| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.06, | |
| "eval_steps": 1000, | |
| "global_step": 150, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4e-05, | |
| "loss": 2.242, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 8e-05, | |
| "loss": 2.0714, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00012, | |
| "loss": 1.8547, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00016, | |
| "loss": 2.1566, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9434, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00019862068965517243, | |
| "loss": 2.6236, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00019724137931034484, | |
| "loss": 2.6954, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00019586206896551723, | |
| "loss": 1.6838, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00019448275862068965, | |
| "loss": 2.2392, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0001931034482758621, | |
| "loss": 2.2795, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0001917241379310345, | |
| "loss": 1.9616, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0001903448275862069, | |
| "loss": 2.3299, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00018896551724137932, | |
| "loss": 2.041, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00018758620689655173, | |
| "loss": 2.0492, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00018620689655172415, | |
| "loss": 2.0077, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00018482758620689654, | |
| "loss": 1.7698, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00018344827586206896, | |
| "loss": 1.9129, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0001820689655172414, | |
| "loss": 2.794, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00018068965517241382, | |
| "loss": 1.75, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0001793103448275862, | |
| "loss": 2.1945, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00017793103448275862, | |
| "loss": 2.2451, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00017655172413793104, | |
| "loss": 2.0672, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00017517241379310346, | |
| "loss": 1.9468, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00017379310344827587, | |
| "loss": 2.2935, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00017241379310344826, | |
| "loss": 1.7946, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0001710344827586207, | |
| "loss": 2.2776, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00016965517241379312, | |
| "loss": 1.6909, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00016827586206896554, | |
| "loss": 1.7678, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00016689655172413793, | |
| "loss": 1.8813, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00016551724137931035, | |
| "loss": 1.8318, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00016413793103448276, | |
| "loss": 2.1097, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00016275862068965518, | |
| "loss": 1.7972, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0001613793103448276, | |
| "loss": 2.4994, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00016, | |
| "loss": 1.9247, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00015862068965517243, | |
| "loss": 2.0211, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00015724137931034485, | |
| "loss": 1.9272, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00015586206896551724, | |
| "loss": 1.841, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00015448275862068965, | |
| "loss": 1.9257, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00015310344827586207, | |
| "loss": 2.5944, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00015172413793103449, | |
| "loss": 1.7023, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001503448275862069, | |
| "loss": 1.6184, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00014896551724137932, | |
| "loss": 1.8313, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00014758620689655174, | |
| "loss": 2.1766, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00014620689655172415, | |
| "loss": 1.8785, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00014482758620689657, | |
| "loss": 1.943, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00014344827586206896, | |
| "loss": 1.7249, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00014206896551724138, | |
| "loss": 2.2218, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001406896551724138, | |
| "loss": 1.7874, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001393103448275862, | |
| "loss": 2.0419, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00013793103448275863, | |
| "loss": 1.6703, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00013655172413793104, | |
| "loss": 2.1259, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00013517241379310346, | |
| "loss": 1.9301, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00013379310344827588, | |
| "loss": 1.7461, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001324137931034483, | |
| "loss": 1.6138, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00013103448275862068, | |
| "loss": 1.7463, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001296551724137931, | |
| "loss": 2.064, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00012827586206896552, | |
| "loss": 1.8975, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00012689655172413793, | |
| "loss": 1.8061, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00012551724137931035, | |
| "loss": 2.108, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00012413793103448277, | |
| "loss": 1.9021, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00012275862068965518, | |
| "loss": 2.0424, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00012137931034482759, | |
| "loss": 1.8546, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00012, | |
| "loss": 1.8161, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001186206896551724, | |
| "loss": 1.5075, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00011724137931034482, | |
| "loss": 1.5969, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00011586206896551725, | |
| "loss": 2.2371, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00011448275862068967, | |
| "loss": 1.7303, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00011310344827586207, | |
| "loss": 1.697, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00011172413793103449, | |
| "loss": 1.8959, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001103448275862069, | |
| "loss": 1.8423, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00010896551724137931, | |
| "loss": 2.3673, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00010758620689655173, | |
| "loss": 2.0706, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00010620689655172413, | |
| "loss": 1.5911, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00010482758620689656, | |
| "loss": 1.5971, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00010344827586206898, | |
| "loss": 1.8699, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001020689655172414, | |
| "loss": 2.0459, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001006896551724138, | |
| "loss": 2.0273, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.931034482758621e-05, | |
| "loss": 1.5643, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.793103448275862e-05, | |
| "loss": 1.6831, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.655172413793105e-05, | |
| "loss": 1.8277, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.517241379310345e-05, | |
| "loss": 1.7535, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.379310344827587e-05, | |
| "loss": 1.9604, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.241379310344827e-05, | |
| "loss": 1.5079, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.10344827586207e-05, | |
| "loss": 2.251, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 8.96551724137931e-05, | |
| "loss": 1.9083, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 8.827586206896552e-05, | |
| "loss": 1.8386, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 8.689655172413794e-05, | |
| "loss": 1.617, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 8.551724137931035e-05, | |
| "loss": 2.0229, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 8.413793103448277e-05, | |
| "loss": 2.0117, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 8.275862068965517e-05, | |
| "loss": 2.0882, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 8.137931034482759e-05, | |
| "loss": 1.7453, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 8e-05, | |
| "loss": 1.225, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 7.862068965517242e-05, | |
| "loss": 1.8628, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 7.724137931034483e-05, | |
| "loss": 2.3614, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 7.586206896551724e-05, | |
| "loss": 1.6226, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 7.448275862068966e-05, | |
| "loss": 1.6608, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 7.310344827586208e-05, | |
| "loss": 1.893, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 7.172413793103448e-05, | |
| "loss": 1.774, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 7.03448275862069e-05, | |
| "loss": 1.9358, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 6.896551724137931e-05, | |
| "loss": 1.6985, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 6.758620689655173e-05, | |
| "loss": 1.7736, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 6.620689655172415e-05, | |
| "loss": 1.8252, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 6.482758620689655e-05, | |
| "loss": 1.8631, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 6.344827586206897e-05, | |
| "loss": 2.0603, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 6.206896551724138e-05, | |
| "loss": 2.0579, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 6.068965517241379e-05, | |
| "loss": 1.6665, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 5.93103448275862e-05, | |
| "loss": 1.7799, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 5.7931034482758627e-05, | |
| "loss": 1.716, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 5.6551724137931037e-05, | |
| "loss": 1.9681, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 5.517241379310345e-05, | |
| "loss": 1.8138, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 5.379310344827586e-05, | |
| "loss": 1.5202, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 5.241379310344828e-05, | |
| "loss": 1.7804, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 5.10344827586207e-05, | |
| "loss": 2.1902, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.9655172413793107e-05, | |
| "loss": 1.9234, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.827586206896552e-05, | |
| "loss": 2.0297, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.689655172413793e-05, | |
| "loss": 2.0227, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.551724137931035e-05, | |
| "loss": 1.858, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.413793103448276e-05, | |
| "loss": 1.8684, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.275862068965518e-05, | |
| "loss": 1.7844, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.1379310344827587e-05, | |
| "loss": 1.6117, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4e-05, | |
| "loss": 1.5404, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 3.862068965517241e-05, | |
| "loss": 1.8653, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 3.724137931034483e-05, | |
| "loss": 1.5691, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 3.586206896551724e-05, | |
| "loss": 1.7276, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 3.4482758620689657e-05, | |
| "loss": 2.044, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 3.310344827586207e-05, | |
| "loss": 1.9005, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 3.172413793103448e-05, | |
| "loss": 1.9146, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 3.0344827586206897e-05, | |
| "loss": 1.3441, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.8965517241379313e-05, | |
| "loss": 1.8642, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.7586206896551727e-05, | |
| "loss": 2.4694, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.620689655172414e-05, | |
| "loss": 2.0665, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.4827586206896553e-05, | |
| "loss": 1.202, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.3448275862068967e-05, | |
| "loss": 1.397, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.206896551724138e-05, | |
| "loss": 1.6153, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.0689655172413793e-05, | |
| "loss": 2.0912, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.9310344827586207e-05, | |
| "loss": 1.9446, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.793103448275862e-05, | |
| "loss": 2.0358, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.6551724137931037e-05, | |
| "loss": 1.4979, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.5172413793103448e-05, | |
| "loss": 1.8927, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.3793103448275863e-05, | |
| "loss": 1.8603, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.2413793103448277e-05, | |
| "loss": 1.7057, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.103448275862069e-05, | |
| "loss": 1.6926, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.655172413793103e-06, | |
| "loss": 1.6525, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 8.275862068965518e-06, | |
| "loss": 1.7455, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 6.896551724137932e-06, | |
| "loss": 1.775, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 5.517241379310345e-06, | |
| "loss": 1.6912, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.137931034482759e-06, | |
| "loss": 1.8147, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 2.7586206896551725e-06, | |
| "loss": 1.8314, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.3793103448275862e-06, | |
| "loss": 1.9115, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0, | |
| "loss": 1.6862, | |
| "step": 150 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 150, | |
| "num_train_epochs": 1, | |
| "save_steps": 10, | |
| "total_flos": 7574630234062848.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |