| { | |
| "best_metric": 0.8135563731193542, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.709849157054126, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.00354924578527063, | |
| "grad_norm": 0.0637773647904396, | |
| "learning_rate": 1e-05, | |
| "loss": 1.008, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.00354924578527063, | |
| "eval_loss": 2.846020221710205, | |
| "eval_runtime": 13.8102, | |
| "eval_samples_per_second": 34.395, | |
| "eval_steps_per_second": 8.617, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.00709849157054126, | |
| "grad_norm": 0.05443616583943367, | |
| "learning_rate": 2e-05, | |
| "loss": 0.8655, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.01064773735581189, | |
| "grad_norm": 0.10277120023965836, | |
| "learning_rate": 3e-05, | |
| "loss": 1.0278, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.01419698314108252, | |
| "grad_norm": 0.13329511880874634, | |
| "learning_rate": 4e-05, | |
| "loss": 0.8857, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.01774622892635315, | |
| "grad_norm": 0.17435085773468018, | |
| "learning_rate": 5e-05, | |
| "loss": 0.9234, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.02129547471162378, | |
| "grad_norm": 0.15340793132781982, | |
| "learning_rate": 6e-05, | |
| "loss": 1.098, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.024844720496894408, | |
| "grad_norm": 0.2555595636367798, | |
| "learning_rate": 7e-05, | |
| "loss": 1.057, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.02839396628216504, | |
| "grad_norm": 0.26080092787742615, | |
| "learning_rate": 8e-05, | |
| "loss": 1.2819, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.03194321206743567, | |
| "grad_norm": 0.2657241225242615, | |
| "learning_rate": 9e-05, | |
| "loss": 1.3631, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.0354924578527063, | |
| "grad_norm": 0.3126522898674011, | |
| "learning_rate": 0.0001, | |
| "loss": 1.3238, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.03904170363797693, | |
| "grad_norm": 0.36192429065704346, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 1.3867, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.04259094942324756, | |
| "grad_norm": 0.4159677028656006, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 1.2123, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.04614019520851819, | |
| "grad_norm": 0.3637048006057739, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 1.7588, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.049689440993788817, | |
| "grad_norm": 0.2503468096256256, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 1.4214, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.05323868677905945, | |
| "grad_norm": 0.25303223729133606, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 1.4821, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.05678793256433008, | |
| "grad_norm": 0.22958573698997498, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 1.2406, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.06033717834960071, | |
| "grad_norm": 0.4449329078197479, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 1.6172, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.06388642413487133, | |
| "grad_norm": 0.5130823850631714, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 1.3429, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.06743566992014197, | |
| "grad_norm": 0.4500519037246704, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 1.5301, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.0709849157054126, | |
| "grad_norm": 0.7953326106071472, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 1.5241, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.07453416149068323, | |
| "grad_norm": 0.6240103244781494, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 1.6553, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.07808340727595386, | |
| "grad_norm": 0.5798241496086121, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 1.4224, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.08163265306122448, | |
| "grad_norm": 0.5938167572021484, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 1.3839, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.08518189884649512, | |
| "grad_norm": 0.5691270232200623, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 1.3292, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.08873114463176575, | |
| "grad_norm": 0.6260955333709717, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 1.1849, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.09228039041703638, | |
| "grad_norm": 0.5980312824249268, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 1.1255, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.09582963620230701, | |
| "grad_norm": 0.663174569606781, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 1.1381, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.09937888198757763, | |
| "grad_norm": 0.7026616334915161, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 0.9606, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.10292812777284827, | |
| "grad_norm": 0.5387756824493408, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 1.0924, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.1064773735581189, | |
| "grad_norm": 0.6604188084602356, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 1.0046, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.11002661934338953, | |
| "grad_norm": 0.6040458679199219, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 0.7719, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.11357586512866016, | |
| "grad_norm": 0.43716174364089966, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 0.5854, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.11712511091393078, | |
| "grad_norm": 0.3770456910133362, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 0.6235, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.12067435669920142, | |
| "grad_norm": 0.4428984522819519, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 0.5695, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.12422360248447205, | |
| "grad_norm": 0.4358701705932617, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 0.5757, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.12777284826974267, | |
| "grad_norm": 0.5269954800605774, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 0.5932, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.1313220940550133, | |
| "grad_norm": 0.38701364398002625, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 0.6304, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.13487133984028393, | |
| "grad_norm": 0.45931679010391235, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 0.4312, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.13842058562555457, | |
| "grad_norm": 0.5391808152198792, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 0.8634, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.1419698314108252, | |
| "grad_norm": 0.4739307761192322, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 0.6318, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.14551907719609583, | |
| "grad_norm": 0.5952427387237549, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 1.0713, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.14906832298136646, | |
| "grad_norm": 0.413387656211853, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 0.5576, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.1526175687666371, | |
| "grad_norm": 0.4225512146949768, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 0.5857, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.15616681455190773, | |
| "grad_norm": 0.44275951385498047, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 0.7878, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.15971606033717836, | |
| "grad_norm": 0.5345599055290222, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 0.9226, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.16326530612244897, | |
| "grad_norm": 0.4804178476333618, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 0.8944, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.1668145519077196, | |
| "grad_norm": 0.6126044392585754, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 1.0157, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.17036379769299023, | |
| "grad_norm": 0.4933929443359375, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 0.6229, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.17391304347826086, | |
| "grad_norm": 0.5160196423530579, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 0.5569, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.1774622892635315, | |
| "grad_norm": 0.4330124855041504, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 0.4704, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.1774622892635315, | |
| "eval_loss": 0.9678934812545776, | |
| "eval_runtime": 13.9565, | |
| "eval_samples_per_second": 34.034, | |
| "eval_steps_per_second": 8.527, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.18101153504880213, | |
| "grad_norm": 0.34766635298728943, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 0.8223, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.18456078083407276, | |
| "grad_norm": 0.33403506875038147, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 0.9585, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.1881100266193434, | |
| "grad_norm": 0.32657602429389954, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 0.8117, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.19165927240461403, | |
| "grad_norm": 0.29387959837913513, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 0.8368, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.19520851818988466, | |
| "grad_norm": 0.2779003977775574, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 0.8776, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.19875776397515527, | |
| "grad_norm": 0.23073570430278778, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 0.8539, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.2023070097604259, | |
| "grad_norm": 0.19577722251415253, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 0.9496, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.20585625554569653, | |
| "grad_norm": 0.17794418334960938, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 1.0701, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.20940550133096716, | |
| "grad_norm": 0.15639470517635345, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 0.9261, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.2129547471162378, | |
| "grad_norm": 0.1904144436120987, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 1.2298, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.21650399290150843, | |
| "grad_norm": 0.16131433844566345, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 1.1945, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.22005323868677906, | |
| "grad_norm": 0.18274268507957458, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 1.1406, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.2236024844720497, | |
| "grad_norm": 0.22157515585422516, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 1.439, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.22715173025732033, | |
| "grad_norm": 0.2080257683992386, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 1.3212, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.23070097604259096, | |
| "grad_norm": 0.243048757314682, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 1.2805, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.23425022182786157, | |
| "grad_norm": 0.260236918926239, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 1.1913, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.2377994676131322, | |
| "grad_norm": 0.28646838665008545, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 1.1914, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.24134871339840283, | |
| "grad_norm": 0.3513297438621521, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 1.2228, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.24489795918367346, | |
| "grad_norm": 0.3464159667491913, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 1.2172, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.2484472049689441, | |
| "grad_norm": 0.3716575801372528, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 1.1029, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.25199645075421473, | |
| "grad_norm": 0.34499573707580566, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 1.1596, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.25554569653948533, | |
| "grad_norm": 0.29060009121894836, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 1.0781, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.259094942324756, | |
| "grad_norm": 0.32497701048851013, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 1.0572, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.2626441881100266, | |
| "grad_norm": 0.2923493981361389, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 0.9816, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.26619343389529726, | |
| "grad_norm": 0.31345832347869873, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 0.8244, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.26974267968056787, | |
| "grad_norm": 0.37524279952049255, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 0.8458, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.2732919254658385, | |
| "grad_norm": 0.30313193798065186, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 0.9002, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.27684117125110913, | |
| "grad_norm": 0.32463523745536804, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 0.6503, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.2803904170363798, | |
| "grad_norm": 0.37203338742256165, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 1.0524, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.2839396628216504, | |
| "grad_norm": 0.2594308853149414, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 0.501, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.28748890860692106, | |
| "grad_norm": 0.28019317984580994, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 0.5391, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.29103815439219166, | |
| "grad_norm": 0.2912014126777649, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 0.5094, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.29458740017746227, | |
| "grad_norm": 0.36028146743774414, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 0.5493, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.2981366459627329, | |
| "grad_norm": 0.3995708227157593, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 0.6742, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.30168589174800353, | |
| "grad_norm": 0.27118754386901855, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 0.4278, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.3052351375332742, | |
| "grad_norm": 0.26455289125442505, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 0.4913, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.3087843833185448, | |
| "grad_norm": 0.28599533438682556, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 0.406, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.31233362910381546, | |
| "grad_norm": 0.23577168583869934, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 0.4546, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.31588287488908606, | |
| "grad_norm": 0.23274150490760803, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 0.3728, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.3194321206743567, | |
| "grad_norm": 0.29830309748649597, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 0.6929, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.32298136645962733, | |
| "grad_norm": 0.282334566116333, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 0.533, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.32653061224489793, | |
| "grad_norm": 0.31844258308410645, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 0.8791, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.3300798580301686, | |
| "grad_norm": 0.3324856460094452, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 0.5428, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.3336291038154392, | |
| "grad_norm": 0.34939199686050415, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 0.8169, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.33717834960070986, | |
| "grad_norm": 0.40445026755332947, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 0.5412, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.34072759538598046, | |
| "grad_norm": 0.401992529630661, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 0.73, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.3442768411712511, | |
| "grad_norm": 0.5313242673873901, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 0.9308, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.34782608695652173, | |
| "grad_norm": 0.41709527373313904, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 0.5725, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.3513753327417924, | |
| "grad_norm": 0.324158638715744, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 0.4545, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.354924578527063, | |
| "grad_norm": 0.2972598671913147, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 0.4169, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.354924578527063, | |
| "eval_loss": 0.872143566608429, | |
| "eval_runtime": 13.7745, | |
| "eval_samples_per_second": 34.484, | |
| "eval_steps_per_second": 8.639, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.35847382431233366, | |
| "grad_norm": 0.06305402517318726, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 0.7376, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.36202307009760426, | |
| "grad_norm": 0.07278456538915634, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 0.9219, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.36557231588287487, | |
| "grad_norm": 0.09090840071439743, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 0.958, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.3691215616681455, | |
| "grad_norm": 0.10562560707330704, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 0.8287, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.37267080745341613, | |
| "grad_norm": 0.14707393944263458, | |
| "learning_rate": 5e-05, | |
| "loss": 0.8819, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.3762200532386868, | |
| "grad_norm": 0.1610407531261444, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 1.116, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.3797692990239574, | |
| "grad_norm": 0.18440082669258118, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 0.9589, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.38331854480922806, | |
| "grad_norm": 0.16845384240150452, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 1.003, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.38686779059449866, | |
| "grad_norm": 0.17622341215610504, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 0.9909, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.3904170363797693, | |
| "grad_norm": 0.226904034614563, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 1.146, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.3939662821650399, | |
| "grad_norm": 0.21845482289791107, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 1.0654, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.39751552795031053, | |
| "grad_norm": 0.2226574569940567, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 1.1656, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.4010647737355812, | |
| "grad_norm": 0.23775829374790192, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 1.1215, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.4046140195208518, | |
| "grad_norm": 0.2184823602437973, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 1.2335, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.40816326530612246, | |
| "grad_norm": 0.23337799310684204, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 1.0803, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.41171251109139306, | |
| "grad_norm": 0.20988346636295319, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 1.0756, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.4152617568766637, | |
| "grad_norm": 0.18923024833202362, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 1.0092, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.41881100266193433, | |
| "grad_norm": 0.25504937767982483, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 1.3158, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.422360248447205, | |
| "grad_norm": 0.2472749501466751, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 1.3696, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.4259094942324756, | |
| "grad_norm": 0.2592163681983948, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 1.2984, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.42945874001774625, | |
| "grad_norm": 0.2626446783542633, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 1.1976, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.43300798580301686, | |
| "grad_norm": 0.35233691334724426, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 1.0556, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.43655723158828746, | |
| "grad_norm": 0.2648612856864929, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 1.0568, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.4401064773735581, | |
| "grad_norm": 0.3184824287891388, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 1.0718, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.44365572315882873, | |
| "grad_norm": 0.35705751180648804, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 0.9463, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.4472049689440994, | |
| "grad_norm": 0.4138638973236084, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 1.201, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.45075421472937, | |
| "grad_norm": 0.32163676619529724, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 0.7824, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.45430346051464066, | |
| "grad_norm": 0.4062424302101135, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 0.8006, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.45785270629991126, | |
| "grad_norm": 0.36579084396362305, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 0.6973, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.4614019520851819, | |
| "grad_norm": 0.3970095217227936, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 0.6416, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.4649511978704525, | |
| "grad_norm": 0.42958080768585205, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 0.8638, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.46850044365572313, | |
| "grad_norm": 0.31910356879234314, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 0.7986, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.4720496894409938, | |
| "grad_norm": 0.2727908194065094, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 0.513, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.4755989352262644, | |
| "grad_norm": 0.2973743677139282, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 0.5747, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.47914818101153506, | |
| "grad_norm": 0.28417930006980896, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 0.5449, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.48269742679680566, | |
| "grad_norm": 0.24878793954849243, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 0.3623, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.4862466725820763, | |
| "grad_norm": 0.2917158603668213, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 0.6692, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.4897959183673469, | |
| "grad_norm": 0.33632251620292664, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 0.5723, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.4933451641526176, | |
| "grad_norm": 0.2910716235637665, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 0.345, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.4968944099378882, | |
| "grad_norm": 0.18995244801044464, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 0.2948, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.5004436557231589, | |
| "grad_norm": 0.35638427734375, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 0.649, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.5039929015084295, | |
| "grad_norm": 0.27931544184684753, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 0.5447, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.5075421472937001, | |
| "grad_norm": 0.338937371969223, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 0.5011, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.5110913930789707, | |
| "grad_norm": 0.4250231981277466, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 0.8355, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.5146406388642414, | |
| "grad_norm": 0.4304811954498291, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 0.7067, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.518189884649512, | |
| "grad_norm": 0.41890445351600647, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 0.6069, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.5217391304347826, | |
| "grad_norm": 0.5019240975379944, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 0.8946, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.5252883762200532, | |
| "grad_norm": 0.42094922065734863, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 0.5552, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.5288376220053239, | |
| "grad_norm": 0.5289052128791809, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 0.5894, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.5323868677905945, | |
| "grad_norm": 0.3233405351638794, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 0.3862, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.5323868677905945, | |
| "eval_loss": 0.8256970643997192, | |
| "eval_runtime": 13.763, | |
| "eval_samples_per_second": 34.513, | |
| "eval_steps_per_second": 8.646, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.5359361135758651, | |
| "grad_norm": 0.04917902871966362, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 0.8672, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.5394853593611357, | |
| "grad_norm": 0.04684300720691681, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 0.8026, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.5430346051464063, | |
| "grad_norm": 0.06626688688993454, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 0.805, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.546583850931677, | |
| "grad_norm": 0.08166276663541794, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 0.8217, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.5501330967169477, | |
| "grad_norm": 0.09098946303129196, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 0.6913, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.5536823425022183, | |
| "grad_norm": 0.10109563916921616, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 0.7733, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.5572315882874889, | |
| "grad_norm": 0.1208694651722908, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 0.8308, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.5607808340727596, | |
| "grad_norm": 0.1328338235616684, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 0.9621, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.5643300798580302, | |
| "grad_norm": 0.1515001654624939, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 0.9747, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.5678793256433008, | |
| "grad_norm": 0.16993801295757294, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 1.055, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.5714285714285714, | |
| "grad_norm": 0.191247820854187, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 1.3157, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.5749778172138421, | |
| "grad_norm": 0.214799165725708, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 1.2441, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.5785270629991127, | |
| "grad_norm": 0.26233160495758057, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 1.3781, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.5820763087843833, | |
| "grad_norm": 0.21403950452804565, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 1.1118, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.5856255545696539, | |
| "grad_norm": 0.2490546852350235, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 1.2944, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.5891748003549245, | |
| "grad_norm": 0.23730029165744781, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 1.0025, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.5927240461401952, | |
| "grad_norm": 0.30522581934928894, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 1.2762, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.5962732919254659, | |
| "grad_norm": 0.27200835943222046, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 1.052, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.5998225377107365, | |
| "grad_norm": 0.28643932938575745, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 1.0601, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.6033717834960071, | |
| "grad_norm": 0.3039252758026123, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 1.063, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.6069210292812778, | |
| "grad_norm": 0.25883403420448303, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 0.8155, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.6104702750665484, | |
| "grad_norm": 0.30000802874565125, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 1.2432, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.614019520851819, | |
| "grad_norm": 0.32206636667251587, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 1.1191, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.6175687666370896, | |
| "grad_norm": 0.27582842111587524, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 0.773, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.6211180124223602, | |
| "grad_norm": 0.321921169757843, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 0.8355, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.6246672582076309, | |
| "grad_norm": 0.3191986083984375, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 0.7699, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.6282165039929015, | |
| "grad_norm": 0.31827062368392944, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 0.8709, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.6317657497781721, | |
| "grad_norm": 0.2712433934211731, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 0.7872, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.6353149955634427, | |
| "grad_norm": 0.36085808277130127, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 0.5401, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.6388642413487134, | |
| "grad_norm": 0.3143174946308136, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 0.8474, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.642413487133984, | |
| "grad_norm": 0.2958793044090271, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 0.6594, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.6459627329192547, | |
| "grad_norm": 0.284658819437027, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 0.8383, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.6495119787045253, | |
| "grad_norm": 0.31372034549713135, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 0.6267, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.6530612244897959, | |
| "grad_norm": 0.34637507796287537, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 0.5881, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.6566104702750666, | |
| "grad_norm": 0.35337427258491516, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 0.3874, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.6601597160603372, | |
| "grad_norm": 0.34344422817230225, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 0.6524, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.6637089618456078, | |
| "grad_norm": 0.30618658661842346, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 0.3648, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.6672582076308784, | |
| "grad_norm": 0.2853531539440155, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 0.4848, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.6708074534161491, | |
| "grad_norm": 0.27215951681137085, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 0.5995, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.6743566992014197, | |
| "grad_norm": 0.35979869961738586, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 0.3727, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.6779059449866903, | |
| "grad_norm": 0.26503661274909973, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 0.498, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.6814551907719609, | |
| "grad_norm": 0.2748235762119293, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 0.6392, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.6850044365572315, | |
| "grad_norm": 0.2541946470737457, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 0.4664, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.6885536823425022, | |
| "grad_norm": 0.2967659831047058, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 0.6572, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.6921029281277729, | |
| "grad_norm": 0.3442312479019165, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 0.6146, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.6956521739130435, | |
| "grad_norm": 0.43777281045913696, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 0.7797, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.6992014196983141, | |
| "grad_norm": 0.40697500109672546, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 0.7005, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.7027506654835848, | |
| "grad_norm": 0.3615904152393341, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 0.4938, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.7062999112688554, | |
| "grad_norm": 0.3695439398288727, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 0.546, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.709849157054126, | |
| "grad_norm": 0.2854368984699249, | |
| "learning_rate": 0.0, | |
| "loss": 0.3884, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.709849157054126, | |
| "eval_loss": 0.8135563731193542, | |
| "eval_runtime": 13.7629, | |
| "eval_samples_per_second": 34.513, | |
| "eval_steps_per_second": 8.646, | |
| "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": 6.816622616091034e+16, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |