| { | |
| "best_metric": 0.037435051053762436, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.07572889057175312, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0003786444528587656, | |
| "grad_norm": 30.30210304260254, | |
| "learning_rate": 1e-05, | |
| "loss": 3.454, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0003786444528587656, | |
| "eval_loss": 1.2271908521652222, | |
| "eval_runtime": 344.36, | |
| "eval_samples_per_second": 12.917, | |
| "eval_steps_per_second": 3.229, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0007572889057175312, | |
| "grad_norm": 28.70562171936035, | |
| "learning_rate": 2e-05, | |
| "loss": 3.055, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.001135933358576297, | |
| "grad_norm": 27.683151245117188, | |
| "learning_rate": 3e-05, | |
| "loss": 3.0755, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.0015145778114350624, | |
| "grad_norm": 21.200279235839844, | |
| "learning_rate": 4e-05, | |
| "loss": 2.3776, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.001893222264293828, | |
| "grad_norm": 20.016876220703125, | |
| "learning_rate": 5e-05, | |
| "loss": 2.1317, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.002271866717152594, | |
| "grad_norm": 13.269105911254883, | |
| "learning_rate": 6e-05, | |
| "loss": 1.7241, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.0026505111700113595, | |
| "grad_norm": 8.53914737701416, | |
| "learning_rate": 7e-05, | |
| "loss": 1.2721, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.003029155622870125, | |
| "grad_norm": 7.634444713592529, | |
| "learning_rate": 8e-05, | |
| "loss": 1.0289, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.0034078000757288905, | |
| "grad_norm": 5.753473281860352, | |
| "learning_rate": 9e-05, | |
| "loss": 0.9522, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.003786444528587656, | |
| "grad_norm": 5.670541286468506, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7795, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0041650889814464215, | |
| "grad_norm": 5.604635238647461, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 0.7602, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.004543733434305188, | |
| "grad_norm": 4.307402610778809, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 0.5068, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.004922377887163953, | |
| "grad_norm": 3.943270444869995, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 0.4283, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.005301022340022719, | |
| "grad_norm": 4.39333963394165, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 0.4252, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.005679666792881484, | |
| "grad_norm": 4.685744285583496, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 0.4325, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.00605831124574025, | |
| "grad_norm": 4.721577167510986, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 0.4541, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.006436955698599016, | |
| "grad_norm": 3.852276086807251, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 0.4821, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.006815600151457781, | |
| "grad_norm": 3.1198885440826416, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 0.3999, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.007194244604316547, | |
| "grad_norm": 3.4822542667388916, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 0.4251, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.007572889057175312, | |
| "grad_norm": 3.086282968521118, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 0.3033, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.007951533510034078, | |
| "grad_norm": 3.2310450077056885, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 0.3489, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.008330177962892843, | |
| "grad_norm": 3.3141283988952637, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 0.3134, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.00870882241575161, | |
| "grad_norm": 3.379727363586426, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 0.2986, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.009087466868610375, | |
| "grad_norm": 3.217503309249878, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 0.3165, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.00946611132146914, | |
| "grad_norm": 9.431246757507324, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 0.4134, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.009844755774327906, | |
| "grad_norm": 4.396425247192383, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 0.4735, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.010223400227186671, | |
| "grad_norm": 3.3436813354492188, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 0.2919, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.010602044680045438, | |
| "grad_norm": 2.918400764465332, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 0.3274, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.010980689132904203, | |
| "grad_norm": 3.9403295516967773, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 0.3571, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.011359333585762969, | |
| "grad_norm": 3.652174949645996, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 0.3791, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.011737978038621734, | |
| "grad_norm": 3.4695796966552734, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 0.3037, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.0121166224914805, | |
| "grad_norm": 4.472967147827148, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 0.2725, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.012495266944339266, | |
| "grad_norm": 3.7442424297332764, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 0.3347, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.012873911397198031, | |
| "grad_norm": 4.048393726348877, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 0.2825, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.013252555850056797, | |
| "grad_norm": 3.2372047901153564, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 0.3224, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.013631200302915562, | |
| "grad_norm": 3.886826276779175, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 0.2502, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.014009844755774327, | |
| "grad_norm": 5.310039520263672, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 0.4031, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.014388489208633094, | |
| "grad_norm": 3.9522817134857178, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 0.2867, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.01476713366149186, | |
| "grad_norm": 4.22894287109375, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 0.273, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.015145778114350625, | |
| "grad_norm": 3.847644329071045, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 0.3251, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.01552442256720939, | |
| "grad_norm": 4.220029830932617, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 0.387, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.015903067020068155, | |
| "grad_norm": 6.5446367263793945, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 0.501, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.01628171147292692, | |
| "grad_norm": 4.978550910949707, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 0.2667, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.016660355925785686, | |
| "grad_norm": 4.060535430908203, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 0.2507, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.01703900037864445, | |
| "grad_norm": 6.107398986816406, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 0.4149, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.01741764483150322, | |
| "grad_norm": 5.923000335693359, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 0.3435, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.017796289284361985, | |
| "grad_norm": 5.507021427154541, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 0.2431, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.01817493373722075, | |
| "grad_norm": 6.118594646453857, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 0.4265, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.018553578190079516, | |
| "grad_norm": 8.140854835510254, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 0.3929, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.01893222264293828, | |
| "grad_norm": 20.28197479248047, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 0.6931, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.01893222264293828, | |
| "eval_loss": 0.10484325140714645, | |
| "eval_runtime": 346.7247, | |
| "eval_samples_per_second": 12.829, | |
| "eval_steps_per_second": 3.207, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.019310867095797046, | |
| "grad_norm": 15.64964771270752, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 1.3934, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.01968951154865581, | |
| "grad_norm": 3.1674065589904785, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 0.3688, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.020068156001514577, | |
| "grad_norm": 4.302602767944336, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 0.3721, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.020446800454373342, | |
| "grad_norm": 2.2917792797088623, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 0.3455, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.020825444907232107, | |
| "grad_norm": 2.2252116203308105, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 0.2578, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.021204089360090876, | |
| "grad_norm": 2.050823211669922, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 0.3477, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.02158273381294964, | |
| "grad_norm": 1.7291784286499023, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 0.1985, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.021961378265808407, | |
| "grad_norm": 1.78282630443573, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 0.2512, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.022340022718667172, | |
| "grad_norm": 1.5709683895111084, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 0.1674, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.022718667171525937, | |
| "grad_norm": 2.580583095550537, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 0.2681, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.023097311624384703, | |
| "grad_norm": 1.7469924688339233, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 0.1837, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.023475956077243468, | |
| "grad_norm": 1.8775490522384644, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 0.16, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.023854600530102233, | |
| "grad_norm": 2.6333093643188477, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 0.2646, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.024233244982961, | |
| "grad_norm": 2.23219895362854, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 0.1811, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.024611889435819764, | |
| "grad_norm": 2.8453469276428223, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 0.2264, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.024990533888678532, | |
| "grad_norm": 1.8574113845825195, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 0.2062, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.025369178341537298, | |
| "grad_norm": 2.065722942352295, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 0.2177, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.025747822794396063, | |
| "grad_norm": 1.9401975870132446, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 0.154, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.026126467247254828, | |
| "grad_norm": 2.319091320037842, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 0.172, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.026505111700113593, | |
| "grad_norm": 1.731372356414795, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 0.1473, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.02688375615297236, | |
| "grad_norm": 1.685765266418457, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 0.127, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.027262400605831124, | |
| "grad_norm": 2.3166072368621826, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 0.1974, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.02764104505868989, | |
| "grad_norm": 3.6142325401306152, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 0.2996, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.028019689511548655, | |
| "grad_norm": 2.9581050872802734, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 0.2075, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.02839833396440742, | |
| "grad_norm": 2.805510997772217, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 0.218, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.02877697841726619, | |
| "grad_norm": 2.2220678329467773, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 0.2563, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.029155622870124954, | |
| "grad_norm": 1.5375562906265259, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 0.1137, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.02953426732298372, | |
| "grad_norm": 2.0002684593200684, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 0.2283, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.029912911775842484, | |
| "grad_norm": 1.6701806783676147, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 0.1793, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.03029155622870125, | |
| "grad_norm": 1.9673099517822266, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 0.1513, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.030670200681560015, | |
| "grad_norm": 2.2128007411956787, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 0.1565, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.03104884513441878, | |
| "grad_norm": 2.8963911533355713, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 0.1956, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.03142748958727755, | |
| "grad_norm": 1.9358999729156494, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 0.1098, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.03180613404013631, | |
| "grad_norm": 2.467489719390869, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 0.1466, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.03218477849299508, | |
| "grad_norm": 2.4192018508911133, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 0.167, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.03256342294585384, | |
| "grad_norm": 3.016098976135254, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 0.1953, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.03294206739871261, | |
| "grad_norm": 5.171764850616455, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 0.2195, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.03332071185157137, | |
| "grad_norm": 3.947794198989868, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 0.2156, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.03369935630443014, | |
| "grad_norm": 2.628844738006592, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 0.1087, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.0340780007572889, | |
| "grad_norm": 2.4610188007354736, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 0.1795, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.03445664521014767, | |
| "grad_norm": 3.3710789680480957, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 0.2278, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.03483528966300644, | |
| "grad_norm": 4.118891716003418, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 0.202, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.0352139341158652, | |
| "grad_norm": 2.6067090034484863, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 0.1338, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.03559257856872397, | |
| "grad_norm": 4.202768802642822, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 0.2661, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.03597122302158273, | |
| "grad_norm": 5.873966693878174, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 0.1864, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.0363498674744415, | |
| "grad_norm": 2.386099338531494, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 0.1341, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.03672851192730026, | |
| "grad_norm": 10.880393028259277, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 0.2691, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.03710715638015903, | |
| "grad_norm": 7.632635116577148, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 0.284, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.03748580083301779, | |
| "grad_norm": 9.015336036682129, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 0.4067, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.03786444528587656, | |
| "grad_norm": 16.10353660583496, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 0.2029, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.03786444528587656, | |
| "eval_loss": 0.056889016181230545, | |
| "eval_runtime": 347.0888, | |
| "eval_samples_per_second": 12.815, | |
| "eval_steps_per_second": 3.204, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.03824308973873533, | |
| "grad_norm": 5.862347602844238, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 1.4201, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.03862173419159409, | |
| "grad_norm": 1.3635225296020508, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 0.2016, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.03900037864445286, | |
| "grad_norm": 1.1730880737304688, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 0.164, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.03937902309731162, | |
| "grad_norm": 1.2722896337509155, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 0.2026, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.03975766755017039, | |
| "grad_norm": 2.723306179046631, | |
| "learning_rate": 5e-05, | |
| "loss": 0.2805, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.040136312003029154, | |
| "grad_norm": 1.37183678150177, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 0.1796, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.04051495645588792, | |
| "grad_norm": 1.303532361984253, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 0.1872, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.040893600908746684, | |
| "grad_norm": 1.6460245847702026, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 0.1762, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.04127224536160545, | |
| "grad_norm": 1.3503190279006958, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 0.1787, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.041650889814464215, | |
| "grad_norm": 1.532721757888794, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 0.1828, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.042029534267322984, | |
| "grad_norm": 1.484273910522461, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 0.1077, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.04240817872018175, | |
| "grad_norm": 1.2090758085250854, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 0.1398, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.042786823173040514, | |
| "grad_norm": 1.1297125816345215, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 0.1165, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.04316546762589928, | |
| "grad_norm": 1.2648597955703735, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 0.1443, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.043544112078758045, | |
| "grad_norm": 1.3545916080474854, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 0.1526, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.04392275653161681, | |
| "grad_norm": 1.4295732975006104, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 0.164, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.044301400984475575, | |
| "grad_norm": 1.6138875484466553, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 0.1634, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.044680045437334344, | |
| "grad_norm": 2.2143564224243164, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 0.1696, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.045058689890193106, | |
| "grad_norm": 1.946824073791504, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 0.1478, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.045437334343051874, | |
| "grad_norm": 1.0318002700805664, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 0.1257, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.04581597879591064, | |
| "grad_norm": 1.3572828769683838, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 0.1411, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.046194623248769405, | |
| "grad_norm": 1.7299364805221558, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 0.138, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.046573267701628174, | |
| "grad_norm": 1.4923510551452637, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 0.1396, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.046951912154486936, | |
| "grad_norm": 1.890297532081604, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 0.1556, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.047330556607345704, | |
| "grad_norm": 1.558786153793335, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 0.158, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.047709201060204466, | |
| "grad_norm": 2.4953885078430176, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 0.2441, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.048087845513063235, | |
| "grad_norm": 1.3862062692642212, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 0.1151, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.048466489965922, | |
| "grad_norm": 1.265024185180664, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 0.0882, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.048845134418780765, | |
| "grad_norm": 2.4402647018432617, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 0.1424, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.04922377887163953, | |
| "grad_norm": 2.3204729557037354, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 0.1781, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.049602423324498296, | |
| "grad_norm": 1.9608105421066284, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 0.1387, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.049981067777357065, | |
| "grad_norm": 1.5260179042816162, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 0.1172, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.050359712230215826, | |
| "grad_norm": 1.8157596588134766, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 0.1537, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.050738356683074595, | |
| "grad_norm": 1.5637553930282593, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 0.1107, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.05111700113593336, | |
| "grad_norm": 2.8638739585876465, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 0.1487, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.051495645588792126, | |
| "grad_norm": 3.0828709602355957, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 0.1259, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.05187429004165089, | |
| "grad_norm": 3.131708860397339, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 0.1774, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.052252934494509656, | |
| "grad_norm": 2.7193596363067627, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 0.1621, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.05263157894736842, | |
| "grad_norm": 3.783799409866333, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 0.1638, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.05301022340022719, | |
| "grad_norm": 2.9036104679107666, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 0.1835, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.053388867853085956, | |
| "grad_norm": 2.245820999145508, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 0.0936, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.05376751230594472, | |
| "grad_norm": 1.060530424118042, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 0.0667, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.054146156758803486, | |
| "grad_norm": 3.1269538402557373, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 0.1015, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.05452480121166225, | |
| "grad_norm": 4.6061882972717285, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 0.2818, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.05490344566452102, | |
| "grad_norm": 3.7940242290496826, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 0.2089, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.05528209011737978, | |
| "grad_norm": 1.856933832168579, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 0.0939, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.05566073457023855, | |
| "grad_norm": 3.9735310077667236, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 0.0632, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.05603937902309731, | |
| "grad_norm": 3.893869638442993, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 0.2492, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.05641802347595608, | |
| "grad_norm": 4.264387130737305, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 0.1116, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.05679666792881484, | |
| "grad_norm": 6.064884185791016, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 0.1896, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.05679666792881484, | |
| "eval_loss": 0.041598498821258545, | |
| "eval_runtime": 346.8167, | |
| "eval_samples_per_second": 12.825, | |
| "eval_steps_per_second": 3.206, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.05717531238167361, | |
| "grad_norm": 1.5904114246368408, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 0.6746, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.05755395683453238, | |
| "grad_norm": 2.7552506923675537, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 0.2464, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.05793260128739114, | |
| "grad_norm": 1.2251074314117432, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 0.1311, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.05831124574024991, | |
| "grad_norm": 1.2176803350448608, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 0.1551, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.05868989019310867, | |
| "grad_norm": 0.821071445941925, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 0.1037, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.05906853464596744, | |
| "grad_norm": 1.3791486024856567, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 0.1771, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.0594471790988262, | |
| "grad_norm": 1.079137921333313, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 0.1266, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.05982582355168497, | |
| "grad_norm": 1.9268596172332764, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 0.1414, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.06020446800454373, | |
| "grad_norm": 1.0557211637496948, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 0.1216, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.0605831124574025, | |
| "grad_norm": 1.3594329357147217, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 0.1289, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.06096175691026127, | |
| "grad_norm": 1.1114915609359741, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 0.1197, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.06134040136312003, | |
| "grad_norm": 1.0788450241088867, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 0.1168, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.0617190458159788, | |
| "grad_norm": 1.244353175163269, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 0.1085, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.06209769026883756, | |
| "grad_norm": 1.3749324083328247, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 0.1714, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.06247633472169633, | |
| "grad_norm": 1.2642157077789307, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 0.1066, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.0628549791745551, | |
| "grad_norm": 1.334326982498169, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 0.1083, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.06323362362741386, | |
| "grad_norm": 1.3576340675354004, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 0.1058, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.06361226808027262, | |
| "grad_norm": 1.2982534170150757, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 0.0936, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.06399091253313138, | |
| "grad_norm": 1.3901177644729614, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 0.1284, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.06436955698599016, | |
| "grad_norm": 2.074124574661255, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 0.1789, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.06474820143884892, | |
| "grad_norm": 1.7676602602005005, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 0.1211, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.06512684589170768, | |
| "grad_norm": 1.2563903331756592, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 0.0983, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.06550549034456646, | |
| "grad_norm": 1.8831913471221924, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 0.1552, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.06588413479742522, | |
| "grad_norm": 1.4794015884399414, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 0.1357, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.06626277925028398, | |
| "grad_norm": 2.2407169342041016, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 0.1771, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.06664142370314274, | |
| "grad_norm": 1.7588889598846436, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 0.1214, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.06702006815600152, | |
| "grad_norm": 1.844901204109192, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 0.1096, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.06739871260886028, | |
| "grad_norm": 1.7757076025009155, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 0.1458, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.06777735706171904, | |
| "grad_norm": 1.5489846467971802, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 0.1577, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.0681560015145778, | |
| "grad_norm": 4.3092193603515625, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 0.1605, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.06853464596743658, | |
| "grad_norm": 1.4136149883270264, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 0.1114, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.06891329042029534, | |
| "grad_norm": 1.2963083982467651, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 0.0779, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.0692919348731541, | |
| "grad_norm": 2.228961706161499, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 0.1804, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.06967057932601288, | |
| "grad_norm": 2.433332920074463, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 0.1264, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.07004922377887164, | |
| "grad_norm": 1.8022562265396118, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 0.093, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.0704278682317304, | |
| "grad_norm": 1.702027678489685, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 0.125, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.07080651268458917, | |
| "grad_norm": 3.649475574493408, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 0.2044, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.07118515713744794, | |
| "grad_norm": 1.7495139837265015, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 0.1047, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.0715638015903067, | |
| "grad_norm": 1.5214550495147705, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 0.0938, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.07194244604316546, | |
| "grad_norm": 1.9874706268310547, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 0.1139, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.07232109049602424, | |
| "grad_norm": 2.840205669403076, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 0.1447, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.072699734948883, | |
| "grad_norm": 2.353154420852661, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 0.1566, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.07307837940174176, | |
| "grad_norm": 3.3914666175842285, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 0.1363, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.07345702385460053, | |
| "grad_norm": 2.1442184448242188, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 0.0832, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.0738356683074593, | |
| "grad_norm": 6.018881797790527, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 0.1511, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.07421431276031806, | |
| "grad_norm": 4.002391815185547, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 0.0893, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.07459295721317682, | |
| "grad_norm": 3.663280487060547, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 0.1031, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.07497160166603559, | |
| "grad_norm": 7.241387367248535, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 0.1845, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.07535024611889436, | |
| "grad_norm": 8.930119514465332, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 0.2253, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.07572889057175312, | |
| "grad_norm": 5.462854385375977, | |
| "learning_rate": 0.0, | |
| "loss": 0.048, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.07572889057175312, | |
| "eval_loss": 0.037435051053762436, | |
| "eval_runtime": 346.4156, | |
| "eval_samples_per_second": 12.84, | |
| "eval_steps_per_second": 3.21, | |
| "step": 200 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 200, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 50, | |
| "stateful_callbacks": { | |
| "EarlyStoppingCallback": { | |
| "args": { | |
| "early_stopping_patience": 5, | |
| "early_stopping_threshold": 0.0 | |
| }, | |
| "attributes": { | |
| "early_stopping_patience_counter": 0 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 2.987208343761715e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |