| { | |
| "best_metric": 1.6393885612487793, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-150", | |
| "epoch": 0.7575757575757576, | |
| "eval_steps": 50, | |
| "global_step": 150, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.005050505050505051, | |
| "grad_norm": 1.1721141338348389, | |
| "learning_rate": 1e-05, | |
| "loss": 1.9173, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.005050505050505051, | |
| "eval_loss": 2.0529778003692627, | |
| "eval_runtime": 5.5459, | |
| "eval_samples_per_second": 60.225, | |
| "eval_steps_per_second": 15.146, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.010101010101010102, | |
| "grad_norm": 1.1800084114074707, | |
| "learning_rate": 2e-05, | |
| "loss": 1.8947, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.015151515151515152, | |
| "grad_norm": 1.2963718175888062, | |
| "learning_rate": 3e-05, | |
| "loss": 1.991, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.020202020202020204, | |
| "grad_norm": 1.1437054872512817, | |
| "learning_rate": 4e-05, | |
| "loss": 1.8894, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.025252525252525252, | |
| "grad_norm": 0.9491212964057922, | |
| "learning_rate": 5e-05, | |
| "loss": 1.8763, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.030303030303030304, | |
| "grad_norm": 1.068749189376831, | |
| "learning_rate": 6e-05, | |
| "loss": 1.8863, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.03535353535353535, | |
| "grad_norm": 1.2480872869491577, | |
| "learning_rate": 7e-05, | |
| "loss": 1.9224, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.04040404040404041, | |
| "grad_norm": 1.1706750392913818, | |
| "learning_rate": 8e-05, | |
| "loss": 1.8594, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.045454545454545456, | |
| "grad_norm": 1.0032029151916504, | |
| "learning_rate": 9e-05, | |
| "loss": 1.8845, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.050505050505050504, | |
| "grad_norm": 0.9176870584487915, | |
| "learning_rate": 0.0001, | |
| "loss": 1.8175, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.05555555555555555, | |
| "grad_norm": 0.9666992425918579, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 1.8403, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.06060606060606061, | |
| "grad_norm": 0.9135011434555054, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 1.7869, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.06565656565656566, | |
| "grad_norm": 0.8850371837615967, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 1.7742, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.0707070707070707, | |
| "grad_norm": 0.8293886184692383, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 1.8337, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.07575757575757576, | |
| "grad_norm": 0.8078172206878662, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 1.8119, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.08080808080808081, | |
| "grad_norm": 0.8534777760505676, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 1.8344, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.08585858585858586, | |
| "grad_norm": 0.869648814201355, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 1.8352, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.09090909090909091, | |
| "grad_norm": 0.8570549488067627, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 1.835, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.09595959595959595, | |
| "grad_norm": 0.8511046171188354, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 1.7984, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.10101010101010101, | |
| "grad_norm": 0.8205245733261108, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 1.7921, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.10606060606060606, | |
| "grad_norm": 0.8118591904640198, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 1.7726, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.1111111111111111, | |
| "grad_norm": 0.7998741865158081, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 1.7515, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.11616161616161616, | |
| "grad_norm": 0.7742809653282166, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 1.7864, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.12121212121212122, | |
| "grad_norm": 0.8047619462013245, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 1.7929, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.12626262626262627, | |
| "grad_norm": 0.8474924564361572, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 1.8146, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.13131313131313133, | |
| "grad_norm": 0.8512980341911316, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 1.7866, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.13636363636363635, | |
| "grad_norm": 0.7851330041885376, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 1.8195, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.1414141414141414, | |
| "grad_norm": 0.820371687412262, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 1.7535, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.14646464646464646, | |
| "grad_norm": 0.8007573485374451, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 1.7925, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.15151515151515152, | |
| "grad_norm": 0.8156882524490356, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 1.766, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.15656565656565657, | |
| "grad_norm": 0.8136966824531555, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 1.7351, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.16161616161616163, | |
| "grad_norm": 0.875393271446228, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 1.7745, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.16666666666666666, | |
| "grad_norm": 0.8639114499092102, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 1.7181, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.1717171717171717, | |
| "grad_norm": 0.879962146282196, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 1.764, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.17676767676767677, | |
| "grad_norm": 0.8434805870056152, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 1.7422, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.18181818181818182, | |
| "grad_norm": 0.8515347242355347, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 1.6956, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.18686868686868688, | |
| "grad_norm": 0.9043070673942566, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 1.7979, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.1919191919191919, | |
| "grad_norm": 0.9440944194793701, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 1.7796, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.19696969696969696, | |
| "grad_norm": 0.9319767951965332, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 1.7241, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.20202020202020202, | |
| "grad_norm": 0.9053051471710205, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 1.7708, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.20707070707070707, | |
| "grad_norm": 0.9685701131820679, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 1.7346, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.21212121212121213, | |
| "grad_norm": 1.0011823177337646, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 1.736, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.21717171717171718, | |
| "grad_norm": 0.9732013940811157, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 1.7814, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.2222222222222222, | |
| "grad_norm": 0.986873209476471, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 1.7565, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.22727272727272727, | |
| "grad_norm": 0.9863841533660889, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 1.7511, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.23232323232323232, | |
| "grad_norm": 1.0697975158691406, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 1.7923, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.23737373737373738, | |
| "grad_norm": 1.1048533916473389, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 1.7154, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.24242424242424243, | |
| "grad_norm": 1.1453745365142822, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 1.7455, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.2474747474747475, | |
| "grad_norm": 1.3200526237487793, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 1.8051, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.25252525252525254, | |
| "grad_norm": 1.4795987606048584, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 1.7385, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.25252525252525254, | |
| "eval_loss": 1.7528727054595947, | |
| "eval_runtime": 5.5196, | |
| "eval_samples_per_second": 60.511, | |
| "eval_steps_per_second": 15.218, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.25757575757575757, | |
| "grad_norm": 1.2490203380584717, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 1.75, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.26262626262626265, | |
| "grad_norm": 0.9804388880729675, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 1.7115, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.2676767676767677, | |
| "grad_norm": 0.7740570902824402, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 1.7264, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.2727272727272727, | |
| "grad_norm": 0.6553235650062561, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 1.7284, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.2777777777777778, | |
| "grad_norm": 0.6601061820983887, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 1.6666, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.2828282828282828, | |
| "grad_norm": 0.6838630437850952, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 1.7001, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.2878787878787879, | |
| "grad_norm": 0.7081226110458374, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 1.6359, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.29292929292929293, | |
| "grad_norm": 0.6920811533927917, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 1.7222, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.29797979797979796, | |
| "grad_norm": 0.6932616829872131, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 1.6801, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.30303030303030304, | |
| "grad_norm": 0.6817681193351746, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 1.6867, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.30808080808080807, | |
| "grad_norm": 0.6873700618743896, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 1.6937, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.31313131313131315, | |
| "grad_norm": 0.7055866718292236, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 1.6666, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.3181818181818182, | |
| "grad_norm": 0.6892226338386536, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 1.7111, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.32323232323232326, | |
| "grad_norm": 0.6525599956512451, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 1.6603, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.3282828282828283, | |
| "grad_norm": 0.6676272749900818, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 1.6603, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.3333333333333333, | |
| "grad_norm": 0.68507319688797, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 1.6492, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.3383838383838384, | |
| "grad_norm": 0.654543399810791, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 1.7116, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.3434343434343434, | |
| "grad_norm": 0.6445846557617188, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 1.7001, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.3484848484848485, | |
| "grad_norm": 0.660938024520874, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 1.6993, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.35353535353535354, | |
| "grad_norm": 0.6602575778961182, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 1.6992, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.35858585858585856, | |
| "grad_norm": 0.6613306999206543, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 1.6887, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.36363636363636365, | |
| "grad_norm": 0.6551235318183899, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 1.6597, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.3686868686868687, | |
| "grad_norm": 0.6808748841285706, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 1.6926, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.37373737373737376, | |
| "grad_norm": 0.6841230392456055, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 1.6876, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.3787878787878788, | |
| "grad_norm": 0.702394962310791, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 1.664, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.3838383838383838, | |
| "grad_norm": 0.719436764717102, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 1.6915, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.3888888888888889, | |
| "grad_norm": 0.7273593544960022, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 1.6838, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.3939393939393939, | |
| "grad_norm": 0.7527819275856018, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 1.7053, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.398989898989899, | |
| "grad_norm": 0.7678869366645813, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 1.7477, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.40404040404040403, | |
| "grad_norm": 0.7475170493125916, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 1.6924, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.4090909090909091, | |
| "grad_norm": 0.7630130648612976, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 1.6604, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.41414141414141414, | |
| "grad_norm": 0.7588295936584473, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 1.7821, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.41919191919191917, | |
| "grad_norm": 0.7793270945549011, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 1.6746, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.42424242424242425, | |
| "grad_norm": 0.813008725643158, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 1.8024, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.4292929292929293, | |
| "grad_norm": 0.7864987254142761, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 1.6815, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.43434343434343436, | |
| "grad_norm": 0.8113416433334351, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 1.6845, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.4393939393939394, | |
| "grad_norm": 0.8136469125747681, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 1.7042, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.4444444444444444, | |
| "grad_norm": 0.8281291127204895, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 1.6419, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.4494949494949495, | |
| "grad_norm": 0.8532263040542603, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 1.6778, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.45454545454545453, | |
| "grad_norm": 0.879052460193634, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 1.7372, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.4595959595959596, | |
| "grad_norm": 0.875851571559906, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 1.659, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.46464646464646464, | |
| "grad_norm": 0.9393147230148315, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 1.7725, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.4696969696969697, | |
| "grad_norm": 0.9405022263526917, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 1.7574, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.47474747474747475, | |
| "grad_norm": 0.9430553913116455, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 1.6683, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.4797979797979798, | |
| "grad_norm": 0.9621241092681885, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 1.6864, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.48484848484848486, | |
| "grad_norm": 1.0094877481460571, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 1.7532, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.4898989898989899, | |
| "grad_norm": 1.0817221403121948, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 1.6942, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.494949494949495, | |
| "grad_norm": 1.2114548683166504, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 1.7071, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "grad_norm": 0.7440630197525024, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 1.644, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.5050505050505051, | |
| "grad_norm": 0.7109785676002502, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 1.6207, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.5050505050505051, | |
| "eval_loss": 1.673848271369934, | |
| "eval_runtime": 5.5217, | |
| "eval_samples_per_second": 60.489, | |
| "eval_steps_per_second": 15.213, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.51010101010101, | |
| "grad_norm": 0.7127423882484436, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 1.6547, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.5151515151515151, | |
| "grad_norm": 0.7262145280838013, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 1.6783, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.5202020202020202, | |
| "grad_norm": 0.6225936412811279, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 1.6037, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.5252525252525253, | |
| "grad_norm": 0.6136009097099304, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 1.6503, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.5303030303030303, | |
| "grad_norm": 0.5995920300483704, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6204, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.5353535353535354, | |
| "grad_norm": 0.6025617718696594, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 1.6848, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.5404040404040404, | |
| "grad_norm": 0.602440595626831, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 1.6538, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.5454545454545454, | |
| "grad_norm": 0.6105862855911255, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 1.5667, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.5505050505050505, | |
| "grad_norm": 0.6314627528190613, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 1.6723, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.5555555555555556, | |
| "grad_norm": 0.6185853481292725, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 1.6417, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.5606060606060606, | |
| "grad_norm": 0.6278685331344604, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 1.6156, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.5656565656565656, | |
| "grad_norm": 0.6393197178840637, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 1.6447, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.5707070707070707, | |
| "grad_norm": 0.6812595725059509, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 1.6311, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.5757575757575758, | |
| "grad_norm": 0.6392555236816406, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 1.6515, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.5808080808080808, | |
| "grad_norm": 0.6752147078514099, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 1.66, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.5858585858585859, | |
| "grad_norm": 0.6572754979133606, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 1.6608, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.5909090909090909, | |
| "grad_norm": 0.6718370914459229, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 1.6199, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.5959595959595959, | |
| "grad_norm": 0.6734743118286133, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 1.6564, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.601010101010101, | |
| "grad_norm": 0.6663838624954224, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 1.6478, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.6060606060606061, | |
| "grad_norm": 0.6840781569480896, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 1.6227, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.6111111111111112, | |
| "grad_norm": 0.7021604776382446, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 1.6286, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.6161616161616161, | |
| "grad_norm": 0.6946828961372375, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 1.6464, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.6212121212121212, | |
| "grad_norm": 0.7025306820869446, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 1.6627, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.6262626262626263, | |
| "grad_norm": 0.7253495454788208, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 1.6408, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.6313131313131313, | |
| "grad_norm": 0.7254654765129089, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 1.6683, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.6363636363636364, | |
| "grad_norm": 0.7109437584877014, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 1.6036, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.6414141414141414, | |
| "grad_norm": 0.7180439233779907, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 1.6427, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.6464646464646465, | |
| "grad_norm": 0.7294493913650513, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 1.6041, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.6515151515151515, | |
| "grad_norm": 0.7492313981056213, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 1.6785, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.6565656565656566, | |
| "grad_norm": 0.75017911195755, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 1.6926, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.6616161616161617, | |
| "grad_norm": 0.7792496085166931, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 1.7436, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.6666666666666666, | |
| "grad_norm": 0.7525399327278137, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 1.6119, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.6717171717171717, | |
| "grad_norm": 0.7805038690567017, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 1.6924, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.6767676767676768, | |
| "grad_norm": 0.7899602055549622, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 1.6468, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.6818181818181818, | |
| "grad_norm": 0.8201916813850403, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 1.6756, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.6868686868686869, | |
| "grad_norm": 0.8109264969825745, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 1.6284, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.6919191919191919, | |
| "grad_norm": 0.8284419178962708, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 1.6605, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.696969696969697, | |
| "grad_norm": 0.8458845019340515, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 1.5937, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.702020202020202, | |
| "grad_norm": 0.8690614104270935, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 1.6476, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.7070707070707071, | |
| "grad_norm": 0.8767363429069519, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 1.6847, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.7121212121212122, | |
| "grad_norm": 0.8925921320915222, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 1.7541, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.7171717171717171, | |
| "grad_norm": 0.9122419953346252, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 1.6584, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.7222222222222222, | |
| "grad_norm": 0.9675278663635254, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 1.7109, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.7272727272727273, | |
| "grad_norm": 0.966988205909729, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 1.6428, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.7323232323232324, | |
| "grad_norm": 1.0217162370681763, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 1.6801, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.7373737373737373, | |
| "grad_norm": 1.0754969120025635, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 1.7185, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.7424242424242424, | |
| "grad_norm": 1.3362473249435425, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 1.7683, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.7474747474747475, | |
| "grad_norm": 0.6028446555137634, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 1.658, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.7525252525252525, | |
| "grad_norm": 0.655256450176239, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 1.645, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.7575757575757576, | |
| "grad_norm": 0.6057277321815491, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 1.6004, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.7575757575757576, | |
| "eval_loss": 1.6393885612487793, | |
| "eval_runtime": 5.5351, | |
| "eval_samples_per_second": 60.343, | |
| "eval_steps_per_second": 15.176, | |
| "step": 150 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 200, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 2, | |
| "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": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 9988164550656000.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |