| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 6.0, | |
| "eval_steps": 500, | |
| "global_step": 210, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 7.142857142857143e-06, | |
| "loss": 2.0436, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.4285714285714285e-05, | |
| "loss": 2.0747, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 2.1428571428571428e-05, | |
| "loss": 1.9406, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 2.857142857142857e-05, | |
| "loss": 2.2945, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 3.571428571428572e-05, | |
| "loss": 2.0633, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.2857142857142856e-05, | |
| "loss": 1.9623, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 5e-05, | |
| "loss": 2.0427, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.9997006294917125e-05, | |
| "loss": 1.687, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.9988025896650096e-05, | |
| "loss": 2.1023, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.997306095597203e-05, | |
| "loss": 2.0172, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.9952115056932445e-05, | |
| "loss": 1.9736, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.99251932159989e-05, | |
| "loss": 2.0402, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.989230188085556e-05, | |
| "loss": 1.9319, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.985344892885899e-05, | |
| "loss": 1.9959, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 4.980864366515159e-05, | |
| "loss": 2.0264, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.975789682043301e-05, | |
| "loss": 1.9565, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.970122054839022e-05, | |
| "loss": 1.7632, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.963862842278669e-05, | |
| "loss": 1.9154, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.957013543421161e-05, | |
| "loss": 1.9812, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.9495757986489614e-05, | |
| "loss": 1.8412, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.941551389275217e-05, | |
| "loss": 1.7311, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 4.932942237117137e-05, | |
| "loss": 1.8453, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 4.9237504040357286e-05, | |
| "loss": 2.0415, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.913978091441985e-05, | |
| "loss": 1.9249, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 4.903627639769656e-05, | |
| "loss": 1.8312, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 4.8927015279147245e-05, | |
| "loss": 1.9261, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 4.881202372641719e-05, | |
| "loss": 1.9458, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 4.869132927957007e-05, | |
| "loss": 1.8691, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 4.856496084449218e-05, | |
| "loss": 1.7991, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 4.8432948685969646e-05, | |
| "loss": 1.8221, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 4.829532442044008e-05, | |
| "loss": 1.9247, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 4.8152121008420524e-05, | |
| "loss": 1.8928, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 4.8003372746613585e-05, | |
| "loss": 1.9307, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 4.784911525969345e-05, | |
| "loss": 1.8312, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.768938549177393e-05, | |
| "loss": 1.975, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 4.752422169756048e-05, | |
| "loss": 1.9311, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 4.735366343318832e-05, | |
| "loss": 1.7779, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 4.7177751546748885e-05, | |
| "loss": 1.8819, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 4.6996528168506864e-05, | |
| "loss": 1.9532, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 4.6810036700810143e-05, | |
| "loss": 1.8492, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 4.6618321807695095e-05, | |
| "loss": 2.0124, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 4.642142940418973e-05, | |
| "loss": 1.7269, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 4.621940664531718e-05, | |
| "loss": 1.9515, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 4.601230191480224e-05, | |
| "loss": 1.8711, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 4.580016481348367e-05, | |
| "loss": 1.94, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 4.558304614743496e-05, | |
| "loss": 1.8501, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 4.536099791579643e-05, | |
| "loss": 1.7961, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 4.5134073298321655e-05, | |
| "loss": 1.8661, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 4.4902326642641095e-05, | |
| "loss": 1.7798, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 4.466581345124604e-05, | |
| "loss": 1.851, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 4.442459036819595e-05, | |
| "loss": 1.8324, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 4.417871516555241e-05, | |
| "loss": 1.7448, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 4.392824672954294e-05, | |
| "loss": 1.7811, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 4.3673245046457924e-05, | |
| "loss": 1.8318, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 4.341377118828415e-05, | |
| "loss": 1.8495, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 4.3149887298078276e-05, | |
| "loss": 1.8712, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 4.288165657508376e-05, | |
| "loss": 1.9371, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 4.2609143259594906e-05, | |
| "loss": 1.9462, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 4.233241261757155e-05, | |
| "loss": 2.0101, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 4.2051530925008053e-05, | |
| "loss": 1.8443, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 4.1766565452060466e-05, | |
| "loss": 1.9715, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 4.147758444693557e-05, | |
| "loss": 1.8938, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 4.118465711954569e-05, | |
| "loss": 1.9364, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 4.088785362493314e-05, | |
| "loss": 1.7777, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 4.058724504646834e-05, | |
| "loss": 1.6773, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 4.028290337882565e-05, | |
| "loss": 1.7501, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 3.997490151074085e-05, | |
| "loss": 1.8379, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 3.9663313207554574e-05, | |
| "loss": 1.8593, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 3.93482130935458e-05, | |
| "loss": 1.8162, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 3.902967663405956e-05, | |
| "loss": 1.8169, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 3.870778011743327e-05, | |
| "loss": 1.8185, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 3.8382600636725984e-05, | |
| "loss": 1.8404, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 3.805421607125482e-05, | |
| "loss": 1.6879, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 3.772270506794323e-05, | |
| "loss": 1.765, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 3.738814702248524e-05, | |
| "loss": 1.81, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 3.705062206033055e-05, | |
| "loss": 1.8509, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 3.671021101749476e-05, | |
| "loss": 1.7403, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 3.636699542119939e-05, | |
| "loss": 1.9653, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 3.602105747034646e-05, | |
| "loss": 1.7267, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 3.5672480015832116e-05, | |
| "loss": 1.8532, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 3.532134654070415e-05, | |
| "loss": 1.7135, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 3.496774114016809e-05, | |
| "loss": 1.862, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 3.461174850144674e-05, | |
| "loss": 1.8093, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 3.425345388349786e-05, | |
| "loss": 1.7221, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 3.3892943096594966e-05, | |
| "loss": 1.7018, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 3.353030248177606e-05, | |
| "loss": 1.7583, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 3.3165618890165304e-05, | |
| "loss": 1.8001, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 3.279897966217245e-05, | |
| "loss": 1.908, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 3.243047260657511e-05, | |
| "loss": 1.6914, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 3.2060185979488925e-05, | |
| "loss": 1.5934, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 3.168820846323053e-05, | |
| "loss": 1.7841, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 3.131462914507838e-05, | |
| "loss": 1.8182, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 3.093953749593678e-05, | |
| "loss": 1.8664, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 3.056302334890786e-05, | |
| "loss": 1.8402, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 3.0185176877776877e-05, | |
| "loss": 1.807, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 2.9806088575415925e-05, | |
| "loss": 1.7117, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 2.9425849232111208e-05, | |
| "loss": 1.7541, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 2.9044549913819124e-05, | |
| "loss": 1.7826, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 2.8662281940356235e-05, | |
| "loss": 1.8804, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 2.827913686352856e-05, | |
| "loss": 1.7024, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 2.7895206445205223e-05, | |
| "loss": 1.8257, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 2.7510582635341814e-05, | |
| "loss": 1.8938, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 2.7125357549958684e-05, | |
| "loss": 1.795, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 2.6739623449079532e-05, | |
| "loss": 1.9122, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 2.635347271463544e-05, | |
| "loss": 1.8717, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 2.5966997828339722e-05, | |
| "loss": 1.8828, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 2.5580291349538895e-05, | |
| "loss": 1.6019, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 2.5193445893045052e-05, | |
| "loss": 1.8392, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 2.480655410695495e-05, | |
| "loss": 1.805, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 2.4419708650461108e-05, | |
| "loss": 1.8106, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 2.403300217166028e-05, | |
| "loss": 1.7194, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 2.3646527285364565e-05, | |
| "loss": 1.7633, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "learning_rate": 2.326037655092047e-05, | |
| "loss": 1.7865, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 3.26, | |
| "learning_rate": 2.287464245004132e-05, | |
| "loss": 1.7708, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 3.29, | |
| "learning_rate": 2.2489417364658192e-05, | |
| "loss": 1.7652, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 3.31, | |
| "learning_rate": 2.210479355479478e-05, | |
| "loss": 1.9105, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 3.34, | |
| "learning_rate": 2.1720863136471444e-05, | |
| "loss": 1.8936, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 3.37, | |
| "learning_rate": 2.133771805964377e-05, | |
| "loss": 1.7525, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 3.4, | |
| "learning_rate": 2.0955450086180882e-05, | |
| "loss": 1.8283, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 3.43, | |
| "learning_rate": 2.0574150767888794e-05, | |
| "loss": 1.663, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 3.46, | |
| "learning_rate": 2.019391142458408e-05, | |
| "loss": 1.705, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 3.49, | |
| "learning_rate": 1.9814823122223126e-05, | |
| "loss": 1.7988, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 3.51, | |
| "learning_rate": 1.9436976651092144e-05, | |
| "loss": 1.7041, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 3.54, | |
| "learning_rate": 1.906046250406323e-05, | |
| "loss": 1.7457, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 3.57, | |
| "learning_rate": 1.868537085492163e-05, | |
| "loss": 1.648, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 3.6, | |
| "learning_rate": 1.8311791536769483e-05, | |
| "loss": 1.7847, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 3.63, | |
| "learning_rate": 1.793981402051107e-05, | |
| "loss": 1.6724, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 3.66, | |
| "learning_rate": 1.7569527393424892e-05, | |
| "loss": 1.8516, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 3.69, | |
| "learning_rate": 1.7201020337827556e-05, | |
| "loss": 1.7543, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 3.71, | |
| "learning_rate": 1.6834381109834695e-05, | |
| "loss": 1.6904, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 3.74, | |
| "learning_rate": 1.646969751822394e-05, | |
| "loss": 1.701, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 3.77, | |
| "learning_rate": 1.6107056903405036e-05, | |
| "loss": 1.6526, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 3.8, | |
| "learning_rate": 1.574654611650214e-05, | |
| "loss": 1.8144, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 3.83, | |
| "learning_rate": 1.5388251498553265e-05, | |
| "loss": 1.6833, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 3.86, | |
| "learning_rate": 1.5032258859831915e-05, | |
| "loss": 1.7892, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 3.89, | |
| "learning_rate": 1.467865345929586e-05, | |
| "loss": 1.6905, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 3.91, | |
| "learning_rate": 1.4327519984167886e-05, | |
| "loss": 1.7325, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 3.94, | |
| "learning_rate": 1.397894252965355e-05, | |
| "loss": 1.7709, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "learning_rate": 1.3633004578800612e-05, | |
| "loss": 1.6669, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 1.328978898250525e-05, | |
| "loss": 1.8431, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 1.294937793966946e-05, | |
| "loss": 1.7394, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 1.261185297751477e-05, | |
| "loss": 1.7378, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 1.2277294932056782e-05, | |
| "loss": 1.7358, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 1.1945783928745186e-05, | |
| "loss": 1.8278, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 4.14, | |
| "learning_rate": 1.1617399363274022e-05, | |
| "loss": 1.7929, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 4.17, | |
| "learning_rate": 1.1292219882566727e-05, | |
| "loss": 1.6347, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 4.2, | |
| "learning_rate": 1.0970323365940444e-05, | |
| "loss": 1.7681, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 4.23, | |
| "learning_rate": 1.0651786906454193e-05, | |
| "loss": 1.7229, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 4.26, | |
| "learning_rate": 1.0336686792445425e-05, | |
| "loss": 1.5717, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 4.29, | |
| "learning_rate": 1.002509848925916e-05, | |
| "loss": 1.6373, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 4.31, | |
| "learning_rate": 9.717096621174354e-06, | |
| "loss": 1.8171, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 4.34, | |
| "learning_rate": 9.412754953531663e-06, | |
| "loss": 1.716, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 4.37, | |
| "learning_rate": 9.112146375066871e-06, | |
| "loss": 1.7158, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 4.4, | |
| "learning_rate": 8.815342880454311e-06, | |
| "loss": 1.7698, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 4.43, | |
| "learning_rate": 8.522415553064433e-06, | |
| "loss": 1.694, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 4.46, | |
| "learning_rate": 8.23343454793954e-06, | |
| "loss": 1.6029, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 4.49, | |
| "learning_rate": 7.948469074991954e-06, | |
| "loss": 1.7488, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 4.51, | |
| "learning_rate": 7.667587382428454e-06, | |
| "loss": 1.7441, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 4.54, | |
| "learning_rate": 7.390856740405092e-06, | |
| "loss": 1.6584, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 4.57, | |
| "learning_rate": 7.118343424916249e-06, | |
| "loss": 1.7259, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 4.6, | |
| "learning_rate": 6.8501127019217346e-06, | |
| "loss": 1.7216, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 4.63, | |
| "learning_rate": 6.586228811715853e-06, | |
| "loss": 1.6678, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 4.66, | |
| "learning_rate": 6.3267549535420855e-06, | |
| "loss": 1.7258, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 4.69, | |
| "learning_rate": 6.071753270457064e-06, | |
| "loss": 1.6373, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 4.71, | |
| "learning_rate": 5.821284834447585e-06, | |
| "loss": 1.6852, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 4.74, | |
| "learning_rate": 5.5754096318040485e-06, | |
| "loss": 1.8066, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 4.77, | |
| "learning_rate": 5.334186548753961e-06, | |
| "loss": 1.7095, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 4.8, | |
| "learning_rate": 5.097673357358907e-06, | |
| "loss": 1.7511, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 4.83, | |
| "learning_rate": 4.865926701678353e-06, | |
| "loss": 1.7555, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 4.86, | |
| "learning_rate": 4.639002084203575e-06, | |
| "loss": 1.5744, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 4.89, | |
| "learning_rate": 4.416953852565045e-06, | |
| "loss": 1.6493, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 4.91, | |
| "learning_rate": 4.199835186516332e-06, | |
| "loss": 1.7753, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 4.94, | |
| "learning_rate": 3.987698085197761e-06, | |
| "loss": 1.7544, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 4.97, | |
| "learning_rate": 3.7805933546828264e-06, | |
| "loss": 1.7049, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 3.578570595810274e-06, | |
| "loss": 1.6084, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 3.3816781923049046e-06, | |
| "loss": 1.8014, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 5.06, | |
| "learning_rate": 3.189963299189863e-06, | |
| "loss": 1.5702, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 5.09, | |
| "learning_rate": 3.0034718314931376e-06, | |
| "loss": 1.7683, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 5.11, | |
| "learning_rate": 2.8222484532511166e-06, | |
| "loss": 1.7868, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 5.14, | |
| "learning_rate": 2.6463365668116857e-06, | |
| "loss": 1.6444, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 5.17, | |
| "learning_rate": 2.475778302439524e-06, | |
| "loss": 1.5679, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 5.2, | |
| "learning_rate": 2.310614508226078e-06, | |
| "loss": 1.7885, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 5.23, | |
| "learning_rate": 2.150884740306558e-06, | |
| "loss": 1.3817, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 5.26, | |
| "learning_rate": 1.9966272533864183e-06, | |
| "loss": 1.6504, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 5.29, | |
| "learning_rate": 1.8478789915794769e-06, | |
| "loss": 1.7736, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 5.31, | |
| "learning_rate": 1.7046755795599222e-06, | |
| "loss": 1.5773, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 5.34, | |
| "learning_rate": 1.567051314030349e-06, | |
| "loss": 1.6765, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 5.37, | |
| "learning_rate": 1.4350391555078251e-06, | |
| "loss": 1.7248, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 5.4, | |
| "learning_rate": 1.3086707204299414e-06, | |
| "loss": 1.7631, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 5.43, | |
| "learning_rate": 1.1879762735828081e-06, | |
| "loss": 1.7554, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 5.46, | |
| "learning_rate": 1.0729847208527517e-06, | |
| "loss": 1.7072, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 5.49, | |
| "learning_rate": 9.6372360230344e-07, | |
| "loss": 1.6505, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 5.51, | |
| "learning_rate": 8.602190855801523e-07, | |
| "loss": 1.7094, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 5.54, | |
| "learning_rate": 7.624959596427145e-07, | |
| "loss": 1.8452, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 5.57, | |
| "learning_rate": 6.70577628828628e-07, | |
| "loss": 1.7456, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 5.6, | |
| "learning_rate": 5.844861072478336e-07, | |
| "loss": 1.7514, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 5.63, | |
| "learning_rate": 5.042420135103865e-07, | |
| "loss": 1.7269, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 5.66, | |
| "learning_rate": 4.298645657883904e-07, | |
| "loss": 1.7689, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 5.69, | |
| "learning_rate": 3.613715772133097e-07, | |
| "loss": 1.7866, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 5.71, | |
| "learning_rate": 2.987794516097875e-07, | |
| "loss": 1.8182, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 5.74, | |
| "learning_rate": 2.4210317956698815e-07, | |
| "loss": 1.2477, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 5.77, | |
| "learning_rate": 1.9135633484841097e-07, | |
| "loss": 1.2515, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 5.8, | |
| "learning_rate": 1.4655107114101007e-07, | |
| "loss": 1.1149, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 5.83, | |
| "learning_rate": 1.0769811914444205e-07, | |
| "loss": 1.1842, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 5.86, | |
| "learning_rate": 7.480678400109963e-08, | |
| "loss": 1.1749, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 5.89, | |
| "learning_rate": 4.7884943067555424e-08, | |
| "loss": 1.1625, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 5.91, | |
| "learning_rate": 2.6939044027973757e-08, | |
| "loss": 1.1117, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 5.94, | |
| "learning_rate": 1.1974103349909893e-08, | |
| "loss": 1.1215, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 5.97, | |
| "learning_rate": 2.993705082879328e-09, | |
| "loss": 1.1948, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "learning_rate": 0.0, | |
| "loss": 1.2632, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "step": 210, | |
| "total_flos": 0.0, | |
| "train_loss": 0.056318444297427224, | |
| "train_runtime": 1472.1497, | |
| "train_samples_per_second": 289.312, | |
| "train_steps_per_second": 0.143 | |
| } | |
| ], | |
| "logging_steps": 1.0, | |
| "max_steps": 210, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 6, | |
| "save_steps": 50, | |
| "total_flos": 0.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |