| { | |
| "best_metric": 3.1179463863372803, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 1.2965964343598055, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.006482982171799027, | |
| "grad_norm": 2.4257895946502686, | |
| "learning_rate": 1e-05, | |
| "loss": 2.8806, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.006482982171799027, | |
| "eval_loss": 4.048314571380615, | |
| "eval_runtime": 4.4067, | |
| "eval_samples_per_second": 59.002, | |
| "eval_steps_per_second": 14.75, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.012965964343598054, | |
| "grad_norm": 2.721057891845703, | |
| "learning_rate": 2e-05, | |
| "loss": 3.1378, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.019448946515397084, | |
| "grad_norm": 2.6624343395233154, | |
| "learning_rate": 3e-05, | |
| "loss": 3.2146, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.02593192868719611, | |
| "grad_norm": 3.1744377613067627, | |
| "learning_rate": 4e-05, | |
| "loss": 3.4243, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.03241491085899514, | |
| "grad_norm": 1.3935495615005493, | |
| "learning_rate": 5e-05, | |
| "loss": 3.3465, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.03889789303079417, | |
| "grad_norm": 1.1647844314575195, | |
| "learning_rate": 6e-05, | |
| "loss": 3.1884, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.04538087520259319, | |
| "grad_norm": 1.2633975744247437, | |
| "learning_rate": 7e-05, | |
| "loss": 3.2955, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.05186385737439222, | |
| "grad_norm": 7.194058418273926, | |
| "learning_rate": 8e-05, | |
| "loss": 3.4416, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.05834683954619125, | |
| "grad_norm": 1.3966301679611206, | |
| "learning_rate": 9e-05, | |
| "loss": 3.3087, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.06482982171799027, | |
| "grad_norm": 1.2901732921600342, | |
| "learning_rate": 0.0001, | |
| "loss": 3.1022, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0713128038897893, | |
| "grad_norm": 1.2181003093719482, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 3.4253, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.07779578606158834, | |
| "grad_norm": 1.2063218355178833, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 3.3823, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.08427876823338736, | |
| "grad_norm": 1.182935357093811, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 3.3402, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.09076175040518639, | |
| "grad_norm": 1.1588680744171143, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 3.3243, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.09724473257698542, | |
| "grad_norm": 1.0938173532485962, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 3.4699, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.10372771474878444, | |
| "grad_norm": 1.1799602508544922, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 3.5748, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.11021069692058347, | |
| "grad_norm": 1.173529863357544, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 3.4397, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.1166936790923825, | |
| "grad_norm": 1.1526814699172974, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 3.5394, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.12317666126418152, | |
| "grad_norm": 1.213952898979187, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 3.549, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.12965964343598055, | |
| "grad_norm": 1.2063002586364746, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 3.446, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.13614262560777957, | |
| "grad_norm": 1.3847075700759888, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 3.5069, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.1426256077795786, | |
| "grad_norm": 1.307205319404602, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 3.5155, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.14910858995137763, | |
| "grad_norm": 1.3520137071609497, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 3.5014, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.15559157212317667, | |
| "grad_norm": 1.3639370203018188, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 3.4481, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.1620745542949757, | |
| "grad_norm": 1.3616282939910889, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 3.4407, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.1685575364667747, | |
| "grad_norm": 1.5210310220718384, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 3.5726, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.17504051863857376, | |
| "grad_norm": 2.013072967529297, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 3.4126, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.18152350081037277, | |
| "grad_norm": 1.7007057666778564, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 3.5318, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.1880064829821718, | |
| "grad_norm": 1.72188138961792, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 3.393, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.19448946515397084, | |
| "grad_norm": 1.9005141258239746, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 3.3745, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.20097244732576985, | |
| "grad_norm": 1.9481499195098877, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 3.3485, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.20745542949756887, | |
| "grad_norm": 2.2559003829956055, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 3.6438, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.21393841166936792, | |
| "grad_norm": 2.5377609729766846, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 3.5355, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.22042139384116693, | |
| "grad_norm": 2.702714204788208, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 3.498, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.22690437601296595, | |
| "grad_norm": 2.9564263820648193, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 3.4006, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.233387358184765, | |
| "grad_norm": 3.5847671031951904, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 3.4421, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.23987034035656402, | |
| "grad_norm": 6.074172496795654, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 4.022, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.24635332252836303, | |
| "grad_norm": 7.933667182922363, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 4.0258, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.25283630470016205, | |
| "grad_norm": 4.544643878936768, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 3.0892, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.2593192868719611, | |
| "grad_norm": 4.53206205368042, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 3.1497, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.26580226904376014, | |
| "grad_norm": 3.3829574584960938, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 3.2652, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.27228525121555913, | |
| "grad_norm": 2.460261821746826, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 3.219, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.2787682333873582, | |
| "grad_norm": 1.6331559419631958, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 2.9834, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.2852512155591572, | |
| "grad_norm": 1.2655614614486694, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 3.0261, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.2917341977309562, | |
| "grad_norm": 1.0338799953460693, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 2.9218, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.29821717990275526, | |
| "grad_norm": 0.9313412308692932, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 3.1315, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.3047001620745543, | |
| "grad_norm": 0.8853776454925537, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 2.9292, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.31118314424635335, | |
| "grad_norm": 0.8572593927383423, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 3.056, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.31766612641815234, | |
| "grad_norm": 0.8959136605262756, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 3.3208, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.3241491085899514, | |
| "grad_norm": 0.9137169122695923, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 3.2281, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.3241491085899514, | |
| "eval_loss": 3.407155990600586, | |
| "eval_runtime": 4.4027, | |
| "eval_samples_per_second": 59.055, | |
| "eval_steps_per_second": 14.764, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.33063209076175043, | |
| "grad_norm": 1.031606912612915, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 3.4814, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.3371150729335494, | |
| "grad_norm": 1.012144684791565, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 3.0972, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.34359805510534847, | |
| "grad_norm": 1.0538456439971924, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 3.4517, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.3500810372771475, | |
| "grad_norm": 1.089782476425171, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 3.3362, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.3565640194489465, | |
| "grad_norm": 1.0148874521255493, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 3.4717, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.36304700162074555, | |
| "grad_norm": 1.0671885013580322, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 3.3098, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.3695299837925446, | |
| "grad_norm": 1.080293893814087, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 3.2466, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.3760129659643436, | |
| "grad_norm": 1.1275016069412231, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 3.2311, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.3824959481361426, | |
| "grad_norm": 1.1703333854675293, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 3.4903, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.3889789303079417, | |
| "grad_norm": 1.1692801713943481, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 3.0968, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.39546191247974066, | |
| "grad_norm": 1.2943613529205322, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 3.3956, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.4019448946515397, | |
| "grad_norm": 1.2076880931854248, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 3.4592, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.40842787682333875, | |
| "grad_norm": 1.3995380401611328, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 3.3807, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.41491085899513774, | |
| "grad_norm": 1.3586690425872803, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 3.2966, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.4213938411669368, | |
| "grad_norm": 1.359994649887085, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 3.3712, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.42787682333873583, | |
| "grad_norm": 1.5300426483154297, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 3.4098, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.4343598055105348, | |
| "grad_norm": 1.7160489559173584, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 3.3961, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.44084278768233387, | |
| "grad_norm": 1.6259511709213257, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 3.6358, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.4473257698541329, | |
| "grad_norm": 1.7580747604370117, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 3.3248, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.4538087520259319, | |
| "grad_norm": 1.9767115116119385, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 3.2761, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.46029173419773095, | |
| "grad_norm": 2.133505344390869, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 3.4892, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.46677471636953, | |
| "grad_norm": 2.5297980308532715, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 3.3459, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.473257698541329, | |
| "grad_norm": 2.724898099899292, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 3.2947, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.47974068071312803, | |
| "grad_norm": 3.1555697917938232, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 3.4557, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.4862236628849271, | |
| "grad_norm": 3.3250153064727783, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 2.9288, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.49270664505672607, | |
| "grad_norm": 5.941638469696045, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 3.1569, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.4991896272285251, | |
| "grad_norm": 1.0539709329605103, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 2.8805, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.5056726094003241, | |
| "grad_norm": 1.1901050806045532, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 2.9988, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.5121555915721232, | |
| "grad_norm": 1.0893672704696655, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 3.1139, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.5186385737439222, | |
| "grad_norm": 1.090607762336731, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 2.9885, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.5251215559157212, | |
| "grad_norm": 0.9755169749259949, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 3.0238, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.5316045380875203, | |
| "grad_norm": 0.9265464544296265, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 3.0336, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.5380875202593193, | |
| "grad_norm": 0.9207394123077393, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 3.1685, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.5445705024311183, | |
| "grad_norm": 0.8276278972625732, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 2.9041, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.5510534846029174, | |
| "grad_norm": 0.7896760106086731, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 3.1384, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.5575364667747164, | |
| "grad_norm": 0.8650602102279663, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 3.1776, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.5640194489465153, | |
| "grad_norm": 0.7942522168159485, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 2.9668, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.5705024311183144, | |
| "grad_norm": 0.868003249168396, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 3.0804, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.5769854132901134, | |
| "grad_norm": 0.9763777852058411, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 3.0785, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.5834683954619124, | |
| "grad_norm": 17.73986053466797, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 3.1134, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.5899513776337115, | |
| "grad_norm": 0.9874743223190308, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 3.3085, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.5964343598055105, | |
| "grad_norm": 0.9981576800346375, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 3.15, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.6029173419773096, | |
| "grad_norm": 1.0464242696762085, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 3.251, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.6094003241491086, | |
| "grad_norm": 1.119978904724121, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 3.1136, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.6158833063209076, | |
| "grad_norm": 1.084312081336975, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 3.2541, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.6223662884927067, | |
| "grad_norm": 1.0629708766937256, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 3.1018, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.6288492706645057, | |
| "grad_norm": 1.205837607383728, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 3.2119, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.6353322528363047, | |
| "grad_norm": 1.4000663757324219, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 3.3012, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.6418152350081038, | |
| "grad_norm": 1.2465779781341553, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 3.2355, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.6482982171799028, | |
| "grad_norm": 1.313493251800537, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 3.3348, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.6482982171799028, | |
| "eval_loss": 3.1993484497070312, | |
| "eval_runtime": 4.3952, | |
| "eval_samples_per_second": 59.155, | |
| "eval_steps_per_second": 14.789, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.6547811993517018, | |
| "grad_norm": 1.3143662214279175, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 3.3285, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.6612641815235009, | |
| "grad_norm": 1.3665472269058228, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 3.2328, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.6677471636952999, | |
| "grad_norm": 1.4826375246047974, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 3.3221, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.6742301458670988, | |
| "grad_norm": 1.5050545930862427, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 3.1805, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.6807131280388979, | |
| "grad_norm": 1.5657821893692017, | |
| "learning_rate": 5e-05, | |
| "loss": 3.4312, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.6871961102106969, | |
| "grad_norm": 1.7497036457061768, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 3.5762, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.6936790923824959, | |
| "grad_norm": 1.7621721029281616, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 3.438, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.700162074554295, | |
| "grad_norm": 2.0215108394622803, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 3.372, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.706645056726094, | |
| "grad_norm": 2.0130550861358643, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 3.0789, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.713128038897893, | |
| "grad_norm": 2.149240255355835, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 3.2946, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.7196110210696921, | |
| "grad_norm": 2.407111406326294, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 3.1611, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.7260940032414911, | |
| "grad_norm": 3.102109432220459, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 2.9546, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.7325769854132901, | |
| "grad_norm": 3.2644224166870117, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 3.5911, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.7390599675850892, | |
| "grad_norm": 5.703332901000977, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 3.5122, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.7455429497568882, | |
| "grad_norm": 0.8429614901542664, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 2.9083, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.7520259319286872, | |
| "grad_norm": 0.9492572546005249, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 2.6933, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.7585089141004863, | |
| "grad_norm": 0.8772708177566528, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 2.8183, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.7649918962722853, | |
| "grad_norm": 0.8874571919441223, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 3.0054, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.7714748784440842, | |
| "grad_norm": 0.8024647831916809, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 2.9317, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.7779578606158833, | |
| "grad_norm": 0.8291574716567993, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 2.8545, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.7844408427876823, | |
| "grad_norm": 0.853552520275116, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 2.9571, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.7909238249594813, | |
| "grad_norm": 1.0540350675582886, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 3.0079, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.7974068071312804, | |
| "grad_norm": 1.0573362112045288, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 3.1414, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.8038897893030794, | |
| "grad_norm": 0.9318277835845947, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 2.9553, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.8103727714748784, | |
| "grad_norm": 0.9680076241493225, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 3.0572, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.8168557536466775, | |
| "grad_norm": 0.9529114961624146, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 3.0945, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.8233387358184765, | |
| "grad_norm": 0.880294144153595, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 3.027, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.8298217179902755, | |
| "grad_norm": 0.8740915060043335, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 3.016, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.8363047001620746, | |
| "grad_norm": 0.9977677464485168, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 3.101, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.8427876823338736, | |
| "grad_norm": 0.9654675722122192, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 3.1626, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.8492706645056726, | |
| "grad_norm": 1.0337480306625366, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 3.0022, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.8557536466774717, | |
| "grad_norm": 1.0083832740783691, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 3.12, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.8622366288492707, | |
| "grad_norm": 1.0241321325302124, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 3.0341, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.8687196110210696, | |
| "grad_norm": 1.093923807144165, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 3.2091, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.8752025931928687, | |
| "grad_norm": 1.2071563005447388, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 3.3929, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.8816855753646677, | |
| "grad_norm": 1.2620857954025269, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 3.2127, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.8881685575364667, | |
| "grad_norm": 1.2649301290512085, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 3.2952, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.8946515397082658, | |
| "grad_norm": 1.2944204807281494, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 3.2841, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.9011345218800648, | |
| "grad_norm": 1.4113136529922485, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 3.368, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.9076175040518638, | |
| "grad_norm": 1.4523754119873047, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 3.1832, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.9141004862236629, | |
| "grad_norm": 1.4591470956802368, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 3.3294, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.9205834683954619, | |
| "grad_norm": 1.5156933069229126, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 3.3472, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.9270664505672609, | |
| "grad_norm": 1.716585397720337, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 3.1546, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.93354943273906, | |
| "grad_norm": 1.792197823524475, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 3.3309, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.940032414910859, | |
| "grad_norm": 2.3100404739379883, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 3.3645, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.946515397082658, | |
| "grad_norm": 2.0696702003479004, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 3.5517, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.9529983792544571, | |
| "grad_norm": 2.2145299911499023, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 3.3537, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.9594813614262561, | |
| "grad_norm": 2.33599853515625, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 2.916, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.965964343598055, | |
| "grad_norm": 2.8635406494140625, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 3.4446, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.9724473257698542, | |
| "grad_norm": 3.043067455291748, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 3.4808, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.9724473257698542, | |
| "eval_loss": 3.1411373615264893, | |
| "eval_runtime": 4.3988, | |
| "eval_samples_per_second": 59.107, | |
| "eval_steps_per_second": 14.777, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.9789303079416531, | |
| "grad_norm": 3.7490005493164062, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 2.9795, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.9854132901134521, | |
| "grad_norm": 7.271495342254639, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 3.3842, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.9918962722852512, | |
| "grad_norm": 0.8125553131103516, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 2.9919, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.9983792544570502, | |
| "grad_norm": 1.4885720014572144, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 3.2925, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 1.0048622366288493, | |
| "grad_norm": 6.7827558517456055, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 5.4402, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 1.0113452188006482, | |
| "grad_norm": 0.5606927275657654, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 2.4702, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 1.0178282009724473, | |
| "grad_norm": 0.586924135684967, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 2.7962, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 1.0243111831442464, | |
| "grad_norm": 0.6142417192459106, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 2.7736, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 1.0307941653160453, | |
| "grad_norm": 0.6449275612831116, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 2.744, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 1.0372771474878444, | |
| "grad_norm": 0.6574861407279968, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 2.9087, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 1.0437601296596435, | |
| "grad_norm": 0.6879081130027771, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 2.8085, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 1.0502431118314424, | |
| "grad_norm": 0.7328092455863953, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 2.831, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 1.0567260940032415, | |
| "grad_norm": 0.8173815011978149, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 3.1008, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 1.0632090761750406, | |
| "grad_norm": 0.767900288105011, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 2.6928, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 1.0696920583468394, | |
| "grad_norm": 0.8631361126899719, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 3.0519, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 1.0761750405186385, | |
| "grad_norm": 0.9234204292297363, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 3.0188, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 1.0826580226904376, | |
| "grad_norm": 0.8644314408302307, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 2.8174, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 1.0891410048622365, | |
| "grad_norm": 1.0139473676681519, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 2.9399, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 1.0956239870340356, | |
| "grad_norm": 0.9893440008163452, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 2.9744, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 1.1021069692058347, | |
| "grad_norm": 1.0184575319290161, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 3.0177, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 1.1085899513776336, | |
| "grad_norm": 1.01087486743927, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 2.9195, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 1.1150729335494327, | |
| "grad_norm": 1.102742075920105, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 2.9046, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 1.1215559157212318, | |
| "grad_norm": 1.203591227531433, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 3.1453, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 1.1280388978930307, | |
| "grad_norm": 1.1001300811767578, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 2.9801, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 1.1345218800648298, | |
| "grad_norm": 1.2257055044174194, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 3.1013, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 1.141004862236629, | |
| "grad_norm": 1.2483285665512085, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 3.3192, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 1.147487844408428, | |
| "grad_norm": 1.2322840690612793, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 3.0454, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 1.1539708265802269, | |
| "grad_norm": 1.2523112297058105, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 2.9435, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 1.160453808752026, | |
| "grad_norm": 1.2575554847717285, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 3.0885, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 1.1669367909238249, | |
| "grad_norm": 1.3601508140563965, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 3.1036, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 1.173419773095624, | |
| "grad_norm": 1.4257776737213135, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 3.2161, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 1.179902755267423, | |
| "grad_norm": 1.491587519645691, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 3.0617, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 1.1863857374392222, | |
| "grad_norm": 1.5730777978897095, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 3.0901, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 1.192868719611021, | |
| "grad_norm": 1.6635804176330566, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 3.2284, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 1.1993517017828201, | |
| "grad_norm": 1.7281324863433838, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 3.1782, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 1.205834683954619, | |
| "grad_norm": 1.966333031654358, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 3.1118, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 1.2123176661264181, | |
| "grad_norm": 2.004605770111084, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 2.8502, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 1.2188006482982172, | |
| "grad_norm": 2.276492118835449, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 3.1829, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 1.2252836304700163, | |
| "grad_norm": 2.4115772247314453, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 2.922, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 1.2317666126418152, | |
| "grad_norm": 2.7553176879882812, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 3.1014, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 1.2382495948136143, | |
| "grad_norm": 3.376315116882324, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 3.0917, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 1.2447325769854132, | |
| "grad_norm": 5.8945817947387695, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 3.3228, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 1.2512155591572123, | |
| "grad_norm": 2.612820863723755, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 2.785, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 1.2576985413290114, | |
| "grad_norm": 0.531866192817688, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 2.7123, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 1.2641815235008105, | |
| "grad_norm": 0.5284538269042969, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 2.5691, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 1.2706645056726094, | |
| "grad_norm": 0.5989671945571899, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 2.9147, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 1.2771474878444085, | |
| "grad_norm": 0.6157123446464539, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 2.9009, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 1.2836304700162073, | |
| "grad_norm": 0.631897509098053, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 2.6657, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 1.2901134521880064, | |
| "grad_norm": 0.6588748693466187, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 2.826, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 1.2965964343598055, | |
| "grad_norm": 0.6760578751564026, | |
| "learning_rate": 0.0, | |
| "loss": 2.616, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.2965964343598055, | |
| "eval_loss": 3.1179463863372803, | |
| "eval_runtime": 4.4033, | |
| "eval_samples_per_second": 59.047, | |
| "eval_steps_per_second": 14.762, | |
| "step": 200 | |
| } | |
| ], | |
| "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": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1.354852903944192e+16, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |