| { | |
| "best_metric": 0.8490656018257141, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.18912529550827423, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0009456264775413711, | |
| "grad_norm": 0.09641855955123901, | |
| "learning_rate": 3.3333333333333335e-07, | |
| "loss": 0.1657, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0009456264775413711, | |
| "eval_loss": 3.487557888031006, | |
| "eval_runtime": 54.6728, | |
| "eval_samples_per_second": 65.17, | |
| "eval_steps_per_second": 16.297, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0018912529550827422, | |
| "grad_norm": 0.1267959028482437, | |
| "learning_rate": 6.666666666666667e-07, | |
| "loss": 0.2459, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0028368794326241137, | |
| "grad_norm": 0.13855446875095367, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 0.2912, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.0037825059101654845, | |
| "grad_norm": 0.15804192423820496, | |
| "learning_rate": 1.3333333333333334e-06, | |
| "loss": 0.3923, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.004728132387706856, | |
| "grad_norm": 0.18095271289348602, | |
| "learning_rate": 1.6666666666666667e-06, | |
| "loss": 0.4793, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.005673758865248227, | |
| "grad_norm": 0.18571291863918304, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 0.585, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.006619385342789598, | |
| "grad_norm": 0.2568407356739044, | |
| "learning_rate": 2.3333333333333336e-06, | |
| "loss": 0.8612, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.007565011820330969, | |
| "grad_norm": 0.3106652498245239, | |
| "learning_rate": 2.666666666666667e-06, | |
| "loss": 1.2384, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.00851063829787234, | |
| "grad_norm": 0.2571044862270355, | |
| "learning_rate": 3e-06, | |
| "loss": 0.8217, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.009456264775413711, | |
| "grad_norm": 0.29234185814857483, | |
| "learning_rate": 3.3333333333333333e-06, | |
| "loss": 0.7603, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.010401891252955082, | |
| "grad_norm": 0.24926097691059113, | |
| "learning_rate": 3.666666666666667e-06, | |
| "loss": 0.6962, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.011347517730496455, | |
| "grad_norm": 0.2906411588191986, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 0.9644, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.012293144208037825, | |
| "grad_norm": 0.49177998304367065, | |
| "learning_rate": 4.333333333333334e-06, | |
| "loss": 1.7599, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.013238770685579196, | |
| "grad_norm": 0.4265400767326355, | |
| "learning_rate": 4.666666666666667e-06, | |
| "loss": 1.493, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.014184397163120567, | |
| "grad_norm": 0.5240830779075623, | |
| "learning_rate": 5e-06, | |
| "loss": 1.9888, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.015130023640661938, | |
| "grad_norm": 0.5526121854782104, | |
| "learning_rate": 5.333333333333334e-06, | |
| "loss": 2.2851, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.01607565011820331, | |
| "grad_norm": 0.5441344976425171, | |
| "learning_rate": 5.666666666666667e-06, | |
| "loss": 1.9884, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.01702127659574468, | |
| "grad_norm": 0.6593867540359497, | |
| "learning_rate": 6e-06, | |
| "loss": 2.5504, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.017966903073286054, | |
| "grad_norm": 0.6868957877159119, | |
| "learning_rate": 6.333333333333334e-06, | |
| "loss": 3.0568, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.018912529550827423, | |
| "grad_norm": 0.43732380867004395, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 1.956, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.019858156028368795, | |
| "grad_norm": 0.8580771088600159, | |
| "learning_rate": 7.000000000000001e-06, | |
| "loss": 2.8262, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.020803782505910164, | |
| "grad_norm": 0.6907251477241516, | |
| "learning_rate": 7.333333333333334e-06, | |
| "loss": 3.0729, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.021749408983451537, | |
| "grad_norm": 0.6712820529937744, | |
| "learning_rate": 7.666666666666667e-06, | |
| "loss": 2.7102, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.02269503546099291, | |
| "grad_norm": 0.6856934428215027, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 2.7061, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.02364066193853428, | |
| "grad_norm": 0.7126139402389526, | |
| "learning_rate": 8.333333333333334e-06, | |
| "loss": 2.6611, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.02458628841607565, | |
| "grad_norm": 0.7721053957939148, | |
| "learning_rate": 8.666666666666668e-06, | |
| "loss": 2.8423, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.02553191489361702, | |
| "grad_norm": 0.883097231388092, | |
| "learning_rate": 9e-06, | |
| "loss": 3.5565, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.026477541371158392, | |
| "grad_norm": 1.0787785053253174, | |
| "learning_rate": 9.333333333333334e-06, | |
| "loss": 3.8743, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.027423167848699765, | |
| "grad_norm": 0.881747305393219, | |
| "learning_rate": 9.666666666666667e-06, | |
| "loss": 3.628, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.028368794326241134, | |
| "grad_norm": 1.07874596118927, | |
| "learning_rate": 1e-05, | |
| "loss": 4.0762, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.029314420803782507, | |
| "grad_norm": 0.8328316807746887, | |
| "learning_rate": 1.0333333333333333e-05, | |
| "loss": 3.512, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.030260047281323876, | |
| "grad_norm": 0.8973889946937561, | |
| "learning_rate": 1.0666666666666667e-05, | |
| "loss": 3.9156, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.031205673758865248, | |
| "grad_norm": 0.7893106937408447, | |
| "learning_rate": 1.1000000000000001e-05, | |
| "loss": 3.3497, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.03215130023640662, | |
| "grad_norm": 0.8067853450775146, | |
| "learning_rate": 1.1333333333333334e-05, | |
| "loss": 3.3522, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.03309692671394799, | |
| "grad_norm": 1.0948940515518188, | |
| "learning_rate": 1.1666666666666668e-05, | |
| "loss": 4.0731, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.03404255319148936, | |
| "grad_norm": 0.8288351893424988, | |
| "learning_rate": 1.2e-05, | |
| "loss": 3.0711, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.034988179669030735, | |
| "grad_norm": 0.7726929187774658, | |
| "learning_rate": 1.2333333333333334e-05, | |
| "loss": 3.0219, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.03593380614657211, | |
| "grad_norm": 0.9294595122337341, | |
| "learning_rate": 1.2666666666666668e-05, | |
| "loss": 4.2024, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.03687943262411347, | |
| "grad_norm": 0.9361292123794556, | |
| "learning_rate": 1.3000000000000001e-05, | |
| "loss": 3.7119, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.037825059101654845, | |
| "grad_norm": 0.9386151432991028, | |
| "learning_rate": 1.3333333333333333e-05, | |
| "loss": 3.7765, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.03877068557919622, | |
| "grad_norm": 0.9665974974632263, | |
| "learning_rate": 1.3666666666666666e-05, | |
| "loss": 3.6401, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.03971631205673759, | |
| "grad_norm": 0.9367566108703613, | |
| "learning_rate": 1.4000000000000001e-05, | |
| "loss": 3.7931, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.04066193853427896, | |
| "grad_norm": 0.763043999671936, | |
| "learning_rate": 1.4333333333333334e-05, | |
| "loss": 3.1378, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.04160756501182033, | |
| "grad_norm": 1.134705901145935, | |
| "learning_rate": 1.4666666666666668e-05, | |
| "loss": 4.1756, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.0425531914893617, | |
| "grad_norm": 1.111829161643982, | |
| "learning_rate": 1.5e-05, | |
| "loss": 4.4766, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.043498817966903074, | |
| "grad_norm": 0.8709962368011475, | |
| "learning_rate": 1.5333333333333334e-05, | |
| "loss": 2.983, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.044444444444444446, | |
| "grad_norm": 1.0141003131866455, | |
| "learning_rate": 1.5666666666666667e-05, | |
| "loss": 4.134, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.04539007092198582, | |
| "grad_norm": 0.9469693303108215, | |
| "learning_rate": 1.6000000000000003e-05, | |
| "loss": 3.1916, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.046335697399527184, | |
| "grad_norm": 0.9570004940032959, | |
| "learning_rate": 1.6333333333333335e-05, | |
| "loss": 3.7659, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.04728132387706856, | |
| "grad_norm": 1.1584417819976807, | |
| "learning_rate": 1.6666666666666667e-05, | |
| "loss": 4.3539, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.04728132387706856, | |
| "eval_loss": 3.263434410095215, | |
| "eval_runtime": 54.1739, | |
| "eval_samples_per_second": 65.77, | |
| "eval_steps_per_second": 16.447, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.04822695035460993, | |
| "grad_norm": 0.1129605770111084, | |
| "learning_rate": 1.7000000000000003e-05, | |
| "loss": 0.175, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.0491725768321513, | |
| "grad_norm": 0.12607906758785248, | |
| "learning_rate": 1.7333333333333336e-05, | |
| "loss": 0.208, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.050118203309692674, | |
| "grad_norm": 0.16476179659366608, | |
| "learning_rate": 1.7666666666666668e-05, | |
| "loss": 0.3471, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.05106382978723404, | |
| "grad_norm": 0.1916539967060089, | |
| "learning_rate": 1.8e-05, | |
| "loss": 0.5088, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.05200945626477541, | |
| "grad_norm": 0.19213494658470154, | |
| "learning_rate": 1.8333333333333333e-05, | |
| "loss": 0.4579, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.052955082742316785, | |
| "grad_norm": 0.1868194341659546, | |
| "learning_rate": 1.866666666666667e-05, | |
| "loss": 0.4327, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.05390070921985816, | |
| "grad_norm": 0.2641420066356659, | |
| "learning_rate": 1.9e-05, | |
| "loss": 0.6682, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.05484633569739953, | |
| "grad_norm": 0.2992357611656189, | |
| "learning_rate": 1.9333333333333333e-05, | |
| "loss": 0.8189, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.055791962174940896, | |
| "grad_norm": 0.2843168377876282, | |
| "learning_rate": 1.9666666666666666e-05, | |
| "loss": 0.7047, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.05673758865248227, | |
| "grad_norm": 0.3119712769985199, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6548, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.05768321513002364, | |
| "grad_norm": 0.4131757616996765, | |
| "learning_rate": 2.0333333333333334e-05, | |
| "loss": 1.0262, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.05862884160756501, | |
| "grad_norm": 0.399532675743103, | |
| "learning_rate": 2.0666666666666666e-05, | |
| "loss": 0.893, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.059574468085106386, | |
| "grad_norm": 0.4315282702445984, | |
| "learning_rate": 2.1e-05, | |
| "loss": 1.1419, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.06052009456264775, | |
| "grad_norm": 0.5299156904220581, | |
| "learning_rate": 2.1333333333333335e-05, | |
| "loss": 1.3849, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.061465721040189124, | |
| "grad_norm": 0.5484279990196228, | |
| "learning_rate": 2.1666666666666667e-05, | |
| "loss": 1.5874, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.062411347517730496, | |
| "grad_norm": 0.4971844255924225, | |
| "learning_rate": 2.2000000000000003e-05, | |
| "loss": 1.3445, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.06335697399527186, | |
| "grad_norm": 0.6058803796768188, | |
| "learning_rate": 2.2333333333333335e-05, | |
| "loss": 1.7419, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.06430260047281323, | |
| "grad_norm": 0.9219263195991516, | |
| "learning_rate": 2.2666666666666668e-05, | |
| "loss": 2.4339, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.06524822695035461, | |
| "grad_norm": 0.8259028196334839, | |
| "learning_rate": 2.3000000000000003e-05, | |
| "loss": 2.2387, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.06619385342789598, | |
| "grad_norm": 0.8110438585281372, | |
| "learning_rate": 2.3333333333333336e-05, | |
| "loss": 2.7578, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.06713947990543735, | |
| "grad_norm": 0.8333178758621216, | |
| "learning_rate": 2.3666666666666668e-05, | |
| "loss": 2.7401, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.06808510638297872, | |
| "grad_norm": 0.6991426944732666, | |
| "learning_rate": 2.4e-05, | |
| "loss": 2.0404, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.0690307328605201, | |
| "grad_norm": 0.8322498798370361, | |
| "learning_rate": 2.4333333333333336e-05, | |
| "loss": 2.5024, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.06997635933806147, | |
| "grad_norm": 0.6573300957679749, | |
| "learning_rate": 2.466666666666667e-05, | |
| "loss": 2.2451, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.07092198581560284, | |
| "grad_norm": 0.9917909502983093, | |
| "learning_rate": 2.5e-05, | |
| "loss": 2.7126, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.07186761229314421, | |
| "grad_norm": 0.9231103658676147, | |
| "learning_rate": 2.5333333333333337e-05, | |
| "loss": 2.4502, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.07281323877068557, | |
| "grad_norm": 0.8092001676559448, | |
| "learning_rate": 2.5666666666666666e-05, | |
| "loss": 2.3372, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.07375886524822695, | |
| "grad_norm": 0.9932003617286682, | |
| "learning_rate": 2.6000000000000002e-05, | |
| "loss": 2.9451, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.07470449172576832, | |
| "grad_norm": 1.0481059551239014, | |
| "learning_rate": 2.633333333333333e-05, | |
| "loss": 3.2932, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.07565011820330969, | |
| "grad_norm": 1.146222472190857, | |
| "learning_rate": 2.6666666666666667e-05, | |
| "loss": 3.2256, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.07659574468085106, | |
| "grad_norm": 1.144538164138794, | |
| "learning_rate": 2.7000000000000002e-05, | |
| "loss": 2.9511, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.07754137115839244, | |
| "grad_norm": 0.8968273997306824, | |
| "learning_rate": 2.733333333333333e-05, | |
| "loss": 2.6522, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.07848699763593381, | |
| "grad_norm": 1.026343822479248, | |
| "learning_rate": 2.7666666666666667e-05, | |
| "loss": 2.9253, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.07943262411347518, | |
| "grad_norm": 0.8586713671684265, | |
| "learning_rate": 2.8000000000000003e-05, | |
| "loss": 2.4545, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.08037825059101655, | |
| "grad_norm": 1.0284498929977417, | |
| "learning_rate": 2.8333333333333335e-05, | |
| "loss": 2.8148, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.08132387706855793, | |
| "grad_norm": 0.8303140997886658, | |
| "learning_rate": 2.8666666666666668e-05, | |
| "loss": 2.3988, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.08226950354609928, | |
| "grad_norm": 0.7819446921348572, | |
| "learning_rate": 2.9e-05, | |
| "loss": 1.9212, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.08321513002364066, | |
| "grad_norm": 1.0593886375427246, | |
| "learning_rate": 2.9333333333333336e-05, | |
| "loss": 2.6381, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.08416075650118203, | |
| "grad_norm": 1.3569374084472656, | |
| "learning_rate": 2.9666666666666672e-05, | |
| "loss": 3.4768, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.0851063829787234, | |
| "grad_norm": 0.878240168094635, | |
| "learning_rate": 3e-05, | |
| "loss": 2.403, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.08605200945626477, | |
| "grad_norm": 1.1406848430633545, | |
| "learning_rate": 3.0333333333333337e-05, | |
| "loss": 3.1817, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.08699763593380615, | |
| "grad_norm": 0.9278729557991028, | |
| "learning_rate": 3.066666666666667e-05, | |
| "loss": 2.433, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.08794326241134752, | |
| "grad_norm": 0.9704206585884094, | |
| "learning_rate": 3.1e-05, | |
| "loss": 2.3984, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.08888888888888889, | |
| "grad_norm": 1.2946637868881226, | |
| "learning_rate": 3.1333333333333334e-05, | |
| "loss": 3.0731, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.08983451536643026, | |
| "grad_norm": 0.9340803623199463, | |
| "learning_rate": 3.1666666666666666e-05, | |
| "loss": 2.2826, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.09078014184397164, | |
| "grad_norm": 1.044756531715393, | |
| "learning_rate": 3.2000000000000005e-05, | |
| "loss": 2.456, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.091725768321513, | |
| "grad_norm": 1.097231388092041, | |
| "learning_rate": 3.233333333333333e-05, | |
| "loss": 2.4481, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.09267139479905437, | |
| "grad_norm": 1.1005371809005737, | |
| "learning_rate": 3.266666666666667e-05, | |
| "loss": 2.3557, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.09361702127659574, | |
| "grad_norm": 0.9912729859352112, | |
| "learning_rate": 3.3e-05, | |
| "loss": 2.1763, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.09456264775413711, | |
| "grad_norm": 1.5450869798660278, | |
| "learning_rate": 3.3333333333333335e-05, | |
| "loss": 3.3969, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.09456264775413711, | |
| "eval_loss": 1.9755831956863403, | |
| "eval_runtime": 54.202, | |
| "eval_samples_per_second": 65.736, | |
| "eval_steps_per_second": 16.439, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.09550827423167849, | |
| "grad_norm": 0.1785513013601303, | |
| "learning_rate": 3.366666666666667e-05, | |
| "loss": 0.1393, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.09645390070921986, | |
| "grad_norm": 0.21543779969215393, | |
| "learning_rate": 3.4000000000000007e-05, | |
| "loss": 0.1675, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.09739952718676123, | |
| "grad_norm": 0.27484190464019775, | |
| "learning_rate": 3.433333333333333e-05, | |
| "loss": 0.2331, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.0983451536643026, | |
| "grad_norm": 0.22659574449062347, | |
| "learning_rate": 3.466666666666667e-05, | |
| "loss": 0.2908, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.09929078014184398, | |
| "grad_norm": 0.28378260135650635, | |
| "learning_rate": 3.5e-05, | |
| "loss": 0.2924, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.10023640661938535, | |
| "grad_norm": 0.2579360902309418, | |
| "learning_rate": 3.5333333333333336e-05, | |
| "loss": 0.3687, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.10118203309692671, | |
| "grad_norm": 0.2652228772640228, | |
| "learning_rate": 3.566666666666667e-05, | |
| "loss": 0.5374, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.10212765957446808, | |
| "grad_norm": 0.32100433111190796, | |
| "learning_rate": 3.6e-05, | |
| "loss": 0.6874, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.10307328605200945, | |
| "grad_norm": 0.30462920665740967, | |
| "learning_rate": 3.633333333333333e-05, | |
| "loss": 0.4254, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.10401891252955082, | |
| "grad_norm": 0.2611568570137024, | |
| "learning_rate": 3.6666666666666666e-05, | |
| "loss": 0.6701, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.1049645390070922, | |
| "grad_norm": 0.29873016476631165, | |
| "learning_rate": 3.7e-05, | |
| "loss": 0.6449, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.10591016548463357, | |
| "grad_norm": 0.2605455219745636, | |
| "learning_rate": 3.733333333333334e-05, | |
| "loss": 0.6369, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.10685579196217494, | |
| "grad_norm": 0.5200917720794678, | |
| "learning_rate": 3.766666666666667e-05, | |
| "loss": 1.1225, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.10780141843971631, | |
| "grad_norm": 0.31477710604667664, | |
| "learning_rate": 3.8e-05, | |
| "loss": 0.772, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.10874704491725769, | |
| "grad_norm": 0.42271727323532104, | |
| "learning_rate": 3.8333333333333334e-05, | |
| "loss": 1.0659, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.10969267139479906, | |
| "grad_norm": 0.37258803844451904, | |
| "learning_rate": 3.866666666666667e-05, | |
| "loss": 0.9266, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.11063829787234042, | |
| "grad_norm": 0.3200523257255554, | |
| "learning_rate": 3.9000000000000006e-05, | |
| "loss": 0.782, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.11158392434988179, | |
| "grad_norm": 0.37096261978149414, | |
| "learning_rate": 3.933333333333333e-05, | |
| "loss": 0.8424, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.11252955082742316, | |
| "grad_norm": 0.3387626111507416, | |
| "learning_rate": 3.966666666666667e-05, | |
| "loss": 0.7941, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.11347517730496454, | |
| "grad_norm": 0.5522480607032776, | |
| "learning_rate": 4e-05, | |
| "loss": 1.3268, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.11442080378250591, | |
| "grad_norm": 0.6289123892784119, | |
| "learning_rate": 4.0333333333333336e-05, | |
| "loss": 1.4588, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.11536643026004728, | |
| "grad_norm": 0.4001694321632385, | |
| "learning_rate": 4.066666666666667e-05, | |
| "loss": 1.056, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.11631205673758865, | |
| "grad_norm": 0.4287935793399811, | |
| "learning_rate": 4.1e-05, | |
| "loss": 1.1171, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.11725768321513003, | |
| "grad_norm": 0.5254130363464355, | |
| "learning_rate": 4.133333333333333e-05, | |
| "loss": 1.2804, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.1182033096926714, | |
| "grad_norm": 0.5890396237373352, | |
| "learning_rate": 4.166666666666667e-05, | |
| "loss": 1.5757, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.11914893617021277, | |
| "grad_norm": 0.5535526275634766, | |
| "learning_rate": 4.2e-05, | |
| "loss": 1.3113, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.12009456264775414, | |
| "grad_norm": 0.5088675618171692, | |
| "learning_rate": 4.233333333333334e-05, | |
| "loss": 1.3381, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.1210401891252955, | |
| "grad_norm": 0.6092506051063538, | |
| "learning_rate": 4.266666666666667e-05, | |
| "loss": 1.5369, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.12198581560283688, | |
| "grad_norm": 0.5836873650550842, | |
| "learning_rate": 4.3e-05, | |
| "loss": 1.3943, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.12293144208037825, | |
| "grad_norm": 0.6445267200469971, | |
| "learning_rate": 4.3333333333333334e-05, | |
| "loss": 1.3146, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.12387706855791962, | |
| "grad_norm": 0.4378418028354645, | |
| "learning_rate": 4.3666666666666666e-05, | |
| "loss": 1.0086, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.12482269503546099, | |
| "grad_norm": 0.6770736575126648, | |
| "learning_rate": 4.4000000000000006e-05, | |
| "loss": 1.5146, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.12576832151300235, | |
| "grad_norm": 0.6039463877677917, | |
| "learning_rate": 4.433333333333334e-05, | |
| "loss": 1.3621, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.12671394799054372, | |
| "grad_norm": 0.6413301229476929, | |
| "learning_rate": 4.466666666666667e-05, | |
| "loss": 1.4514, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.1276595744680851, | |
| "grad_norm": 0.6788724660873413, | |
| "learning_rate": 4.5e-05, | |
| "loss": 1.3301, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.12860520094562647, | |
| "grad_norm": 0.5223095417022705, | |
| "learning_rate": 4.5333333333333335e-05, | |
| "loss": 1.2304, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.12955082742316784, | |
| "grad_norm": 0.6495351195335388, | |
| "learning_rate": 4.566666666666667e-05, | |
| "loss": 1.4054, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.13049645390070921, | |
| "grad_norm": 0.6079133749008179, | |
| "learning_rate": 4.600000000000001e-05, | |
| "loss": 1.3832, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.1314420803782506, | |
| "grad_norm": 0.5553860664367676, | |
| "learning_rate": 4.633333333333333e-05, | |
| "loss": 1.0288, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.13238770685579196, | |
| "grad_norm": 0.550632894039154, | |
| "learning_rate": 4.666666666666667e-05, | |
| "loss": 1.2453, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.13333333333333333, | |
| "grad_norm": 0.9021189212799072, | |
| "learning_rate": 4.7e-05, | |
| "loss": 1.6022, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.1342789598108747, | |
| "grad_norm": 0.6802974939346313, | |
| "learning_rate": 4.7333333333333336e-05, | |
| "loss": 1.5432, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.13522458628841608, | |
| "grad_norm": 0.5788073539733887, | |
| "learning_rate": 4.766666666666667e-05, | |
| "loss": 1.2525, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.13617021276595745, | |
| "grad_norm": 0.6477014422416687, | |
| "learning_rate": 4.8e-05, | |
| "loss": 1.0959, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.13711583924349882, | |
| "grad_norm": 0.4955628216266632, | |
| "learning_rate": 4.8333333333333334e-05, | |
| "loss": 1.1016, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.1380614657210402, | |
| "grad_norm": 0.4967166483402252, | |
| "learning_rate": 4.866666666666667e-05, | |
| "loss": 1.0116, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.13900709219858157, | |
| "grad_norm": 0.5145598649978638, | |
| "learning_rate": 4.9e-05, | |
| "loss": 1.0286, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.13995271867612294, | |
| "grad_norm": 0.7154728174209595, | |
| "learning_rate": 4.933333333333334e-05, | |
| "loss": 1.2846, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.1408983451536643, | |
| "grad_norm": 0.6433387994766235, | |
| "learning_rate": 4.966666666666667e-05, | |
| "loss": 1.0688, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.14184397163120568, | |
| "grad_norm": 0.8238207697868347, | |
| "learning_rate": 5e-05, | |
| "loss": 1.694, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.14184397163120568, | |
| "eval_loss": 1.0205990076065063, | |
| "eval_runtime": 54.4047, | |
| "eval_samples_per_second": 65.491, | |
| "eval_steps_per_second": 16.377, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.14278959810874706, | |
| "grad_norm": 0.45450514554977417, | |
| "learning_rate": 4.995066821070679e-05, | |
| "loss": 0.1451, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.14373522458628843, | |
| "grad_norm": 0.5641042590141296, | |
| "learning_rate": 4.980286753286195e-05, | |
| "loss": 0.19, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.14468085106382977, | |
| "grad_norm": 0.6517515182495117, | |
| "learning_rate": 4.9557181268217227e-05, | |
| "loss": 0.2196, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.14562647754137115, | |
| "grad_norm": 0.5124554634094238, | |
| "learning_rate": 4.9214579028215776e-05, | |
| "loss": 0.2285, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.14657210401891252, | |
| "grad_norm": 0.4333844482898712, | |
| "learning_rate": 4.877641290737884e-05, | |
| "loss": 0.2272, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.1475177304964539, | |
| "grad_norm": 0.3877612352371216, | |
| "learning_rate": 4.8244412147206284e-05, | |
| "loss": 0.3144, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.14846335697399526, | |
| "grad_norm": 0.3211458921432495, | |
| "learning_rate": 4.762067631165049e-05, | |
| "loss": 0.3578, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.14940898345153664, | |
| "grad_norm": 0.24879109859466553, | |
| "learning_rate": 4.690766700109659e-05, | |
| "loss": 0.2756, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.150354609929078, | |
| "grad_norm": 0.22662314772605896, | |
| "learning_rate": 4.610819813755038e-05, | |
| "loss": 0.2839, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.15130023640661938, | |
| "grad_norm": 0.23902210593223572, | |
| "learning_rate": 4.522542485937369e-05, | |
| "loss": 0.3703, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.15224586288416075, | |
| "grad_norm": 0.23963478207588196, | |
| "learning_rate": 4.426283106939474e-05, | |
| "loss": 0.3972, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.15319148936170213, | |
| "grad_norm": 0.23576639592647552, | |
| "learning_rate": 4.3224215685535294e-05, | |
| "loss": 0.3027, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.1541371158392435, | |
| "grad_norm": 0.33861732482910156, | |
| "learning_rate": 4.211367764821722e-05, | |
| "loss": 0.481, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.15508274231678487, | |
| "grad_norm": 0.34786704182624817, | |
| "learning_rate": 4.093559974371725e-05, | |
| "loss": 0.441, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.15602836879432624, | |
| "grad_norm": 0.5184177160263062, | |
| "learning_rate": 3.969463130731183e-05, | |
| "loss": 0.5629, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.15697399527186762, | |
| "grad_norm": 0.38847318291664124, | |
| "learning_rate": 3.8395669874474915e-05, | |
| "loss": 0.5856, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.157919621749409, | |
| "grad_norm": 0.4933064877986908, | |
| "learning_rate": 3.704384185254288e-05, | |
| "loss": 0.6682, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.15886524822695036, | |
| "grad_norm": 0.4688006043434143, | |
| "learning_rate": 3.564448228912682e-05, | |
| "loss": 0.6222, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.15981087470449173, | |
| "grad_norm": 0.7410577535629272, | |
| "learning_rate": 3.4203113817116957e-05, | |
| "loss": 0.8418, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.1607565011820331, | |
| "grad_norm": 0.7604397535324097, | |
| "learning_rate": 3.272542485937369e-05, | |
| "loss": 1.1219, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.16170212765957448, | |
| "grad_norm": 0.6462581753730774, | |
| "learning_rate": 3.121724717912138e-05, | |
| "loss": 0.9533, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.16264775413711585, | |
| "grad_norm": 0.2718241810798645, | |
| "learning_rate": 2.9684532864643122e-05, | |
| "loss": 0.5001, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.1635933806146572, | |
| "grad_norm": 0.35057616233825684, | |
| "learning_rate": 2.8133330839107608e-05, | |
| "loss": 0.5601, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.16453900709219857, | |
| "grad_norm": 0.5394273996353149, | |
| "learning_rate": 2.656976298823284e-05, | |
| "loss": 0.8314, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.16548463356973994, | |
| "grad_norm": 0.3971342444419861, | |
| "learning_rate": 2.5e-05, | |
| "loss": 0.7177, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.16643026004728131, | |
| "grad_norm": 0.3941769003868103, | |
| "learning_rate": 2.3430237011767167e-05, | |
| "loss": 0.6324, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.1673758865248227, | |
| "grad_norm": 0.45238277316093445, | |
| "learning_rate": 2.186666916089239e-05, | |
| "loss": 0.7953, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.16832151300236406, | |
| "grad_norm": 0.5700445771217346, | |
| "learning_rate": 2.031546713535688e-05, | |
| "loss": 0.859, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.16926713947990543, | |
| "grad_norm": 0.47867706418037415, | |
| "learning_rate": 1.8782752820878634e-05, | |
| "loss": 0.8502, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.1702127659574468, | |
| "grad_norm": 0.42379331588745117, | |
| "learning_rate": 1.7274575140626318e-05, | |
| "loss": 0.8364, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.17115839243498818, | |
| "grad_norm": 0.4100590944290161, | |
| "learning_rate": 1.5796886182883053e-05, | |
| "loss": 0.7299, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.17210401891252955, | |
| "grad_norm": 0.5437663197517395, | |
| "learning_rate": 1.4355517710873184e-05, | |
| "loss": 0.7789, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.17304964539007092, | |
| "grad_norm": 0.47485315799713135, | |
| "learning_rate": 1.2956158147457115e-05, | |
| "loss": 0.8491, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.1739952718676123, | |
| "grad_norm": 0.48820245265960693, | |
| "learning_rate": 1.1604330125525079e-05, | |
| "loss": 0.926, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.17494089834515367, | |
| "grad_norm": 0.32581862807273865, | |
| "learning_rate": 1.0305368692688174e-05, | |
| "loss": 0.5398, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.17588652482269504, | |
| "grad_norm": 0.4086638391017914, | |
| "learning_rate": 9.064400256282757e-06, | |
| "loss": 0.8114, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.1768321513002364, | |
| "grad_norm": 0.4679095447063446, | |
| "learning_rate": 7.886322351782783e-06, | |
| "loss": 0.8871, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.17777777777777778, | |
| "grad_norm": 0.4896736145019531, | |
| "learning_rate": 6.775784314464717e-06, | |
| "loss": 0.81, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.17872340425531916, | |
| "grad_norm": 0.4382385015487671, | |
| "learning_rate": 5.737168930605272e-06, | |
| "loss": 0.8284, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.17966903073286053, | |
| "grad_norm": 0.5600558519363403, | |
| "learning_rate": 4.7745751406263165e-06, | |
| "loss": 1.0058, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.1806146572104019, | |
| "grad_norm": 0.5055614113807678, | |
| "learning_rate": 3.891801862449629e-06, | |
| "loss": 0.8065, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.18156028368794327, | |
| "grad_norm": 0.6402313113212585, | |
| "learning_rate": 3.092332998903416e-06, | |
| "loss": 1.0421, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.18250591016548465, | |
| "grad_norm": 0.5336678624153137, | |
| "learning_rate": 2.379323688349516e-06, | |
| "loss": 0.8897, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.183451536643026, | |
| "grad_norm": 0.5516868829727173, | |
| "learning_rate": 1.7555878527937164e-06, | |
| "loss": 0.8386, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.18439716312056736, | |
| "grad_norm": 0.6657341718673706, | |
| "learning_rate": 1.2235870926211619e-06, | |
| "loss": 1.1191, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.18534278959810874, | |
| "grad_norm": 0.5750419497489929, | |
| "learning_rate": 7.854209717842231e-07, | |
| "loss": 1.0913, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.1862884160756501, | |
| "grad_norm": 0.4884861409664154, | |
| "learning_rate": 4.4281873178278475e-07, | |
| "loss": 0.976, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.18723404255319148, | |
| "grad_norm": 0.5465104579925537, | |
| "learning_rate": 1.9713246713805588e-07, | |
| "loss": 0.8359, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.18817966903073285, | |
| "grad_norm": 0.6062586903572083, | |
| "learning_rate": 4.9331789293211026e-08, | |
| "loss": 0.9467, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.18912529550827423, | |
| "grad_norm": 0.7774025201797485, | |
| "learning_rate": 0.0, | |
| "loss": 1.1784, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.18912529550827423, | |
| "eval_loss": 0.8490656018257141, | |
| "eval_runtime": 54.2258, | |
| "eval_samples_per_second": 65.707, | |
| "eval_steps_per_second": 16.431, | |
| "step": 200 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 200, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 50, | |
| "stateful_callbacks": { | |
| "EarlyStoppingCallback": { | |
| "args": { | |
| "early_stopping_patience": 4, | |
| "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": 2.614235570896896e+16, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |