| { | |
| "best_metric": 0.0036001873668283224, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 2.9304029304029307, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.014652014652014652, | |
| "grad_norm": 1.4051328897476196, | |
| "learning_rate": 1e-05, | |
| "loss": 1.7838, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.014652014652014652, | |
| "eval_loss": 2.120868682861328, | |
| "eval_runtime": 3.4162, | |
| "eval_samples_per_second": 33.663, | |
| "eval_steps_per_second": 8.489, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.029304029304029304, | |
| "grad_norm": 1.4766777753829956, | |
| "learning_rate": 2e-05, | |
| "loss": 1.8249, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.04395604395604396, | |
| "grad_norm": 1.403977394104004, | |
| "learning_rate": 3e-05, | |
| "loss": 1.9306, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.05860805860805861, | |
| "grad_norm": 1.3712867498397827, | |
| "learning_rate": 4e-05, | |
| "loss": 1.9566, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.07326007326007326, | |
| "grad_norm": 1.811954140663147, | |
| "learning_rate": 5e-05, | |
| "loss": 2.0415, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.08791208791208792, | |
| "grad_norm": 1.8069602251052856, | |
| "learning_rate": 6e-05, | |
| "loss": 1.7068, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.10256410256410256, | |
| "grad_norm": 1.489194393157959, | |
| "learning_rate": 7e-05, | |
| "loss": 1.7191, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.11721611721611722, | |
| "grad_norm": 1.5241328477859497, | |
| "learning_rate": 8e-05, | |
| "loss": 1.6456, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.13186813186813187, | |
| "grad_norm": 1.6911565065383911, | |
| "learning_rate": 9e-05, | |
| "loss": 1.6776, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.14652014652014653, | |
| "grad_norm": 1.7715837955474854, | |
| "learning_rate": 0.0001, | |
| "loss": 1.755, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.16117216117216118, | |
| "grad_norm": 1.8528773784637451, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 1.7202, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.17582417582417584, | |
| "grad_norm": 1.8174198865890503, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 1.4741, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.19047619047619047, | |
| "grad_norm": 1.6014569997787476, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 1.3885, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.20512820512820512, | |
| "grad_norm": 1.8772008419036865, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 1.2719, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.21978021978021978, | |
| "grad_norm": 1.9175750017166138, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 1.3249, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.23443223443223443, | |
| "grad_norm": 2.942521095275879, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 1.5371, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.2490842490842491, | |
| "grad_norm": 4.0308451652526855, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 1.533, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.26373626373626374, | |
| "grad_norm": 3.8481459617614746, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 1.5169, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.2783882783882784, | |
| "grad_norm": 3.228616237640381, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 1.4294, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.29304029304029305, | |
| "grad_norm": 1.9547224044799805, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 1.244, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.3076923076923077, | |
| "grad_norm": 1.5362334251403809, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 1.2708, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.32234432234432236, | |
| "grad_norm": 1.726875901222229, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 1.2775, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.336996336996337, | |
| "grad_norm": 1.7407375574111938, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 1.1433, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.3516483516483517, | |
| "grad_norm": 1.725243091583252, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 0.8104, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.3663003663003663, | |
| "grad_norm": 1.4295977354049683, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 0.7726, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.38095238095238093, | |
| "grad_norm": 1.4068007469177246, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 0.6804, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.3956043956043956, | |
| "grad_norm": 1.8039186000823975, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 0.6846, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.41025641025641024, | |
| "grad_norm": 1.5268199443817139, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 0.672, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.4249084249084249, | |
| "grad_norm": 1.4601694345474243, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 0.5472, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.43956043956043955, | |
| "grad_norm": 1.17428457736969, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 0.4946, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.4542124542124542, | |
| "grad_norm": 1.2569117546081543, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 0.3962, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.46886446886446886, | |
| "grad_norm": 1.4331339597702026, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 0.3742, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.4835164835164835, | |
| "grad_norm": 2.530585765838623, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 0.4389, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.4981684981684982, | |
| "grad_norm": 2.3833696842193604, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 0.3445, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.5128205128205128, | |
| "grad_norm": 3.746232271194458, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 1.1835, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.5274725274725275, | |
| "grad_norm": 3.533053159713745, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 1.0275, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.5421245421245421, | |
| "grad_norm": 2.9000489711761475, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 0.7839, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.5567765567765568, | |
| "grad_norm": 2.057802677154541, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 0.7407, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.5714285714285714, | |
| "grad_norm": 2.0854156017303467, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 0.6865, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.5860805860805861, | |
| "grad_norm": 1.5644015073776245, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 0.3948, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.6007326007326007, | |
| "grad_norm": 1.4325916767120361, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 0.2805, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.6153846153846154, | |
| "grad_norm": 1.347711205482483, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 0.2394, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.63003663003663, | |
| "grad_norm": 1.5750563144683838, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 0.2054, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.6446886446886447, | |
| "grad_norm": 1.460623025894165, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 0.2156, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.6593406593406593, | |
| "grad_norm": 0.9425587058067322, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 0.1474, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.673992673992674, | |
| "grad_norm": 0.6663263440132141, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 0.1191, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.6886446886446886, | |
| "grad_norm": 0.7461935877799988, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 0.0977, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.7032967032967034, | |
| "grad_norm": 0.6047072410583496, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 0.0952, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.717948717948718, | |
| "grad_norm": 0.5037960410118103, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 0.0753, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.7326007326007326, | |
| "grad_norm": 1.2110044956207275, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 0.0982, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.7326007326007326, | |
| "eval_loss": 0.24422450363636017, | |
| "eval_runtime": 3.4154, | |
| "eval_samples_per_second": 33.671, | |
| "eval_steps_per_second": 8.491, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.7472527472527473, | |
| "grad_norm": 1.0307632684707642, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 0.0878, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.7619047619047619, | |
| "grad_norm": 2.2227706909179688, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 0.8299, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.7765567765567766, | |
| "grad_norm": 1.6425745487213135, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 0.4696, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.7912087912087912, | |
| "grad_norm": 1.4220668077468872, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 0.2931, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.8058608058608059, | |
| "grad_norm": 1.6686886548995972, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 0.2843, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.8205128205128205, | |
| "grad_norm": 1.6092051267623901, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 0.2222, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.8351648351648352, | |
| "grad_norm": 1.701392650604248, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 0.1314, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.8498168498168498, | |
| "grad_norm": 1.7303084135055542, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 0.1046, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.8644688644688645, | |
| "grad_norm": 2.248420476913452, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 0.1109, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.8791208791208791, | |
| "grad_norm": 2.016070604324341, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 0.1013, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.8937728937728938, | |
| "grad_norm": 1.4205291271209717, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 0.0675, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.9084249084249084, | |
| "grad_norm": 0.6473082900047302, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 0.049, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.9230769230769231, | |
| "grad_norm": 0.44803163409233093, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 0.0386, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.9377289377289377, | |
| "grad_norm": 0.5309455394744873, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 0.0414, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.9523809523809523, | |
| "grad_norm": 0.43542081117630005, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 0.0297, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.967032967032967, | |
| "grad_norm": 0.27672839164733887, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 0.0266, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.9816849816849816, | |
| "grad_norm": 0.4447838068008423, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 0.0287, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.9963369963369964, | |
| "grad_norm": 0.7602549195289612, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 0.0226, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 1.010989010989011, | |
| "grad_norm": 1.9219874143600464, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 0.5078, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 1.0256410256410255, | |
| "grad_norm": 2.5693531036376953, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 0.3999, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 1.0402930402930404, | |
| "grad_norm": 1.3853951692581177, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 0.154, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 1.054945054945055, | |
| "grad_norm": 1.0422284603118896, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 0.1412, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 1.0695970695970696, | |
| "grad_norm": 1.0995287895202637, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 0.1278, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 1.0842490842490842, | |
| "grad_norm": 0.8546093106269836, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 0.082, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 1.098901098901099, | |
| "grad_norm": 0.7979779839515686, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 0.0707, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 1.1135531135531136, | |
| "grad_norm": 0.5793243646621704, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 0.0489, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 1.1282051282051282, | |
| "grad_norm": 0.8931896686553955, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 0.0484, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 1.1428571428571428, | |
| "grad_norm": 1.081650733947754, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 0.0425, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 1.1575091575091574, | |
| "grad_norm": 0.6045519113540649, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 0.0268, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 1.1721611721611722, | |
| "grad_norm": 0.5937008857727051, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 0.0254, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 1.1868131868131868, | |
| "grad_norm": 0.5532675385475159, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 0.0246, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 1.2014652014652014, | |
| "grad_norm": 0.33690696954727173, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 0.0199, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 1.2161172161172162, | |
| "grad_norm": 0.33141854405403137, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 0.018, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 1.2307692307692308, | |
| "grad_norm": 0.2195536345243454, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 0.0102, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 1.2454212454212454, | |
| "grad_norm": 0.1930108666419983, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 0.0107, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 1.26007326007326, | |
| "grad_norm": 1.6896013021469116, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 0.3249, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 1.2747252747252746, | |
| "grad_norm": 2.0226798057556152, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 0.2439, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 1.2893772893772895, | |
| "grad_norm": 1.3606441020965576, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 0.1206, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 1.304029304029304, | |
| "grad_norm": 0.6428670287132263, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 0.0794, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 1.3186813186813187, | |
| "grad_norm": 0.6892333626747131, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 0.0823, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 1.3333333333333333, | |
| "grad_norm": 0.5389428734779358, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 0.0576, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 1.347985347985348, | |
| "grad_norm": 0.6178902387619019, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 0.0459, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 1.3626373626373627, | |
| "grad_norm": 0.5471458435058594, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 0.0374, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 1.3772893772893773, | |
| "grad_norm": 0.5344598889350891, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 0.0312, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 1.3919413919413919, | |
| "grad_norm": 0.43834978342056274, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 0.0199, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 1.4065934065934065, | |
| "grad_norm": 0.5209547281265259, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 0.022, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 1.4212454212454213, | |
| "grad_norm": 0.3357657194137573, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 0.0114, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 1.435897435897436, | |
| "grad_norm": 0.18191751837730408, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 0.0058, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 1.4505494505494505, | |
| "grad_norm": 0.25728464126586914, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 0.0048, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 1.4652014652014653, | |
| "grad_norm": 0.21814337372779846, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 0.0041, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.4652014652014653, | |
| "eval_loss": 0.03212686628103256, | |
| "eval_runtime": 3.4462, | |
| "eval_samples_per_second": 33.37, | |
| "eval_steps_per_second": 8.415, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.47985347985348, | |
| "grad_norm": 0.22860318422317505, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 0.0025, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 1.4945054945054945, | |
| "grad_norm": 0.3643300235271454, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 0.0029, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 1.5091575091575091, | |
| "grad_norm": 1.091467022895813, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 0.1556, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 1.5238095238095237, | |
| "grad_norm": 0.8354416489601135, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 0.0959, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 1.5384615384615383, | |
| "grad_norm": 1.269963264465332, | |
| "learning_rate": 5e-05, | |
| "loss": 0.0634, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 1.5531135531135531, | |
| "grad_norm": 0.41050803661346436, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 0.0348, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 1.5677655677655677, | |
| "grad_norm": 0.381136029958725, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 0.0258, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 1.5824175824175826, | |
| "grad_norm": 0.3236844837665558, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 0.026, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 1.5970695970695972, | |
| "grad_norm": 0.6296079754829407, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 0.0352, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 1.6117216117216118, | |
| "grad_norm": 0.43410807847976685, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 0.0241, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 1.6263736263736264, | |
| "grad_norm": 0.5278913378715515, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 0.0235, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 1.641025641025641, | |
| "grad_norm": 0.4789339005947113, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 0.0134, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 1.6556776556776556, | |
| "grad_norm": 0.14194229245185852, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 0.0037, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 1.6703296703296702, | |
| "grad_norm": 0.06887496262788773, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 0.0027, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 1.684981684981685, | |
| "grad_norm": 0.1691749542951584, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 0.0041, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 1.6996336996336996, | |
| "grad_norm": 0.32316136360168457, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 0.0087, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 1.7142857142857144, | |
| "grad_norm": 0.20666715502738953, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 0.0069, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 1.728937728937729, | |
| "grad_norm": 0.1677791327238083, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 0.0041, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 1.7435897435897436, | |
| "grad_norm": 0.2694739103317261, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 0.0149, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 1.7582417582417582, | |
| "grad_norm": 0.5503838658332825, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 0.0483, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 1.7728937728937728, | |
| "grad_norm": 0.8629624843597412, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 0.0514, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 1.7875457875457874, | |
| "grad_norm": 0.7726304531097412, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 0.0419, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 1.8021978021978022, | |
| "grad_norm": 0.4805305004119873, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 0.02, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 1.8168498168498168, | |
| "grad_norm": 0.2010086476802826, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 0.0109, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 1.8315018315018317, | |
| "grad_norm": 0.20322047173976898, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 0.0144, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 1.8461538461538463, | |
| "grad_norm": 0.8529400825500488, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 0.0378, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 1.8608058608058609, | |
| "grad_norm": 0.5172489881515503, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 0.018, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 1.8754578754578755, | |
| "grad_norm": 0.6631051898002625, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 0.0155, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 1.89010989010989, | |
| "grad_norm": 0.19856305420398712, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 0.0065, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 1.9047619047619047, | |
| "grad_norm": 0.2891700267791748, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 0.0053, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 1.9194139194139193, | |
| "grad_norm": 0.15734972059726715, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 0.0022, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 1.934065934065934, | |
| "grad_norm": 0.048469528555870056, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 0.0018, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 1.9487179487179487, | |
| "grad_norm": 0.05388571694493294, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 0.0018, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 1.9633699633699635, | |
| "grad_norm": 0.05657542869448662, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 0.0017, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 1.978021978021978, | |
| "grad_norm": 0.24223290383815765, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 0.0061, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 1.9926739926739927, | |
| "grad_norm": 0.1939966380596161, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 0.0055, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 2.0073260073260073, | |
| "grad_norm": 0.27042555809020996, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 0.017, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 2.021978021978022, | |
| "grad_norm": 0.32032516598701477, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 0.0222, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 2.0366300366300365, | |
| "grad_norm": 0.3312196433544159, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 0.0122, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 2.051282051282051, | |
| "grad_norm": 0.35154223442077637, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 0.0074, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 2.065934065934066, | |
| "grad_norm": 0.16670069098472595, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 0.0063, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 2.0805860805860807, | |
| "grad_norm": 0.06954033672809601, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 0.0041, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 2.0952380952380953, | |
| "grad_norm": 0.08824688196182251, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 0.0042, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 2.10989010989011, | |
| "grad_norm": 0.11472660303115845, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 0.0044, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 2.1245421245421245, | |
| "grad_norm": 0.20808860659599304, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 0.0049, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 2.139194139194139, | |
| "grad_norm": 0.15601789951324463, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 0.0053, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 2.1538461538461537, | |
| "grad_norm": 0.07053323835134506, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 0.0019, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 2.1684981684981683, | |
| "grad_norm": 0.04543652758002281, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 0.0016, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 2.183150183150183, | |
| "grad_norm": 0.02706839144229889, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 0.0011, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 2.197802197802198, | |
| "grad_norm": 0.028567973524332047, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 0.0012, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 2.197802197802198, | |
| "eval_loss": 0.0043426197953522205, | |
| "eval_runtime": 3.4484, | |
| "eval_samples_per_second": 33.348, | |
| "eval_steps_per_second": 8.41, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 2.2124542124542126, | |
| "grad_norm": 0.025643927976489067, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 0.0011, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 2.227106227106227, | |
| "grad_norm": 0.028219111263751984, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 0.0012, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 2.241758241758242, | |
| "grad_norm": 0.07654520124197006, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 0.0016, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 2.2564102564102564, | |
| "grad_norm": 0.12539993226528168, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 0.0124, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 2.271062271062271, | |
| "grad_norm": 0.17886903882026672, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 0.0158, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 2.2857142857142856, | |
| "grad_norm": 0.061994776129722595, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 0.0043, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 2.3003663003663, | |
| "grad_norm": 0.0980483740568161, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 0.0037, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 2.315018315018315, | |
| "grad_norm": 0.10740351676940918, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 0.0038, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 2.32967032967033, | |
| "grad_norm": 0.11715144664049149, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 0.0039, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 2.3443223443223444, | |
| "grad_norm": 0.12940457463264465, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 0.004, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 2.358974358974359, | |
| "grad_norm": 0.08561872690916061, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 0.0033, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 2.3736263736263736, | |
| "grad_norm": 0.06982971727848053, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 0.0039, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 2.3882783882783882, | |
| "grad_norm": 0.05386628210544586, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 0.0019, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 2.402930402930403, | |
| "grad_norm": 0.053777169436216354, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 0.0018, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 2.4175824175824174, | |
| "grad_norm": 0.03311780095100403, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 0.0013, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 2.4322344322344325, | |
| "grad_norm": 0.019654465839266777, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 0.001, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 2.446886446886447, | |
| "grad_norm": 0.020478205755352974, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 0.001, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 2.4615384615384617, | |
| "grad_norm": 0.09313038736581802, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 0.002, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 2.4761904761904763, | |
| "grad_norm": 0.031488507986068726, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 0.001, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 2.490842490842491, | |
| "grad_norm": 0.02000267431139946, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 0.0007, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 2.5054945054945055, | |
| "grad_norm": 0.1078575849533081, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 0.0094, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 2.52014652014652, | |
| "grad_norm": 0.11918904632329941, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 0.0125, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 2.5347985347985347, | |
| "grad_norm": 0.08137286454439163, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 0.0061, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 2.5494505494505493, | |
| "grad_norm": 0.04404880106449127, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 0.003, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 2.564102564102564, | |
| "grad_norm": 0.06080146133899689, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 0.0052, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 2.578754578754579, | |
| "grad_norm": 0.04211973398923874, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 0.0026, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 2.5934065934065935, | |
| "grad_norm": 0.1295069009065628, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 0.0029, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 2.608058608058608, | |
| "grad_norm": 0.06931787729263306, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 0.0025, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 2.6227106227106227, | |
| "grad_norm": 0.10869062691926956, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 0.005, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 2.6373626373626373, | |
| "grad_norm": 0.05849863588809967, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 0.0019, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 2.652014652014652, | |
| "grad_norm": 0.029386775568127632, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 0.0013, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 2.6666666666666665, | |
| "grad_norm": 0.023392314091324806, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 0.0011, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 2.6813186813186816, | |
| "grad_norm": 0.017162322998046875, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 0.001, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 2.695970695970696, | |
| "grad_norm": 0.025244666263461113, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 0.001, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 2.7106227106227108, | |
| "grad_norm": 0.018578652292490005, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 0.0008, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 2.7252747252747254, | |
| "grad_norm": 0.01392692793160677, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 0.0008, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 2.73992673992674, | |
| "grad_norm": 0.01738346926867962, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 0.0007, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 2.7545787545787546, | |
| "grad_norm": 0.10509523749351501, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 0.0096, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 2.769230769230769, | |
| "grad_norm": 0.12073224037885666, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 0.0113, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 2.7838827838827838, | |
| "grad_norm": 0.07615958899259567, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 0.0054, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 2.7985347985347984, | |
| "grad_norm": 0.03053397871553898, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 0.0025, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 2.813186813186813, | |
| "grad_norm": 0.033863477408885956, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 0.0025, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 2.8278388278388276, | |
| "grad_norm": 0.13150383532047272, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 0.004, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 2.8424908424908426, | |
| "grad_norm": 0.10117244720458984, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 0.0024, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 2.857142857142857, | |
| "grad_norm": 0.08274346590042114, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 0.0028, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 2.871794871794872, | |
| "grad_norm": 0.06518623977899551, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 0.0021, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 2.8864468864468864, | |
| "grad_norm": 0.03132093325257301, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 0.0012, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 2.901098901098901, | |
| "grad_norm": 0.02533079870045185, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 0.0011, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 2.9157509157509156, | |
| "grad_norm": 0.022283339872956276, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 0.0011, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 2.9304029304029307, | |
| "grad_norm": 0.014048873446881771, | |
| "learning_rate": 0.0, | |
| "loss": 0.0009, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 2.9304029304029307, | |
| "eval_loss": 0.0036001873668283224, | |
| "eval_runtime": 3.4212, | |
| "eval_samples_per_second": 33.614, | |
| "eval_steps_per_second": 8.476, | |
| "step": 200 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 200, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "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": 5.45139913654272e+16, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |