| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.3346303501945527, | |
| "eval_steps": 500, | |
| "global_step": 300, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 8.333333333333333e-07, | |
| "loss": 2.0738, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.6666666666666667e-06, | |
| "loss": 2.0648, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 2.5e-06, | |
| "loss": 2.0626, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 3.3333333333333333e-06, | |
| "loss": 2.0344, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.166666666666667e-06, | |
| "loss": 2.0071, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 5e-06, | |
| "loss": 1.9524, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 5.833333333333334e-06, | |
| "loss": 1.934, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 1.922, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 1.9299, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 8.333333333333334e-06, | |
| "loss": 1.8746, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.166666666666666e-06, | |
| "loss": 1.8904, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1e-05, | |
| "loss": 1.8897, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.99982170002055e-06, | |
| "loss": 1.8508, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.99928681279855e-06, | |
| "loss": 1.8484, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.998395376482152e-06, | |
| "loss": 1.8205, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.99714745464859e-06, | |
| "loss": 1.8729, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.995543136299636e-06, | |
| "loss": 1.8406, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.993582535855265e-06, | |
| "loss": 1.8408, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.99126579314548e-06, | |
| "loss": 1.8171, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.988593073400354e-06, | |
| "loss": 1.8545, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.985564567238237e-06, | |
| "loss": 1.8217, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.982180490652165e-06, | |
| "loss": 1.8156, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.978441084994452e-06, | |
| "loss": 1.8182, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.974346616959476e-06, | |
| "loss": 1.808, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.969897378564667e-06, | |
| "loss": 1.8158, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.965093687129669e-06, | |
| "loss": 1.8113, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.959935885253715e-06, | |
| "loss": 1.7973, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.954424340791195e-06, | |
| "loss": 1.8063, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.948559446825412e-06, | |
| "loss": 1.7525, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.942341621640558e-06, | |
| "loss": 1.8202, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9.935771308691872e-06, | |
| "loss": 1.7508, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 9.92884897657402e-06, | |
| "loss": 1.8029, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 9.921575118987672e-06, | |
| "loss": 1.8176, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 9.913950254704291e-06, | |
| "loss": 1.8108, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 9.905974927529135e-06, | |
| "loss": 1.7801, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 9.897649706262474e-06, | |
| "loss": 1.8001, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 9.888975184659018e-06, | |
| "loss": 1.7995, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 9.879951981385577e-06, | |
| "loss": 1.749, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 9.870580739976936e-06, | |
| "loss": 1.7826, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 9.860862128789954e-06, | |
| "loss": 1.7918, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 9.8507968409559e-06, | |
| "loss": 1.7737, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 9.840385594331022e-06, | |
| "loss": 1.7833, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 9.829629131445342e-06, | |
| "loss": 1.7565, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 9.818528219449705e-06, | |
| "loss": 1.8134, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 9.807083650061063e-06, | |
| "loss": 1.7334, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 9.795296239506011e-06, | |
| "loss": 1.7593, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 9.783166828462573e-06, | |
| "loss": 1.7548, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 9.770696282000245e-06, | |
| "loss": 1.7676, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 9.757885489518296e-06, | |
| "loss": 1.7263, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 9.744735364682347e-06, | |
| "loss": 1.7695, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 9.731246845359187e-06, | |
| "loss": 1.7692, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 9.717420893549902e-06, | |
| "loss": 1.7829, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 9.703258495321267e-06, | |
| "loss": 1.7795, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 9.688760660735403e-06, | |
| "loss": 1.8092, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 9.673928423777757e-06, | |
| "loss": 1.8045, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 9.658762842283343e-06, | |
| "loss": 1.7408, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 9.643264997861312e-06, | |
| "loss": 1.7909, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 9.627435995817799e-06, | |
| "loss": 1.7728, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 9.611276965077099e-06, | |
| "loss": 1.7853, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 9.594789058101154e-06, | |
| "loss": 1.772, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 9.577973450807352e-06, | |
| "loss": 1.7973, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 9.560831342484668e-06, | |
| "loss": 1.7499, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 9.543363955708124e-06, | |
| "loss": 1.7405, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 9.525572536251608e-06, | |
| "loss": 1.7678, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 9.507458352999002e-06, | |
| "loss": 1.7726, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 9.48902269785371e-06, | |
| "loss": 1.7626, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 9.470266885646504e-06, | |
| "loss": 1.7545, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 9.451192254041759e-06, | |
| "loss": 1.7615, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 9.431800163442043e-06, | |
| "loss": 1.7661, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 9.412091996891097e-06, | |
| "loss": 1.7408, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 9.392069159975199e-06, | |
| "loss": 1.7684, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 9.371733080722911e-06, | |
| "loss": 1.7349, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 9.351085209503242e-06, | |
| "loss": 1.7502, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 9.330127018922195e-06, | |
| "loss": 1.7951, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 9.308860003717748e-06, | |
| "loss": 1.7644, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 9.287285680653254e-06, | |
| "loss": 1.7552, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 9.265405588409258e-06, | |
| "loss": 1.7459, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 9.243221287473755e-06, | |
| "loss": 1.7434, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 9.220734360030908e-06, | |
| "loss": 1.7695, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 9.197946409848196e-06, | |
| "loss": 1.7681, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 9.174859062162037e-06, | |
| "loss": 1.7225, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 9.151473963561884e-06, | |
| "loss": 1.7267, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 9.12779278187277e-06, | |
| "loss": 1.7213, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 9.103817206036383e-06, | |
| "loss": 1.7147, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 9.079548945990593e-06, | |
| "loss": 1.7904, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 9.054989732547507e-06, | |
| "loss": 1.7604, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 9.030141317270026e-06, | |
| "loss": 1.7312, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 9.005005472346923e-06, | |
| "loss": 1.714, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 8.979583990466454e-06, | |
| "loss": 1.7428, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 8.953878684688492e-06, | |
| "loss": 1.7253, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 8.927891388315241e-06, | |
| "loss": 1.7196, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 8.90162395476046e-06, | |
| "loss": 1.764, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 8.875078257417294e-06, | |
| "loss": 1.7197, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 8.848256189524661e-06, | |
| "loss": 1.7544, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 8.821159664032223e-06, | |
| "loss": 1.7138, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 8.793790613463956e-06, | |
| "loss": 1.716, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 8.766150989780317e-06, | |
| "loss": 1.7636, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 8.738242764239046e-06, | |
| "loss": 1.7617, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 8.710067927254555e-06, | |
| "loss": 1.7321, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 8.681628488255986e-06, | |
| "loss": 1.7598, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 8.6529264755439e-06, | |
| "loss": 1.7286, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 8.6239639361456e-06, | |
| "loss": 1.7361, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 8.594742935669164e-06, | |
| "loss": 1.7589, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 8.565265558156101e-06, | |
| "loss": 1.7667, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 8.535533905932739e-06, | |
| "loss": 1.7755, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 8.505550099460264e-06, | |
| "loss": 1.7496, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 8.47531627718351e-06, | |
| "loss": 1.7133, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 8.444834595378434e-06, | |
| "loss": 1.7038, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 8.414107227998329e-06, | |
| "loss": 1.7706, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 8.383136366518788e-06, | |
| "loss": 1.7657, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 8.351924219781393e-06, | |
| "loss": 1.7443, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 8.320473013836197e-06, | |
| "loss": 1.7503, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 8.288784991782945e-06, | |
| "loss": 1.7287, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 8.256862413611113e-06, | |
| "loss": 1.723, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 8.224707556038712e-06, | |
| "loss": 1.6972, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 8.192322712349917e-06, | |
| "loss": 1.7595, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 8.15971019223152e-06, | |
| "loss": 1.7356, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 8.126872321608185e-06, | |
| "loss": 1.7603, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 8.093811442476573e-06, | |
| "loss": 1.75, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 8.060529912738316e-06, | |
| "loss": 1.7136, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 8.027030106031835e-06, | |
| "loss": 1.7263, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 7.993314411563075e-06, | |
| "loss": 1.708, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 7.959385233935087e-06, | |
| "loss": 1.7134, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 7.925244992976538e-06, | |
| "loss": 1.7252, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 7.890896123569136e-06, | |
| "loss": 1.7254, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 7.856341075473963e-06, | |
| "loss": 1.7556, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 7.821582313156763e-06, | |
| "loss": 1.7285, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 7.786622315612182e-06, | |
| "loss": 1.7187, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 7.751463576186957e-06, | |
| "loss": 1.7248, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 7.716108602402094e-06, | |
| "loss": 1.6812, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 7.680559915774033e-06, | |
| "loss": 1.6401, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 7.644820051634813e-06, | |
| "loss": 1.6003, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 7.608891558951249e-06, | |
| "loss": 1.5809, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 7.572777000143145e-06, | |
| "loss": 1.5726, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 7.536478950900537e-06, | |
| "loss": 1.5202, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 1.5497, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 7.463342749120014e-06, | |
| "loss": 1.5906, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 7.4265098126554065e-06, | |
| "loss": 1.5538, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 7.389503817530905e-06, | |
| "loss": 1.5802, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 7.352327403013779e-06, | |
| "loss": 1.4782, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 7.314983220525604e-06, | |
| "loss": 1.4827, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 7.27747393345317e-06, | |
| "loss": 1.4746, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 7.239802216958522e-06, | |
| "loss": 1.4494, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 7.201970757788172e-06, | |
| "loss": 1.453, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 7.163982254081475e-06, | |
| "loss": 1.4673, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 7.125839415178204e-06, | |
| "loss": 1.4538, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 7.087544961425317e-06, | |
| "loss": 1.4482, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 7.049101623982938e-06, | |
| "loss": 1.4647, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 7.01051214462958e-06, | |
| "loss": 1.4409, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 6.971779275566593e-06, | |
| "loss": 1.4329, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 6.93290577922188e-06, | |
| "loss": 1.4008, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 6.893894428052881e-06, | |
| "loss": 1.4162, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 6.85474800434884e-06, | |
| "loss": 1.4537, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 6.815469300032374e-06, | |
| "loss": 1.4155, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 6.776061116460353e-06, | |
| "loss": 1.4145, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 6.736526264224101e-06, | |
| "loss": 1.404, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 6.696867562948962e-06, | |
| "loss": 1.4105, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 6.657087841093179e-06, | |
| "loss": 1.4168, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 6.617189935746191e-06, | |
| "loss": 1.422, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 6.5771766924262795e-06, | |
| "loss": 1.3664, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 6.537050964877626e-06, | |
| "loss": 1.4378, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 6.496815614866792e-06, | |
| "loss": 1.4579, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 6.456473511978607e-06, | |
| "loss": 1.4332, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 6.41602753341152e-06, | |
| "loss": 1.4323, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 6.375480563772391e-06, | |
| "loss": 1.4353, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 6.334835494870759e-06, | |
| "loss": 1.4135, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 6.294095225512604e-06, | |
| "loss": 1.4072, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 6.2532626612936035e-06, | |
| "loss": 1.4231, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 6.2123407143919e-06, | |
| "loss": 1.43, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 6.171332303360411e-06, | |
| "loss": 1.4465, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 6.130240352918675e-06, | |
| "loss": 1.4003, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 6.089067793744258e-06, | |
| "loss": 1.422, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 6.047817562263744e-06, | |
| "loss": 1.3963, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 6.006492600443301e-06, | |
| "loss": 1.3816, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 5.965095855578869e-06, | |
| "loss": 1.3685, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 5.923630280085948e-06, | |
| "loss": 1.4241, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 5.882098831289044e-06, | |
| "loss": 1.3942, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 5.840504471210742e-06, | |
| "loss": 1.3724, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 5.798850166360461e-06, | |
| "loss": 1.44, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 5.757138887522884e-06, | |
| "loss": 1.4123, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 5.715373609546071e-06, | |
| "loss": 1.4263, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 5.673557311129306e-06, | |
| "loss": 1.4496, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 5.631692974610647e-06, | |
| "loss": 1.4563, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 5.5897835857542315e-06, | |
| "loss": 1.4167, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 5.547832133537327e-06, | |
| "loss": 1.4244, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 5.505841609937162e-06, | |
| "loss": 1.4246, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 5.463815009717533e-06, | |
| "loss": 1.411, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 5.421755330215223e-06, | |
| "loss": 1.4305, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 5.379665571126232e-06, | |
| "loss": 1.4052, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 5.337548734291827e-06, | |
| "loss": 1.4724, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 5.2954078234844675e-06, | |
| "loss": 1.3773, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 5.253245844193564e-06, | |
| "loss": 1.405, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 5.211065803411135e-06, | |
| "loss": 1.4034, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 5.168870709417342e-06, | |
| "loss": 1.4115, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 5.12666357156594e-06, | |
| "loss": 1.4241, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 5.084447400069656e-06, | |
| "loss": 1.4277, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 5.042225205785492e-06, | |
| "loss": 1.417, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 5e-06, | |
| "loss": 1.4028, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 4.957774794214509e-06, | |
| "loss": 1.4323, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 4.915552599930345e-06, | |
| "loss": 1.3995, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 4.873336428434062e-06, | |
| "loss": 1.3889, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 4.83112929058266e-06, | |
| "loss": 1.4481, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 4.788934196588867e-06, | |
| "loss": 1.4132, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 4.746754155806437e-06, | |
| "loss": 1.4706, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 4.704592176515534e-06, | |
| "loss": 1.4062, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 4.662451265708174e-06, | |
| "loss": 1.3887, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 4.62033442887377e-06, | |
| "loss": 1.4452, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 4.5782446697847775e-06, | |
| "loss": 1.4136, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 4.5361849902824675e-06, | |
| "loss": 1.4201, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 4.49415839006284e-06, | |
| "loss": 1.4218, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 4.452167866462675e-06, | |
| "loss": 1.3894, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 4.410216414245771e-06, | |
| "loss": 1.3553, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 4.368307025389355e-06, | |
| "loss": 1.4651, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 4.326442688870697e-06, | |
| "loss": 1.4534, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 4.284626390453931e-06, | |
| "loss": 1.4167, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 4.2428611124771184e-06, | |
| "loss": 1.4228, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 4.20114983363954e-06, | |
| "loss": 1.4128, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 4.15949552878926e-06, | |
| "loss": 1.4264, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 4.11790116871096e-06, | |
| "loss": 1.3946, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 4.076369719914055e-06, | |
| "loss": 1.4421, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 4.0349041444211345e-06, | |
| "loss": 1.4095, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 3.993507399556699e-06, | |
| "loss": 1.4119, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 3.952182437736257e-06, | |
| "loss": 1.447, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 3.910932206255742e-06, | |
| "loss": 1.4229, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 3.869759647081326e-06, | |
| "loss": 1.4475, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 3.8286676966395895e-06, | |
| "loss": 1.4489, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 3.7876592856081e-06, | |
| "loss": 1.5203, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 3.7467373387063973e-06, | |
| "loss": 1.4133, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 3.705904774487396e-06, | |
| "loss": 1.4024, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 3.6651645051292415e-06, | |
| "loss": 1.4147, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 3.62451943622761e-06, | |
| "loss": 1.4528, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 3.58397246658848e-06, | |
| "loss": 1.4602, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 3.5435264880213943e-06, | |
| "loss": 1.4765, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 3.5031843851332105e-06, | |
| "loss": 1.4762, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 3.462949035122376e-06, | |
| "loss": 1.4487, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 3.4228233075737225e-06, | |
| "loss": 1.4065, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 3.3828100642538097e-06, | |
| "loss": 1.4639, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 3.3429121589068213e-06, | |
| "loss": 1.468, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 3.3031324370510397e-06, | |
| "loss": 1.4578, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 3.2634737357758994e-06, | |
| "loss": 1.4218, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 3.2239388835396492e-06, | |
| "loss": 1.4411, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 3.1845306999676274e-06, | |
| "loss": 1.3912, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 3.145251995651162e-06, | |
| "loss": 1.4242, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 3.10610557194712e-06, | |
| "loss": 1.407, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 3.0670942207781205e-06, | |
| "loss": 1.4504, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 3.0282207244334084e-06, | |
| "loss": 1.4735, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 2.989487855370421e-06, | |
| "loss": 1.4596, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 2.950898376017064e-06, | |
| "loss": 1.4357, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 2.912455038574686e-06, | |
| "loss": 1.4556, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 2.874160584821798e-06, | |
| "loss": 1.4569, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 2.8360177459185267e-06, | |
| "loss": 1.3932, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 2.7980292422118282e-06, | |
| "loss": 1.376, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 2.760197783041477e-06, | |
| "loss": 1.4691, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 2.72252606654683e-06, | |
| "loss": 1.4805, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 2.6850167794743966e-06, | |
| "loss": 1.4235, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 2.6476725969862227e-06, | |
| "loss": 1.4494, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 2.6104961824690967e-06, | |
| "loss": 1.4556, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 2.573490187344596e-06, | |
| "loss": 1.3941, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 2.5366572508799884e-06, | |
| "loss": 1.3611, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 2.5000000000000015e-06, | |
| "loss": 1.3544, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 2.4635210490994648e-06, | |
| "loss": 1.2959, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 2.4272229998568576e-06, | |
| "loss": 1.3098, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 2.3911084410487534e-06, | |
| "loss": 1.2575, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 2.3551799483651894e-06, | |
| "loss": 1.2199, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 2.319440084225969e-06, | |
| "loss": 1.2828, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 2.283891397597908e-06, | |
| "loss": 1.2908, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 2.2485364238130435e-06, | |
| "loss": 1.2341, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 2.2133776843878185e-06, | |
| "loss": 1.2635, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 2.1784176868432376e-06, | |
| "loss": 1.2174, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 2.1436589245260375e-06, | |
| "loss": 1.1737, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 2.109103876430864e-06, | |
| "loss": 1.2123, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 2.074755007023461e-06, | |
| "loss": 1.145, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 2.040614766064913e-06, | |
| "loss": 1.2068, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 2.0066855884369246e-06, | |
| "loss": 1.1631, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 1.9729698939681645e-06, | |
| "loss": 1.1414, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 1.9394700872616856e-06, | |
| "loss": 1.1388, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 1.9061885575234273e-06, | |
| "loss": 1.1535, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 1.8731276783918162e-06, | |
| "loss": 1.1254, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 1.8402898077684806e-06, | |
| "loss": 1.0936, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 1.8076772876500831e-06, | |
| "loss": 1.1191, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 1.775292443961291e-06, | |
| "loss": 1.122, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 1.74313758638889e-06, | |
| "loss": 1.1, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 1.711215008217057e-06, | |
| "loss": 1.1018, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 1.6795269861638041e-06, | |
| "loss": 1.1004, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 1.648075780218607e-06, | |
| "loss": 1.1071, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 1.6168636334812126e-06, | |
| "loss": 1.1005, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 1.585892772001671e-06, | |
| "loss": 1.1549, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 1.555165404621567e-06, | |
| "loss": 1.0458, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 1.5246837228164907e-06, | |
| "loss": 1.1283, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 1.4944499005397372e-06, | |
| "loss": 1.1476, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 1.4644660940672628e-06, | |
| "loss": 1.1383, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 1.434734441843899e-06, | |
| "loss": 1.1161, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 1.4052570643308377e-06, | |
| "loss": 1.1763, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 1.3760360638544012e-06, | |
| "loss": 1.1215, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 1.3470735244561028e-06, | |
| "loss": 1.0899, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 1.3183715117440143e-06, | |
| "loss": 1.1554, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 1.2899320727454472e-06, | |
| "loss": 1.1121, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 1.2617572357609565e-06, | |
| "loss": 1.1645, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 1.2338490102196827e-06, | |
| "loss": 1.1297, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 1.2062093865360458e-06, | |
| "loss": 1.0955, | |
| "step": 300 | |
| } | |
| ], | |
| "logging_steps": 1.0, | |
| "max_steps": 384, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 100, | |
| "total_flos": 0.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |