norm": 0.19593051075935364, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 0.5562, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 2.090225563909774, | |
| "grad_norm": 0.20538921654224396, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 0.4145, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 2.1052631578947367, | |
| "grad_norm": 0.22212770581245422, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 0.4293, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 2.1203007518796992, | |
| "grad_norm": 0.24110625684261322, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 0.49, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 2.1353383458646618, | |
| "grad_norm": 0.25020718574523926, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 0.4459, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 2.1503759398496243, | |
| "grad_norm": 0.2709634602069855, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 0.4458, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 2.1654135338345863, | |
| "grad_norm": 0.28327417373657227, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 0.4517, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 2.180451127819549, | |
| "grad_norm": 0.2878214120864868, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 0.3378, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 2.1954887218045114, | |
| "grad_norm": 0.29141637682914734, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 0.3382, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 2.2105263157894735, | |
| "grad_norm": 0.30836647748947144, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 0.2915, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 2.225563909774436, | |
| "grad_norm": 0.3543986976146698, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 0.2575, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 2.2406015037593985, | |
| "grad_norm": 0.5044798851013184, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 0.2565, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 2.255639097744361, | |
| "grad_norm": 0.14180827140808105, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 0.5729, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 2.255639097744361, | |
| "eval_loss": 0.555241584777832, | |
| "eval_runtime": 3.3881, | |
| "eval_samples_per_second": 33.057, | |
| "eval_steps_per_second": 8.264, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 2.2706766917293235, | |
| "grad_norm": 0.162590891122818, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 0.6461, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 2.2857142857142856, | |
| "grad_norm": 0.17810818552970886, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 0.5882, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 2.300751879699248, | |
| "grad_norm": 0.1994241625070572, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 0.5266, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 2.3157894736842106, | |
| "grad_norm": 0.20960594713687897, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 0.5328, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 2.3308270676691727, | |
| "grad_norm": 0.23722250759601593, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 0.4816, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 2.345864661654135, | |
| "grad_norm": 0.2501429319381714, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 0.4603, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 2.3609022556390977, | |
| "grad_norm": 0.2598116397857666, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 0.4422, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 2.3759398496240602, | |
| "grad_norm": 0.26544246077537537, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 0.4352, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 2.3909774436090228, | |
| "grad_norm": 0.28918272256851196, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 0.4043, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 2.406015037593985, | |
| "grad_norm": 0.2864091694355011, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 0.3689, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 2.4210526315789473, | |
| "grad_norm": 0.3314032256603241, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 0.3901, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 2.43609022556391, | |
| "grad_norm": 0.2950114905834198, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 0.2565, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 2.451127819548872, | |
| "grad_norm": 0.36117634177207947, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 0.3079, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 2.4661654135338344, | |
| "grad_norm": 0.40990445017814636, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 0.3051, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 2.481203007518797, | |
| "grad_norm": 0.435947984457016, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 0.1591, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 2.4962406015037595, | |
| "grad_norm": 0.15773995220661163, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 0.6107, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 2.511278195488722, | |
| "grad_norm": 0.18130731582641602, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 0.6521, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 2.526315789473684, | |
| "grad_norm": 0.1928928941488266, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 0.584, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 2.5413533834586466, | |
| "grad_norm": 0.22154775261878967, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 0.536, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 2.556390977443609, | |
| "grad_norm": 0.24150554835796356, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 0.5197, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 2.571428571428571, | |
| "grad_norm": 0.2482544332742691, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 0.4437, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 2.5864661654135337, | |
| "grad_norm": 0.2693168818950653, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 0.4924, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 2.601503759398496, | |
| "grad_norm": 0.26239049434661865, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 0.3955, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 2.6165413533834587, | |
| "grad_norm": 0.28949567675590515, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 0.4173, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 2.6315789473684212, | |
| "grad_norm": 0.2859938144683838, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 0.388, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 2.6466165413533833, | |
| "grad_norm": 0.2940588891506195, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 0.3952, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 2.661654135338346, | |
| "grad_norm": 0.32735714316368103, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 0.3912, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 2.6766917293233083, | |
| "grad_norm": 0.32457244396209717, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 0.318, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 2.6917293233082704, | |
| "grad_norm": 0.3507591485977173, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 0.3252, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 2.706766917293233, | |
| "grad_norm": 0.41136404871940613, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 0.3047, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 2.7218045112781954, | |
| "grad_norm": 0.4661613702774048, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 0.1978, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 2.736842105263158, | |
| "grad_norm": 0.14506597816944122, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 0.5766, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 2.7518796992481205, | |
| "grad_norm": 0.1723233461380005, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 0.612, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 2.7669172932330826, | |
| "grad_norm": 0.18912146985530853, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 0.55, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 2.781954887218045, | |
| "grad_norm": 0.2082255482673645, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 0.6049, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 2.7969924812030076, | |
| "grad_norm": 0.21885396540164948, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 0.5299, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 2.8120300751879697, | |
| "grad_norm": 0.23263628780841827, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 0.5243, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 2.827067669172932, | |
| "grad_norm": 0.24706439673900604, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 0.4521, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 2.8421052631578947, | |
| "grad_norm": 0.26898086071014404, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 0.4638, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 2.857142857142857, | |
| "grad_norm": 0.277619332075119, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 0.4816, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 2.8721804511278197, | |
| "grad_norm": 0.28311529755592346, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 0.4499, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 2.887218045112782, | |
| "grad_norm": 0.2803266644477844, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 0.3319, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 2.9022556390977443, | |
| "grad_norm": 0.32511386275291443, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 0.4078, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 2.917293233082707, | |
| "grad_norm": 0.35642704367637634, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 0.3913, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 2.932330827067669, | |
| "grad_norm": 0.3581446707248688, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 0.3243, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 2.9473684210526314, | |
| "grad_norm": 0.4329220950603485, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 0.277, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 2.962406015037594, | |
| "grad_norm": 0.45859774947166443, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 0.1649, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 2.9774436090225564, | |
| "grad_norm": 0.19787120819091797, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 0.5944, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 2.992481203007519, | |
| "grad_norm": 0.2929736077785492, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 0.3918, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 3.007518796992481, | |
| "grad_norm": 0.6807894706726074, | |
| "learning_rate": 0.0, | |
| "loss": 0.4638, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 3.007518796992481, | |
| "eval_loss": 0.5553240776062012, | |
| "eval_runtime": 3.3823, | |
| "eval_samples_per_second": 33.113, | |
| "eval_steps_per_second": 8.278, | |
| "step": 200 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 200, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 4, | |
| "save_steps": 50, | |
| "stateful_callbacks": { | |
| "EarlyStoppingCallback": { | |
| "args": { | |
| "early_stopping_patience": 5, | |
| "early_stopping_threshold": 0.0 | |
| }, | |
| "attributes": { | |
| "early_stopping_patience_counter": 1 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 5.542925435928576e+16, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |