llama-3-8b-4-layer-glide / trainer_state.json
OamPatel's picture
Upload folder using huggingface_hub
74b483b verified
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.0,
"eval_steps": 100,
"global_step": 2000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 5.000000000000001e-07,
"loss": 2.7669,
"step": 1
},
{
"epoch": 0.0,
"learning_rate": 1.0000000000000002e-06,
"loss": 2.782,
"step": 2
},
{
"epoch": 0.0,
"learning_rate": 1.5e-06,
"loss": 2.7742,
"step": 3
},
{
"epoch": 0.0,
"learning_rate": 2.0000000000000003e-06,
"loss": 2.6943,
"step": 4
},
{
"epoch": 0.0,
"learning_rate": 2.5e-06,
"loss": 2.5422,
"step": 5
},
{
"epoch": 0.0,
"learning_rate": 3e-06,
"loss": 2.3945,
"step": 6
},
{
"epoch": 0.0,
"learning_rate": 3.5000000000000004e-06,
"loss": 2.3055,
"step": 7
},
{
"epoch": 0.0,
"learning_rate": 4.000000000000001e-06,
"loss": 2.2746,
"step": 8
},
{
"epoch": 0.0,
"learning_rate": 4.5e-06,
"loss": 2.1986,
"step": 9
},
{
"epoch": 0.01,
"learning_rate": 5e-06,
"loss": 2.1832,
"step": 10
},
{
"epoch": 0.01,
"learning_rate": 5.500000000000001e-06,
"loss": 2.122,
"step": 11
},
{
"epoch": 0.01,
"learning_rate": 6e-06,
"loss": 2.0978,
"step": 12
},
{
"epoch": 0.01,
"learning_rate": 6.5000000000000004e-06,
"loss": 2.1149,
"step": 13
},
{
"epoch": 0.01,
"learning_rate": 7.000000000000001e-06,
"loss": 2.0188,
"step": 14
},
{
"epoch": 0.01,
"learning_rate": 7.5e-06,
"loss": 1.9337,
"step": 15
},
{
"epoch": 0.01,
"learning_rate": 8.000000000000001e-06,
"loss": 1.9943,
"step": 16
},
{
"epoch": 0.01,
"learning_rate": 8.500000000000002e-06,
"loss": 1.9404,
"step": 17
},
{
"epoch": 0.01,
"learning_rate": 9e-06,
"loss": 1.9063,
"step": 18
},
{
"epoch": 0.01,
"learning_rate": 9.5e-06,
"loss": 1.9297,
"step": 19
},
{
"epoch": 0.01,
"learning_rate": 1e-05,
"loss": 1.9235,
"step": 20
},
{
"epoch": 0.01,
"learning_rate": 1.05e-05,
"loss": 1.7984,
"step": 21
},
{
"epoch": 0.01,
"learning_rate": 1.1000000000000001e-05,
"loss": 1.8552,
"step": 22
},
{
"epoch": 0.01,
"learning_rate": 1.1500000000000002e-05,
"loss": 1.8237,
"step": 23
},
{
"epoch": 0.01,
"learning_rate": 1.2e-05,
"loss": 1.7755,
"step": 24
},
{
"epoch": 0.01,
"learning_rate": 1.25e-05,
"loss": 1.7878,
"step": 25
},
{
"epoch": 0.01,
"learning_rate": 1.3000000000000001e-05,
"loss": 1.7853,
"step": 26
},
{
"epoch": 0.01,
"learning_rate": 1.3500000000000001e-05,
"loss": 1.7371,
"step": 27
},
{
"epoch": 0.01,
"learning_rate": 1.4000000000000001e-05,
"loss": 1.789,
"step": 28
},
{
"epoch": 0.01,
"learning_rate": 1.45e-05,
"loss": 1.7502,
"step": 29
},
{
"epoch": 0.01,
"learning_rate": 1.5e-05,
"loss": 1.7098,
"step": 30
},
{
"epoch": 0.02,
"learning_rate": 1.55e-05,
"loss": 1.7511,
"step": 31
},
{
"epoch": 0.02,
"learning_rate": 1.6000000000000003e-05,
"loss": 1.7727,
"step": 32
},
{
"epoch": 0.02,
"learning_rate": 1.65e-05,
"loss": 1.7315,
"step": 33
},
{
"epoch": 0.02,
"learning_rate": 1.7000000000000003e-05,
"loss": 1.7496,
"step": 34
},
{
"epoch": 0.02,
"learning_rate": 1.75e-05,
"loss": 1.6851,
"step": 35
},
{
"epoch": 0.02,
"learning_rate": 1.8e-05,
"loss": 1.6738,
"step": 36
},
{
"epoch": 0.02,
"learning_rate": 1.85e-05,
"loss": 1.6682,
"step": 37
},
{
"epoch": 0.02,
"learning_rate": 1.9e-05,
"loss": 1.6803,
"step": 38
},
{
"epoch": 0.02,
"learning_rate": 1.9500000000000003e-05,
"loss": 1.6743,
"step": 39
},
{
"epoch": 0.02,
"learning_rate": 2e-05,
"loss": 1.6752,
"step": 40
},
{
"epoch": 0.02,
"learning_rate": 2.05e-05,
"loss": 1.6464,
"step": 41
},
{
"epoch": 0.02,
"learning_rate": 2.1e-05,
"loss": 1.6593,
"step": 42
},
{
"epoch": 0.02,
"learning_rate": 2.15e-05,
"loss": 1.6329,
"step": 43
},
{
"epoch": 0.02,
"learning_rate": 2.2000000000000003e-05,
"loss": 1.6469,
"step": 44
},
{
"epoch": 0.02,
"learning_rate": 2.25e-05,
"loss": 1.6659,
"step": 45
},
{
"epoch": 0.02,
"learning_rate": 2.3000000000000003e-05,
"loss": 1.6389,
"step": 46
},
{
"epoch": 0.02,
"learning_rate": 2.35e-05,
"loss": 1.6583,
"step": 47
},
{
"epoch": 0.02,
"learning_rate": 2.4e-05,
"loss": 1.6079,
"step": 48
},
{
"epoch": 0.02,
"learning_rate": 2.45e-05,
"loss": 1.6551,
"step": 49
},
{
"epoch": 0.03,
"learning_rate": 2.5e-05,
"loss": 1.643,
"step": 50
},
{
"epoch": 0.03,
"learning_rate": 2.5500000000000003e-05,
"loss": 1.5879,
"step": 51
},
{
"epoch": 0.03,
"learning_rate": 2.6000000000000002e-05,
"loss": 1.681,
"step": 52
},
{
"epoch": 0.03,
"learning_rate": 2.6500000000000004e-05,
"loss": 1.5742,
"step": 53
},
{
"epoch": 0.03,
"learning_rate": 2.7000000000000002e-05,
"loss": 1.6519,
"step": 54
},
{
"epoch": 0.03,
"learning_rate": 2.7500000000000004e-05,
"loss": 1.5784,
"step": 55
},
{
"epoch": 0.03,
"learning_rate": 2.8000000000000003e-05,
"loss": 1.5889,
"step": 56
},
{
"epoch": 0.03,
"learning_rate": 2.8499999999999998e-05,
"loss": 1.5601,
"step": 57
},
{
"epoch": 0.03,
"learning_rate": 2.9e-05,
"loss": 1.586,
"step": 58
},
{
"epoch": 0.03,
"learning_rate": 2.95e-05,
"loss": 1.5837,
"step": 59
},
{
"epoch": 0.03,
"learning_rate": 3e-05,
"loss": 1.5894,
"step": 60
},
{
"epoch": 0.03,
"learning_rate": 3.05e-05,
"loss": 1.6306,
"step": 61
},
{
"epoch": 0.03,
"learning_rate": 3.1e-05,
"loss": 1.5699,
"step": 62
},
{
"epoch": 0.03,
"learning_rate": 3.15e-05,
"loss": 1.6232,
"step": 63
},
{
"epoch": 0.03,
"learning_rate": 3.2000000000000005e-05,
"loss": 1.6154,
"step": 64
},
{
"epoch": 0.03,
"learning_rate": 3.2500000000000004e-05,
"loss": 1.5821,
"step": 65
},
{
"epoch": 0.03,
"learning_rate": 3.3e-05,
"loss": 1.6217,
"step": 66
},
{
"epoch": 0.03,
"learning_rate": 3.35e-05,
"loss": 1.5406,
"step": 67
},
{
"epoch": 0.03,
"learning_rate": 3.4000000000000007e-05,
"loss": 1.5578,
"step": 68
},
{
"epoch": 0.03,
"learning_rate": 3.45e-05,
"loss": 1.5548,
"step": 69
},
{
"epoch": 0.04,
"learning_rate": 3.5e-05,
"loss": 1.6024,
"step": 70
},
{
"epoch": 0.04,
"learning_rate": 3.55e-05,
"loss": 1.5928,
"step": 71
},
{
"epoch": 0.04,
"learning_rate": 3.6e-05,
"loss": 1.6047,
"step": 72
},
{
"epoch": 0.04,
"learning_rate": 3.65e-05,
"loss": 1.5717,
"step": 73
},
{
"epoch": 0.04,
"learning_rate": 3.7e-05,
"loss": 1.5685,
"step": 74
},
{
"epoch": 0.04,
"learning_rate": 3.7500000000000003e-05,
"loss": 1.5319,
"step": 75
},
{
"epoch": 0.04,
"learning_rate": 3.8e-05,
"loss": 1.545,
"step": 76
},
{
"epoch": 0.04,
"learning_rate": 3.85e-05,
"loss": 1.5537,
"step": 77
},
{
"epoch": 0.04,
"learning_rate": 3.9000000000000006e-05,
"loss": 1.5679,
"step": 78
},
{
"epoch": 0.04,
"learning_rate": 3.9500000000000005e-05,
"loss": 1.5901,
"step": 79
},
{
"epoch": 0.04,
"learning_rate": 4e-05,
"loss": 1.5615,
"step": 80
},
{
"epoch": 0.04,
"learning_rate": 4.05e-05,
"loss": 1.5512,
"step": 81
},
{
"epoch": 0.04,
"learning_rate": 4.1e-05,
"loss": 1.5537,
"step": 82
},
{
"epoch": 0.04,
"learning_rate": 4.15e-05,
"loss": 1.5052,
"step": 83
},
{
"epoch": 0.04,
"learning_rate": 4.2e-05,
"loss": 1.5378,
"step": 84
},
{
"epoch": 0.04,
"learning_rate": 4.25e-05,
"loss": 1.5745,
"step": 85
},
{
"epoch": 0.04,
"learning_rate": 4.3e-05,
"loss": 1.5877,
"step": 86
},
{
"epoch": 0.04,
"learning_rate": 4.35e-05,
"loss": 1.5404,
"step": 87
},
{
"epoch": 0.04,
"learning_rate": 4.4000000000000006e-05,
"loss": 1.5483,
"step": 88
},
{
"epoch": 0.04,
"learning_rate": 4.4500000000000004e-05,
"loss": 1.5316,
"step": 89
},
{
"epoch": 0.04,
"learning_rate": 4.5e-05,
"loss": 1.5617,
"step": 90
},
{
"epoch": 0.05,
"learning_rate": 4.55e-05,
"loss": 1.5685,
"step": 91
},
{
"epoch": 0.05,
"learning_rate": 4.600000000000001e-05,
"loss": 1.5747,
"step": 92
},
{
"epoch": 0.05,
"learning_rate": 4.6500000000000005e-05,
"loss": 1.5253,
"step": 93
},
{
"epoch": 0.05,
"learning_rate": 4.7e-05,
"loss": 1.5176,
"step": 94
},
{
"epoch": 0.05,
"learning_rate": 4.75e-05,
"loss": 1.5684,
"step": 95
},
{
"epoch": 0.05,
"learning_rate": 4.8e-05,
"loss": 1.5148,
"step": 96
},
{
"epoch": 0.05,
"learning_rate": 4.85e-05,
"loss": 1.5389,
"step": 97
},
{
"epoch": 0.05,
"learning_rate": 4.9e-05,
"loss": 1.5272,
"step": 98
},
{
"epoch": 0.05,
"learning_rate": 4.9500000000000004e-05,
"loss": 1.5633,
"step": 99
},
{
"epoch": 0.05,
"learning_rate": 5e-05,
"loss": 1.519,
"step": 100
},
{
"epoch": 0.05,
"eval_acceptance_rate": 0.6372560858726501,
"eval_acceptance_rate_1": 0.6372109651565552,
"eval_acceptance_rate_2": 0.6359634399414062,
"eval_acceptance_rate_3": 0.6371628642082214,
"eval_acceptance_rate_4": 0.6373171806335449,
"eval_avg_generation_length": 215.4989013671875,
"eval_avg_input_length": 365.6648864746094,
"eval_expected_speedup": 1.9736412211350056,
"eval_expected_tokens": 2.467051526418757,
"eval_loss": 1.508664608001709,
"eval_max_generation_length": 424.7227783203125,
"eval_max_input_length": 630.4989624023438,
"eval_min_generation_length": 59.60786437988281,
"eval_min_input_length": 160.72378540039062,
"eval_pulsar_acceptance_rate": 0.36676288160468923,
"eval_runtime": 121.2026,
"eval_samples_per_second": 4.084,
"eval_steps_per_second": 0.132,
"step": 100
},
{
"epoch": 0.05,
"learning_rate": 4.9999965825477314e-05,
"loss": 1.4603,
"step": 101
},
{
"epoch": 0.05,
"learning_rate": 4.999986330200268e-05,
"loss": 1.5116,
"step": 102
},
{
"epoch": 0.05,
"learning_rate": 4.999969242985639e-05,
"loss": 1.5292,
"step": 103
},
{
"epoch": 0.05,
"learning_rate": 4.999945320950562e-05,
"loss": 1.5279,
"step": 104
},
{
"epoch": 0.05,
"learning_rate": 4.999914564160437e-05,
"loss": 1.4961,
"step": 105
},
{
"epoch": 0.05,
"learning_rate": 4.999876972699352e-05,
"loss": 1.5504,
"step": 106
},
{
"epoch": 0.05,
"learning_rate": 4.999832546670082e-05,
"loss": 1.5631,
"step": 107
},
{
"epoch": 0.05,
"learning_rate": 4.999781286194085e-05,
"loss": 1.4472,
"step": 108
},
{
"epoch": 0.05,
"learning_rate": 4.9997231914115064e-05,
"loss": 1.4779,
"step": 109
},
{
"epoch": 0.06,
"learning_rate": 4.9996582624811725e-05,
"loss": 1.5227,
"step": 110
},
{
"epoch": 0.06,
"learning_rate": 4.999586499580599e-05,
"loss": 1.5161,
"step": 111
},
{
"epoch": 0.06,
"learning_rate": 4.9995079029059824e-05,
"loss": 1.5485,
"step": 112
},
{
"epoch": 0.06,
"learning_rate": 4.999422472672202e-05,
"loss": 1.5273,
"step": 113
},
{
"epoch": 0.06,
"learning_rate": 4.999330209112822e-05,
"loss": 1.5809,
"step": 114
},
{
"epoch": 0.06,
"learning_rate": 4.9992311124800875e-05,
"loss": 1.523,
"step": 115
},
{
"epoch": 0.06,
"learning_rate": 4.999125183044924e-05,
"loss": 1.4913,
"step": 116
},
{
"epoch": 0.06,
"learning_rate": 4.999012421096938e-05,
"loss": 1.4649,
"step": 117
},
{
"epoch": 0.06,
"learning_rate": 4.998892826944418e-05,
"loss": 1.4986,
"step": 118
},
{
"epoch": 0.06,
"learning_rate": 4.998766400914329e-05,
"loss": 1.5137,
"step": 119
},
{
"epoch": 0.06,
"learning_rate": 4.9986331433523156e-05,
"loss": 1.4823,
"step": 120
},
{
"epoch": 0.06,
"learning_rate": 4.9984930546226975e-05,
"loss": 1.4663,
"step": 121
},
{
"epoch": 0.06,
"learning_rate": 4.9983461351084735e-05,
"loss": 1.4819,
"step": 122
},
{
"epoch": 0.06,
"learning_rate": 4.9981923852113145e-05,
"loss": 1.5535,
"step": 123
},
{
"epoch": 0.06,
"learning_rate": 4.9980318053515686e-05,
"loss": 1.4416,
"step": 124
},
{
"epoch": 0.06,
"learning_rate": 4.997864395968252e-05,
"loss": 1.4651,
"step": 125
},
{
"epoch": 0.06,
"learning_rate": 4.997690157519059e-05,
"loss": 1.4648,
"step": 126
},
{
"epoch": 0.06,
"learning_rate": 4.99750909048035e-05,
"loss": 1.5525,
"step": 127
},
{
"epoch": 0.06,
"learning_rate": 4.997321195347154e-05,
"loss": 1.501,
"step": 128
},
{
"epoch": 0.06,
"learning_rate": 4.99712647263317e-05,
"loss": 1.4397,
"step": 129
},
{
"epoch": 0.07,
"learning_rate": 4.996924922870762e-05,
"loss": 1.4644,
"step": 130
},
{
"epoch": 0.07,
"learning_rate": 4.99671654661096e-05,
"loss": 1.4249,
"step": 131
},
{
"epoch": 0.07,
"learning_rate": 4.996501344423456e-05,
"loss": 1.4381,
"step": 132
},
{
"epoch": 0.07,
"learning_rate": 4.996279316896606e-05,
"loss": 1.463,
"step": 133
},
{
"epoch": 0.07,
"learning_rate": 4.996050464637423e-05,
"loss": 1.4517,
"step": 134
},
{
"epoch": 0.07,
"learning_rate": 4.995814788271582e-05,
"loss": 1.5317,
"step": 135
},
{
"epoch": 0.07,
"learning_rate": 4.9955722884434114e-05,
"loss": 1.4306,
"step": 136
},
{
"epoch": 0.07,
"learning_rate": 4.995322965815898e-05,
"loss": 1.4642,
"step": 137
},
{
"epoch": 0.07,
"learning_rate": 4.995066821070679e-05,
"loss": 1.5041,
"step": 138
},
{
"epoch": 0.07,
"learning_rate": 4.9948038549080456e-05,
"loss": 1.4405,
"step": 139
},
{
"epoch": 0.07,
"learning_rate": 4.994534068046937e-05,
"loss": 1.4534,
"step": 140
},
{
"epoch": 0.07,
"learning_rate": 4.9942574612249394e-05,
"loss": 1.4645,
"step": 141
},
{
"epoch": 0.07,
"learning_rate": 4.9939740351982856e-05,
"loss": 1.4777,
"step": 142
},
{
"epoch": 0.07,
"learning_rate": 4.993683790741852e-05,
"loss": 1.487,
"step": 143
},
{
"epoch": 0.07,
"learning_rate": 4.993386728649156e-05,
"loss": 1.4415,
"step": 144
},
{
"epoch": 0.07,
"learning_rate": 4.9930828497323526e-05,
"loss": 1.406,
"step": 145
},
{
"epoch": 0.07,
"learning_rate": 4.9927721548222374e-05,
"loss": 1.4193,
"step": 146
},
{
"epoch": 0.07,
"learning_rate": 4.992454644768236e-05,
"loss": 1.4266,
"step": 147
},
{
"epoch": 0.07,
"learning_rate": 4.9921303204384104e-05,
"loss": 1.4054,
"step": 148
},
{
"epoch": 0.07,
"learning_rate": 4.991799182719451e-05,
"loss": 1.4332,
"step": 149
},
{
"epoch": 0.07,
"learning_rate": 4.991461232516675e-05,
"loss": 1.4127,
"step": 150
},
{
"epoch": 0.08,
"learning_rate": 4.991116470754025e-05,
"loss": 1.4296,
"step": 151
},
{
"epoch": 0.08,
"learning_rate": 4.990764898374067e-05,
"loss": 1.4264,
"step": 152
},
{
"epoch": 0.08,
"learning_rate": 4.990406516337987e-05,
"loss": 1.4147,
"step": 153
},
{
"epoch": 0.08,
"learning_rate": 4.9900413256255876e-05,
"loss": 1.4413,
"step": 154
},
{
"epoch": 0.08,
"learning_rate": 4.9896693272352846e-05,
"loss": 1.4793,
"step": 155
},
{
"epoch": 0.08,
"learning_rate": 4.98929052218411e-05,
"loss": 1.4607,
"step": 156
},
{
"epoch": 0.08,
"learning_rate": 4.9889049115077005e-05,
"loss": 1.4065,
"step": 157
},
{
"epoch": 0.08,
"learning_rate": 4.988512496260301e-05,
"loss": 1.4479,
"step": 158
},
{
"epoch": 0.08,
"learning_rate": 4.988113277514761e-05,
"loss": 1.396,
"step": 159
},
{
"epoch": 0.08,
"learning_rate": 4.9877072563625285e-05,
"loss": 1.4241,
"step": 160
},
{
"epoch": 0.08,
"learning_rate": 4.9872944339136503e-05,
"loss": 1.3647,
"step": 161
},
{
"epoch": 0.08,
"learning_rate": 4.986874811296767e-05,
"loss": 1.4595,
"step": 162
},
{
"epoch": 0.08,
"learning_rate": 4.9864483896591094e-05,
"loss": 1.4043,
"step": 163
},
{
"epoch": 0.08,
"learning_rate": 4.9860151701665004e-05,
"loss": 1.4215,
"step": 164
},
{
"epoch": 0.08,
"learning_rate": 4.985575154003345e-05,
"loss": 1.3932,
"step": 165
},
{
"epoch": 0.08,
"learning_rate": 4.985128342372629e-05,
"loss": 1.4363,
"step": 166
},
{
"epoch": 0.08,
"learning_rate": 4.984674736495919e-05,
"loss": 1.4503,
"step": 167
},
{
"epoch": 0.08,
"learning_rate": 4.984214337613357e-05,
"loss": 1.4012,
"step": 168
},
{
"epoch": 0.08,
"learning_rate": 4.983747146983656e-05,
"loss": 1.3571,
"step": 169
},
{
"epoch": 0.09,
"learning_rate": 4.9832731658840956e-05,
"loss": 1.3973,
"step": 170
},
{
"epoch": 0.09,
"learning_rate": 4.982792395610524e-05,
"loss": 1.4165,
"step": 171
},
{
"epoch": 0.09,
"learning_rate": 4.982304837477348e-05,
"loss": 1.3847,
"step": 172
},
{
"epoch": 0.09,
"learning_rate": 4.981810492817532e-05,
"loss": 1.426,
"step": 173
},
{
"epoch": 0.09,
"learning_rate": 4.981309362982598e-05,
"loss": 1.3925,
"step": 174
},
{
"epoch": 0.09,
"learning_rate": 4.980801449342613e-05,
"loss": 1.4092,
"step": 175
},
{
"epoch": 0.09,
"learning_rate": 4.980286753286195e-05,
"loss": 1.3513,
"step": 176
},
{
"epoch": 0.09,
"learning_rate": 4.9797652762205025e-05,
"loss": 1.4103,
"step": 177
},
{
"epoch": 0.09,
"learning_rate": 4.979237019571235e-05,
"loss": 1.4159,
"step": 178
},
{
"epoch": 0.09,
"learning_rate": 4.978701984782625e-05,
"loss": 1.4049,
"step": 179
},
{
"epoch": 0.09,
"learning_rate": 4.978160173317438e-05,
"loss": 1.3426,
"step": 180
},
{
"epoch": 0.09,
"learning_rate": 4.9776115866569654e-05,
"loss": 1.3792,
"step": 181
},
{
"epoch": 0.09,
"learning_rate": 4.9770562263010225e-05,
"loss": 1.3671,
"step": 182
},
{
"epoch": 0.09,
"learning_rate": 4.976494093767943e-05,
"loss": 1.4226,
"step": 183
},
{
"epoch": 0.09,
"learning_rate": 4.975925190594575e-05,
"loss": 1.3503,
"step": 184
},
{
"epoch": 0.09,
"learning_rate": 4.9753495183362796e-05,
"loss": 1.3956,
"step": 185
},
{
"epoch": 0.09,
"learning_rate": 4.974767078566922e-05,
"loss": 1.3911,
"step": 186
},
{
"epoch": 0.09,
"learning_rate": 4.9741778728788705e-05,
"loss": 1.3835,
"step": 187
},
{
"epoch": 0.09,
"learning_rate": 4.97358190288299e-05,
"loss": 1.4129,
"step": 188
},
{
"epoch": 0.09,
"learning_rate": 4.9729791702086414e-05,
"loss": 1.3966,
"step": 189
},
{
"epoch": 0.1,
"learning_rate": 4.972369676503672e-05,
"loss": 1.3682,
"step": 190
},
{
"epoch": 0.1,
"learning_rate": 4.971753423434413e-05,
"loss": 1.343,
"step": 191
},
{
"epoch": 0.1,
"learning_rate": 4.971130412685679e-05,
"loss": 1.3597,
"step": 192
},
{
"epoch": 0.1,
"learning_rate": 4.970500645960756e-05,
"loss": 1.396,
"step": 193
},
{
"epoch": 0.1,
"learning_rate": 4.9698641249814036e-05,
"loss": 1.3425,
"step": 194
},
{
"epoch": 0.1,
"learning_rate": 4.9692208514878444e-05,
"loss": 1.3531,
"step": 195
},
{
"epoch": 0.1,
"learning_rate": 4.968570827238764e-05,
"loss": 1.3967,
"step": 196
},
{
"epoch": 0.1,
"learning_rate": 4.967914054011305e-05,
"loss": 1.4169,
"step": 197
},
{
"epoch": 0.1,
"learning_rate": 4.967250533601059e-05,
"loss": 1.3789,
"step": 198
},
{
"epoch": 0.1,
"learning_rate": 4.966580267822065e-05,
"loss": 1.368,
"step": 199
},
{
"epoch": 0.1,
"learning_rate": 4.965903258506806e-05,
"loss": 1.3936,
"step": 200
},
{
"epoch": 0.1,
"eval_acceptance_rate": 0.6973184943199158,
"eval_acceptance_rate_1": 0.6964073777198792,
"eval_acceptance_rate_2": 0.696760356426239,
"eval_acceptance_rate_3": 0.6975700855255127,
"eval_acceptance_rate_4": 0.6972681879997253,
"eval_avg_generation_length": 215.4989013671875,
"eval_avg_input_length": 365.6648864746094,
"eval_expected_speedup": 2.2072695235703748,
"eval_expected_tokens": 2.759086904462969,
"eval_loss": 1.2887985706329346,
"eval_max_generation_length": 424.7227783203125,
"eval_max_input_length": 630.4989624023438,
"eval_min_generation_length": 59.60786437988281,
"eval_min_input_length": 160.72378540039062,
"eval_pulsar_acceptance_rate": 0.4397717261157423,
"eval_runtime": 114.4635,
"eval_samples_per_second": 4.325,
"eval_steps_per_second": 0.14,
"step": 200
},
{
"epoch": 0.1,
"learning_rate": 4.965219507506198e-05,
"loss": 1.3912,
"step": 201
},
{
"epoch": 0.1,
"learning_rate": 4.9645290166895906e-05,
"loss": 1.4015,
"step": 202
},
{
"epoch": 0.1,
"learning_rate": 4.9638317879447606e-05,
"loss": 1.3633,
"step": 203
},
{
"epoch": 0.1,
"learning_rate": 4.963127823177902e-05,
"loss": 1.3772,
"step": 204
},
{
"epoch": 0.1,
"learning_rate": 4.96241712431363e-05,
"loss": 1.357,
"step": 205
},
{
"epoch": 0.1,
"learning_rate": 4.9616996932949666e-05,
"loss": 1.2806,
"step": 206
},
{
"epoch": 0.1,
"learning_rate": 4.960975532083342e-05,
"loss": 1.3648,
"step": 207
},
{
"epoch": 0.1,
"learning_rate": 4.960244642658585e-05,
"loss": 1.3354,
"step": 208
},
{
"epoch": 0.1,
"learning_rate": 4.959507027018918e-05,
"loss": 1.3347,
"step": 209
},
{
"epoch": 0.1,
"learning_rate": 4.958762687180956e-05,
"loss": 1.344,
"step": 210
},
{
"epoch": 0.11,
"learning_rate": 4.958011625179695e-05,
"loss": 1.3561,
"step": 211
},
{
"epoch": 0.11,
"learning_rate": 4.9572538430685094e-05,
"loss": 1.3511,
"step": 212
},
{
"epoch": 0.11,
"learning_rate": 4.956489342919147e-05,
"loss": 1.3671,
"step": 213
},
{
"epoch": 0.11,
"learning_rate": 4.9557181268217227e-05,
"loss": 1.3954,
"step": 214
},
{
"epoch": 0.11,
"learning_rate": 4.95494019688471e-05,
"loss": 1.3535,
"step": 215
},
{
"epoch": 0.11,
"learning_rate": 4.95415555523494e-05,
"loss": 1.3675,
"step": 216
},
{
"epoch": 0.11,
"learning_rate": 4.953364204017595e-05,
"loss": 1.3354,
"step": 217
},
{
"epoch": 0.11,
"learning_rate": 4.952566145396197e-05,
"loss": 1.3595,
"step": 218
},
{
"epoch": 0.11,
"learning_rate": 4.951761381552609e-05,
"loss": 1.3392,
"step": 219
},
{
"epoch": 0.11,
"learning_rate": 4.9509499146870236e-05,
"loss": 1.3648,
"step": 220
},
{
"epoch": 0.11,
"learning_rate": 4.9501317470179606e-05,
"loss": 1.3735,
"step": 221
},
{
"epoch": 0.11,
"learning_rate": 4.94930688078226e-05,
"loss": 1.3218,
"step": 222
},
{
"epoch": 0.11,
"learning_rate": 4.948475318235073e-05,
"loss": 1.3222,
"step": 223
},
{
"epoch": 0.11,
"learning_rate": 4.9476370616498617e-05,
"loss": 1.3465,
"step": 224
},
{
"epoch": 0.11,
"learning_rate": 4.946792113318386e-05,
"loss": 1.3207,
"step": 225
},
{
"epoch": 0.11,
"learning_rate": 4.945940475550703e-05,
"loss": 1.3665,
"step": 226
},
{
"epoch": 0.11,
"learning_rate": 4.945082150675159e-05,
"loss": 1.3443,
"step": 227
},
{
"epoch": 0.11,
"learning_rate": 4.944217141038379e-05,
"loss": 1.3433,
"step": 228
},
{
"epoch": 0.11,
"learning_rate": 4.9433454490052675e-05,
"loss": 1.3194,
"step": 229
},
{
"epoch": 0.12,
"learning_rate": 4.9424670769589984e-05,
"loss": 1.3737,
"step": 230
},
{
"epoch": 0.12,
"learning_rate": 4.941582027301006e-05,
"loss": 1.3505,
"step": 231
},
{
"epoch": 0.12,
"learning_rate": 4.940690302450982e-05,
"loss": 1.2845,
"step": 232
},
{
"epoch": 0.12,
"learning_rate": 4.939791904846869e-05,
"loss": 1.3478,
"step": 233
},
{
"epoch": 0.12,
"learning_rate": 4.938886836944851e-05,
"loss": 1.3168,
"step": 234
},
{
"epoch": 0.12,
"learning_rate": 4.93797510121935e-05,
"loss": 1.2953,
"step": 235
},
{
"epoch": 0.12,
"learning_rate": 4.937056700163015e-05,
"loss": 1.3489,
"step": 236
},
{
"epoch": 0.12,
"learning_rate": 4.9361316362867215e-05,
"loss": 1.3853,
"step": 237
},
{
"epoch": 0.12,
"learning_rate": 4.935199912119558e-05,
"loss": 1.3054,
"step": 238
},
{
"epoch": 0.12,
"learning_rate": 4.934261530208823e-05,
"loss": 1.3143,
"step": 239
},
{
"epoch": 0.12,
"learning_rate": 4.933316493120015e-05,
"loss": 1.3086,
"step": 240
},
{
"epoch": 0.12,
"learning_rate": 4.9323648034368316e-05,
"loss": 1.3011,
"step": 241
},
{
"epoch": 0.12,
"learning_rate": 4.931406463761154e-05,
"loss": 1.2483,
"step": 242
},
{
"epoch": 0.12,
"learning_rate": 4.930441476713049e-05,
"loss": 1.302,
"step": 243
},
{
"epoch": 0.12,
"learning_rate": 4.929469844930753e-05,
"loss": 1.3263,
"step": 244
},
{
"epoch": 0.12,
"learning_rate": 4.9284915710706695e-05,
"loss": 1.387,
"step": 245
},
{
"epoch": 0.12,
"learning_rate": 4.9275066578073626e-05,
"loss": 1.3013,
"step": 246
},
{
"epoch": 0.12,
"learning_rate": 4.926515107833547e-05,
"loss": 1.3018,
"step": 247
},
{
"epoch": 0.12,
"learning_rate": 4.925516923860083e-05,
"loss": 1.331,
"step": 248
},
{
"epoch": 0.12,
"learning_rate": 4.9245121086159674e-05,
"loss": 1.3337,
"step": 249
},
{
"epoch": 0.12,
"learning_rate": 4.923500664848326e-05,
"loss": 1.2895,
"step": 250
},
{
"epoch": 0.13,
"learning_rate": 4.9224825953224084e-05,
"loss": 1.318,
"step": 251
},
{
"epoch": 0.13,
"learning_rate": 4.9214579028215776e-05,
"loss": 1.2952,
"step": 252
},
{
"epoch": 0.13,
"learning_rate": 4.920426590147304e-05,
"loss": 1.3644,
"step": 253
},
{
"epoch": 0.13,
"learning_rate": 4.919388660119156e-05,
"loss": 1.3444,
"step": 254
},
{
"epoch": 0.13,
"learning_rate": 4.918344115574796e-05,
"loss": 1.361,
"step": 255
},
{
"epoch": 0.13,
"learning_rate": 4.917292959369968e-05,
"loss": 1.274,
"step": 256
},
{
"epoch": 0.13,
"learning_rate": 4.9162351943784935e-05,
"loss": 1.3637,
"step": 257
},
{
"epoch": 0.13,
"learning_rate": 4.9151708234922603e-05,
"loss": 1.2705,
"step": 258
},
{
"epoch": 0.13,
"learning_rate": 4.91409984962122e-05,
"loss": 1.3022,
"step": 259
},
{
"epoch": 0.13,
"learning_rate": 4.913022275693372e-05,
"loss": 1.3317,
"step": 260
},
{
"epoch": 0.13,
"learning_rate": 4.9119381046547636e-05,
"loss": 1.3392,
"step": 261
},
{
"epoch": 0.13,
"learning_rate": 4.910847339469477e-05,
"loss": 1.3158,
"step": 262
},
{
"epoch": 0.13,
"learning_rate": 4.9097499831196216e-05,
"loss": 1.3178,
"step": 263
},
{
"epoch": 0.13,
"learning_rate": 4.908646038605329e-05,
"loss": 1.2786,
"step": 264
},
{
"epoch": 0.13,
"learning_rate": 4.907535508944741e-05,
"loss": 1.2851,
"step": 265
},
{
"epoch": 0.13,
"learning_rate": 4.906418397174002e-05,
"loss": 1.2989,
"step": 266
},
{
"epoch": 0.13,
"learning_rate": 4.905294706347255e-05,
"loss": 1.263,
"step": 267
},
{
"epoch": 0.13,
"learning_rate": 4.904164439536626e-05,
"loss": 1.3328,
"step": 268
},
{
"epoch": 0.13,
"learning_rate": 4.903027599832223e-05,
"loss": 1.2766,
"step": 269
},
{
"epoch": 0.14,
"learning_rate": 4.901884190342121e-05,
"loss": 1.3073,
"step": 270
},
{
"epoch": 0.14,
"learning_rate": 4.900734214192358e-05,
"loss": 1.3622,
"step": 271
},
{
"epoch": 0.14,
"learning_rate": 4.8995776745269254e-05,
"loss": 1.344,
"step": 272
},
{
"epoch": 0.14,
"learning_rate": 4.8984145745077584e-05,
"loss": 1.3522,
"step": 273
},
{
"epoch": 0.14,
"learning_rate": 4.8972449173147276e-05,
"loss": 1.2848,
"step": 274
},
{
"epoch": 0.14,
"learning_rate": 4.8960687061456324e-05,
"loss": 1.3402,
"step": 275
},
{
"epoch": 0.14,
"learning_rate": 4.8948859442161874e-05,
"loss": 1.3081,
"step": 276
},
{
"epoch": 0.14,
"learning_rate": 4.893696634760019e-05,
"loss": 1.2651,
"step": 277
},
{
"epoch": 0.14,
"learning_rate": 4.8925007810286546e-05,
"loss": 1.3102,
"step": 278
},
{
"epoch": 0.14,
"learning_rate": 4.891298386291513e-05,
"loss": 1.3975,
"step": 279
},
{
"epoch": 0.14,
"learning_rate": 4.8900894538358944e-05,
"loss": 1.3574,
"step": 280
},
{
"epoch": 0.14,
"learning_rate": 4.888873986966974e-05,
"loss": 1.2603,
"step": 281
},
{
"epoch": 0.14,
"learning_rate": 4.8876519890077926e-05,
"loss": 1.3324,
"step": 282
},
{
"epoch": 0.14,
"learning_rate": 4.8864234632992457e-05,
"loss": 1.3366,
"step": 283
},
{
"epoch": 0.14,
"learning_rate": 4.885188413200075e-05,
"loss": 1.3091,
"step": 284
},
{
"epoch": 0.14,
"learning_rate": 4.8839468420868606e-05,
"loss": 1.3026,
"step": 285
},
{
"epoch": 0.14,
"learning_rate": 4.8826987533540114e-05,
"loss": 1.2572,
"step": 286
},
{
"epoch": 0.14,
"learning_rate": 4.8814441504137534e-05,
"loss": 1.2946,
"step": 287
},
{
"epoch": 0.14,
"learning_rate": 4.880183036696123e-05,
"loss": 1.3107,
"step": 288
},
{
"epoch": 0.14,
"learning_rate": 4.878915415648957e-05,
"loss": 1.28,
"step": 289
},
{
"epoch": 0.14,
"learning_rate": 4.877641290737884e-05,
"loss": 1.2945,
"step": 290
},
{
"epoch": 0.15,
"learning_rate": 4.876360665446312e-05,
"loss": 1.2664,
"step": 291
},
{
"epoch": 0.15,
"learning_rate": 4.875073543275422e-05,
"loss": 1.2705,
"step": 292
},
{
"epoch": 0.15,
"learning_rate": 4.8737799277441566e-05,
"loss": 1.2677,
"step": 293
},
{
"epoch": 0.15,
"learning_rate": 4.872479822389211e-05,
"loss": 1.278,
"step": 294
},
{
"epoch": 0.15,
"learning_rate": 4.871173230765024e-05,
"loss": 1.3051,
"step": 295
},
{
"epoch": 0.15,
"learning_rate": 4.8698601564437675e-05,
"loss": 1.3449,
"step": 296
},
{
"epoch": 0.15,
"learning_rate": 4.868540603015335e-05,
"loss": 1.3078,
"step": 297
},
{
"epoch": 0.15,
"learning_rate": 4.8672145740873374e-05,
"loss": 1.2072,
"step": 298
},
{
"epoch": 0.15,
"learning_rate": 4.865882073285086e-05,
"loss": 1.3015,
"step": 299
},
{
"epoch": 0.15,
"learning_rate": 4.864543104251587e-05,
"loss": 1.2994,
"step": 300
},
{
"epoch": 0.15,
"eval_acceptance_rate": 0.7273669242858887,
"eval_acceptance_rate_1": 0.7265855073928833,
"eval_acceptance_rate_2": 0.7276649475097656,
"eval_acceptance_rate_3": 0.7265880703926086,
"eval_acceptance_rate_4": 0.7270616292953491,
"eval_avg_generation_length": 215.4989013671875,
"eval_avg_input_length": 365.6648864746094,
"eval_expected_speedup": 2.3369276109043398,
"eval_expected_tokens": 2.9211595136304247,
"eval_loss": 1.1850454807281494,
"eval_max_generation_length": 424.7227783203125,
"eval_max_input_length": 630.4989624023438,
"eval_min_generation_length": 59.60786437988281,
"eval_min_input_length": 160.72378540039062,
"eval_pulsar_acceptance_rate": 0.4802898784076063,
"eval_runtime": 123.3228,
"eval_samples_per_second": 4.014,
"eval_steps_per_second": 0.13,
"step": 300
},
{
"epoch": 0.15,
"learning_rate": 4.863197670647531e-05,
"loss": 1.2556,
"step": 301
},
{
"epoch": 0.15,
"learning_rate": 4.861845776151281e-05,
"loss": 1.282,
"step": 302
},
{
"epoch": 0.15,
"learning_rate": 4.860487424458867e-05,
"loss": 1.2981,
"step": 303
},
{
"epoch": 0.15,
"learning_rate": 4.8591226192839696e-05,
"loss": 1.2678,
"step": 304
},
{
"epoch": 0.15,
"learning_rate": 4.857751364357913e-05,
"loss": 1.2905,
"step": 305
},
{
"epoch": 0.15,
"learning_rate": 4.856373663429657e-05,
"loss": 1.3769,
"step": 306
},
{
"epoch": 0.15,
"learning_rate": 4.8549895202657844e-05,
"loss": 1.3233,
"step": 307
},
{
"epoch": 0.15,
"learning_rate": 4.853598938650487e-05,
"loss": 1.2517,
"step": 308
},
{
"epoch": 0.15,
"learning_rate": 4.852201922385564e-05,
"loss": 1.2923,
"step": 309
},
{
"epoch": 0.15,
"learning_rate": 4.850798475290403e-05,
"loss": 1.3197,
"step": 310
},
{
"epoch": 0.16,
"learning_rate": 4.8493886012019765e-05,
"loss": 1.3041,
"step": 311
},
{
"epoch": 0.16,
"learning_rate": 4.8479723039748246e-05,
"loss": 1.3085,
"step": 312
},
{
"epoch": 0.16,
"learning_rate": 4.846549587481052e-05,
"loss": 1.2549,
"step": 313
},
{
"epoch": 0.16,
"learning_rate": 4.845120455610309e-05,
"loss": 1.2561,
"step": 314
},
{
"epoch": 0.16,
"learning_rate": 4.8436849122697883e-05,
"loss": 1.2595,
"step": 315
},
{
"epoch": 0.16,
"learning_rate": 4.842242961384211e-05,
"loss": 1.1942,
"step": 316
},
{
"epoch": 0.16,
"learning_rate": 4.8407946068958165e-05,
"loss": 1.271,
"step": 317
},
{
"epoch": 0.16,
"learning_rate": 4.83933985276435e-05,
"loss": 1.3538,
"step": 318
},
{
"epoch": 0.16,
"learning_rate": 4.837878702967052e-05,
"loss": 1.2728,
"step": 319
},
{
"epoch": 0.16,
"learning_rate": 4.8364111614986527e-05,
"loss": 1.2275,
"step": 320
},
{
"epoch": 0.16,
"learning_rate": 4.834937232371353e-05,
"loss": 1.2514,
"step": 321
},
{
"epoch": 0.16,
"learning_rate": 4.83345691961482e-05,
"loss": 1.3102,
"step": 322
},
{
"epoch": 0.16,
"learning_rate": 4.831970227276171e-05,
"loss": 1.2467,
"step": 323
},
{
"epoch": 0.16,
"learning_rate": 4.830477159419966e-05,
"loss": 1.3083,
"step": 324
},
{
"epoch": 0.16,
"learning_rate": 4.8289777201281974e-05,
"loss": 1.2467,
"step": 325
},
{
"epoch": 0.16,
"learning_rate": 4.827471913500274e-05,
"loss": 1.2743,
"step": 326
},
{
"epoch": 0.16,
"learning_rate": 4.8259597436530125e-05,
"loss": 1.2776,
"step": 327
},
{
"epoch": 0.16,
"learning_rate": 4.8244412147206284e-05,
"loss": 1.3385,
"step": 328
},
{
"epoch": 0.16,
"learning_rate": 4.822916330854722e-05,
"loss": 1.2746,
"step": 329
},
{
"epoch": 0.17,
"learning_rate": 4.821385096224268e-05,
"loss": 1.2242,
"step": 330
},
{
"epoch": 0.17,
"learning_rate": 4.819847515015602e-05,
"loss": 1.2818,
"step": 331
},
{
"epoch": 0.17,
"learning_rate": 4.8183035914324136e-05,
"loss": 1.2727,
"step": 332
},
{
"epoch": 0.17,
"learning_rate": 4.81675332969573e-05,
"loss": 1.2286,
"step": 333
},
{
"epoch": 0.17,
"learning_rate": 4.815196734043909e-05,
"loss": 1.2432,
"step": 334
},
{
"epoch": 0.17,
"learning_rate": 4.8136338087326216e-05,
"loss": 1.275,
"step": 335
},
{
"epoch": 0.17,
"learning_rate": 4.812064558034847e-05,
"loss": 1.2502,
"step": 336
},
{
"epoch": 0.17,
"learning_rate": 4.810488986240858e-05,
"loss": 1.2694,
"step": 337
},
{
"epoch": 0.17,
"learning_rate": 4.8089070976582054e-05,
"loss": 1.2416,
"step": 338
},
{
"epoch": 0.17,
"learning_rate": 4.8073188966117126e-05,
"loss": 1.2345,
"step": 339
},
{
"epoch": 0.17,
"learning_rate": 4.805724387443462e-05,
"loss": 1.244,
"step": 340
},
{
"epoch": 0.17,
"learning_rate": 4.80412357451278e-05,
"loss": 1.2536,
"step": 341
},
{
"epoch": 0.17,
"learning_rate": 4.8025164621962284e-05,
"loss": 1.262,
"step": 342
},
{
"epoch": 0.17,
"learning_rate": 4.8009030548875896e-05,
"loss": 1.2491,
"step": 343
},
{
"epoch": 0.17,
"learning_rate": 4.799283356997859e-05,
"loss": 1.3092,
"step": 344
},
{
"epoch": 0.17,
"learning_rate": 4.797657372955228e-05,
"loss": 1.2588,
"step": 345
},
{
"epoch": 0.17,
"learning_rate": 4.796025107205075e-05,
"loss": 1.223,
"step": 346
},
{
"epoch": 0.17,
"learning_rate": 4.794386564209953e-05,
"loss": 1.2916,
"step": 347
},
{
"epoch": 0.17,
"learning_rate": 4.792741748449575e-05,
"loss": 1.2725,
"step": 348
},
{
"epoch": 0.17,
"learning_rate": 4.7910906644208054e-05,
"loss": 1.2933,
"step": 349
},
{
"epoch": 0.17,
"learning_rate": 4.789433316637644e-05,
"loss": 1.2601,
"step": 350
},
{
"epoch": 0.18,
"learning_rate": 4.787769709631217e-05,
"loss": 1.298,
"step": 351
},
{
"epoch": 0.18,
"learning_rate": 4.786099847949761e-05,
"loss": 1.2316,
"step": 352
},
{
"epoch": 0.18,
"learning_rate": 4.784423736158616e-05,
"loss": 1.2318,
"step": 353
},
{
"epoch": 0.18,
"learning_rate": 4.7827413788402077e-05,
"loss": 1.2691,
"step": 354
},
{
"epoch": 0.18,
"learning_rate": 4.781052780594034e-05,
"loss": 1.2232,
"step": 355
},
{
"epoch": 0.18,
"learning_rate": 4.779357946036661e-05,
"loss": 1.2884,
"step": 356
},
{
"epoch": 0.18,
"learning_rate": 4.777656879801701e-05,
"loss": 1.2336,
"step": 357
},
{
"epoch": 0.18,
"learning_rate": 4.775949586539803e-05,
"loss": 1.2298,
"step": 358
},
{
"epoch": 0.18,
"learning_rate": 4.774236070918643e-05,
"loss": 1.2701,
"step": 359
},
{
"epoch": 0.18,
"learning_rate": 4.7725163376229064e-05,
"loss": 1.2777,
"step": 360
},
{
"epoch": 0.18,
"learning_rate": 4.770790391354279e-05,
"loss": 1.2515,
"step": 361
},
{
"epoch": 0.18,
"learning_rate": 4.7690582368314304e-05,
"loss": 1.2428,
"step": 362
},
{
"epoch": 0.18,
"learning_rate": 4.7673198787900063e-05,
"loss": 1.2477,
"step": 363
},
{
"epoch": 0.18,
"learning_rate": 4.7655753219826114e-05,
"loss": 1.2741,
"step": 364
},
{
"epoch": 0.18,
"learning_rate": 4.763824571178798e-05,
"loss": 1.236,
"step": 365
},
{
"epoch": 0.18,
"learning_rate": 4.762067631165049e-05,
"loss": 1.243,
"step": 366
},
{
"epoch": 0.18,
"learning_rate": 4.760304506744774e-05,
"loss": 1.2307,
"step": 367
},
{
"epoch": 0.18,
"learning_rate": 4.7585352027382877e-05,
"loss": 1.2442,
"step": 368
},
{
"epoch": 0.18,
"learning_rate": 4.7567597239827974e-05,
"loss": 1.221,
"step": 369
},
{
"epoch": 0.18,
"learning_rate": 4.754978075332398e-05,
"loss": 1.1956,
"step": 370
},
{
"epoch": 0.19,
"learning_rate": 4.753190261658045e-05,
"loss": 1.1754,
"step": 371
},
{
"epoch": 0.19,
"learning_rate": 4.751396287847556e-05,
"loss": 1.2031,
"step": 372
},
{
"epoch": 0.19,
"learning_rate": 4.7495961588055836e-05,
"loss": 1.2026,
"step": 373
},
{
"epoch": 0.19,
"learning_rate": 4.747789879453615e-05,
"loss": 1.1978,
"step": 374
},
{
"epoch": 0.19,
"learning_rate": 4.7459774547299475e-05,
"loss": 1.2589,
"step": 375
},
{
"epoch": 0.19,
"learning_rate": 4.7441588895896805e-05,
"loss": 1.2708,
"step": 376
},
{
"epoch": 0.19,
"learning_rate": 4.742334189004704e-05,
"loss": 1.2407,
"step": 377
},
{
"epoch": 0.19,
"learning_rate": 4.7405033579636756e-05,
"loss": 1.2278,
"step": 378
},
{
"epoch": 0.19,
"learning_rate": 4.738666401472022e-05,
"loss": 1.2109,
"step": 379
},
{
"epoch": 0.19,
"learning_rate": 4.736823324551909e-05,
"loss": 1.1796,
"step": 380
},
{
"epoch": 0.19,
"learning_rate": 4.73497413224224e-05,
"loss": 1.241,
"step": 381
},
{
"epoch": 0.19,
"learning_rate": 4.733118829598635e-05,
"loss": 1.2697,
"step": 382
},
{
"epoch": 0.19,
"learning_rate": 4.7312574216934225e-05,
"loss": 1.2478,
"step": 383
},
{
"epoch": 0.19,
"learning_rate": 4.7293899136156184e-05,
"loss": 1.2407,
"step": 384
},
{
"epoch": 0.19,
"learning_rate": 4.72751631047092e-05,
"loss": 1.1911,
"step": 385
},
{
"epoch": 0.19,
"learning_rate": 4.725636617381686e-05,
"loss": 1.2688,
"step": 386
},
{
"epoch": 0.19,
"learning_rate": 4.723750839486926e-05,
"loss": 1.239,
"step": 387
},
{
"epoch": 0.19,
"learning_rate": 4.721858981942284e-05,
"loss": 1.2337,
"step": 388
},
{
"epoch": 0.19,
"learning_rate": 4.719961049920027e-05,
"loss": 1.2041,
"step": 389
},
{
"epoch": 0.2,
"learning_rate": 4.71805704860903e-05,
"loss": 1.2164,
"step": 390
},
{
"epoch": 0.2,
"learning_rate": 4.716146983214757e-05,
"loss": 1.223,
"step": 391
},
{
"epoch": 0.2,
"learning_rate": 4.7142308589592556e-05,
"loss": 1.2392,
"step": 392
},
{
"epoch": 0.2,
"learning_rate": 4.7123086810811356e-05,
"loss": 1.1836,
"step": 393
},
{
"epoch": 0.2,
"learning_rate": 4.710380454835559e-05,
"loss": 1.2215,
"step": 394
},
{
"epoch": 0.2,
"learning_rate": 4.708446185494222e-05,
"loss": 1.208,
"step": 395
},
{
"epoch": 0.2,
"learning_rate": 4.7065058783453424e-05,
"loss": 1.2059,
"step": 396
},
{
"epoch": 0.2,
"learning_rate": 4.704559538693647e-05,
"loss": 1.2306,
"step": 397
},
{
"epoch": 0.2,
"learning_rate": 4.7026071718603536e-05,
"loss": 1.2309,
"step": 398
},
{
"epoch": 0.2,
"learning_rate": 4.700648783183159e-05,
"loss": 1.2496,
"step": 399
},
{
"epoch": 0.2,
"learning_rate": 4.698684378016222e-05,
"loss": 1.223,
"step": 400
},
{
"epoch": 0.2,
"eval_acceptance_rate": 0.7480568289756775,
"eval_acceptance_rate_1": 0.7477021813392639,
"eval_acceptance_rate_2": 0.7478978633880615,
"eval_acceptance_rate_3": 0.7468696236610413,
"eval_acceptance_rate_4": 0.7482213377952576,
"eval_avg_generation_length": 215.4989013671875,
"eval_avg_input_length": 365.6648864746094,
"eval_expected_speedup": 2.431512085060141,
"eval_expected_tokens": 3.039390106325176,
"eval_loss": 1.1151978969573975,
"eval_max_generation_length": 424.7227783203125,
"eval_max_input_length": 630.4989624023438,
"eval_min_generation_length": 59.60786437988281,
"eval_min_input_length": 160.72378540039062,
"eval_pulsar_acceptance_rate": 0.509847526581294,
"eval_runtime": 117.7232,
"eval_samples_per_second": 4.205,
"eval_steps_per_second": 0.136,
"step": 400
},
{
"epoch": 0.2,
"learning_rate": 4.696713961730154e-05,
"loss": 1.2496,
"step": 401
},
{
"epoch": 0.2,
"learning_rate": 4.694737539711994e-05,
"loss": 1.2241,
"step": 402
},
{
"epoch": 0.2,
"learning_rate": 4.6927551173652075e-05,
"loss": 1.2261,
"step": 403
},
{
"epoch": 0.2,
"learning_rate": 4.690766700109659e-05,
"loss": 1.2289,
"step": 404
},
{
"epoch": 0.2,
"learning_rate": 4.6887722933816076e-05,
"loss": 1.2335,
"step": 405
},
{
"epoch": 0.2,
"learning_rate": 4.686771902633684e-05,
"loss": 1.2269,
"step": 406
},
{
"epoch": 0.2,
"learning_rate": 4.684765533334879e-05,
"loss": 1.1942,
"step": 407
},
{
"epoch": 0.2,
"learning_rate": 4.682753190970533e-05,
"loss": 1.1907,
"step": 408
},
{
"epoch": 0.2,
"learning_rate": 4.68073488104231e-05,
"loss": 1.233,
"step": 409
},
{
"epoch": 0.2,
"learning_rate": 4.678710609068193e-05,
"loss": 1.268,
"step": 410
},
{
"epoch": 0.21,
"learning_rate": 4.6766803805824655e-05,
"loss": 1.1887,
"step": 411
},
{
"epoch": 0.21,
"learning_rate": 4.674644201135694e-05,
"loss": 1.1867,
"step": 412
},
{
"epoch": 0.21,
"learning_rate": 4.672602076294714e-05,
"loss": 1.2096,
"step": 413
},
{
"epoch": 0.21,
"learning_rate": 4.670554011642619e-05,
"loss": 1.1924,
"step": 414
},
{
"epoch": 0.21,
"learning_rate": 4.668500012778738e-05,
"loss": 1.2003,
"step": 415
},
{
"epoch": 0.21,
"learning_rate": 4.666440085318626e-05,
"loss": 1.2171,
"step": 416
},
{
"epoch": 0.21,
"learning_rate": 4.6643742348940464e-05,
"loss": 1.1882,
"step": 417
},
{
"epoch": 0.21,
"learning_rate": 4.662302467152955e-05,
"loss": 1.2127,
"step": 418
},
{
"epoch": 0.21,
"learning_rate": 4.660224787759486e-05,
"loss": 1.2136,
"step": 419
},
{
"epoch": 0.21,
"learning_rate": 4.6581412023939354e-05,
"loss": 1.2099,
"step": 420
},
{
"epoch": 0.21,
"learning_rate": 4.656051716752745e-05,
"loss": 1.2396,
"step": 421
},
{
"epoch": 0.21,
"learning_rate": 4.65395633654849e-05,
"loss": 1.1966,
"step": 422
},
{
"epoch": 0.21,
"learning_rate": 4.65185506750986e-05,
"loss": 1.2578,
"step": 423
},
{
"epoch": 0.21,
"learning_rate": 4.649747915381643e-05,
"loss": 1.222,
"step": 424
},
{
"epoch": 0.21,
"learning_rate": 4.6476348859247134e-05,
"loss": 1.1825,
"step": 425
},
{
"epoch": 0.21,
"learning_rate": 4.645515984916013e-05,
"loss": 1.2415,
"step": 426
},
{
"epoch": 0.21,
"learning_rate": 4.643391218148536e-05,
"loss": 1.1667,
"step": 427
},
{
"epoch": 0.21,
"learning_rate": 4.6412605914313144e-05,
"loss": 1.2437,
"step": 428
},
{
"epoch": 0.21,
"learning_rate": 4.639124110589399e-05,
"loss": 1.1796,
"step": 429
},
{
"epoch": 0.21,
"learning_rate": 4.6369817814638475e-05,
"loss": 1.2253,
"step": 430
},
{
"epoch": 0.22,
"learning_rate": 4.634833609911706e-05,
"loss": 1.2117,
"step": 431
},
{
"epoch": 0.22,
"learning_rate": 4.6326796018059936e-05,
"loss": 1.2087,
"step": 432
},
{
"epoch": 0.22,
"learning_rate": 4.630519763035687e-05,
"loss": 1.205,
"step": 433
},
{
"epoch": 0.22,
"learning_rate": 4.6283540995057004e-05,
"loss": 1.2467,
"step": 434
},
{
"epoch": 0.22,
"learning_rate": 4.6261826171368774e-05,
"loss": 1.1599,
"step": 435
},
{
"epoch": 0.22,
"learning_rate": 4.6240053218659674e-05,
"loss": 1.1935,
"step": 436
},
{
"epoch": 0.22,
"learning_rate": 4.621822219645612e-05,
"loss": 1.2211,
"step": 437
},
{
"epoch": 0.22,
"learning_rate": 4.6196333164443295e-05,
"loss": 1.1631,
"step": 438
},
{
"epoch": 0.22,
"learning_rate": 4.617438618246498e-05,
"loss": 1.2275,
"step": 439
},
{
"epoch": 0.22,
"learning_rate": 4.6152381310523387e-05,
"loss": 1.1759,
"step": 440
},
{
"epoch": 0.22,
"learning_rate": 4.6130318608778995e-05,
"loss": 1.2127,
"step": 441
},
{
"epoch": 0.22,
"learning_rate": 4.610819813755038e-05,
"loss": 1.1816,
"step": 442
},
{
"epoch": 0.22,
"learning_rate": 4.608601995731407e-05,
"loss": 1.2486,
"step": 443
},
{
"epoch": 0.22,
"learning_rate": 4.6063784128704367e-05,
"loss": 1.1841,
"step": 444
},
{
"epoch": 0.22,
"learning_rate": 4.604149071251318e-05,
"loss": 1.2022,
"step": 445
},
{
"epoch": 0.22,
"learning_rate": 4.601913976968985e-05,
"loss": 1.2143,
"step": 446
},
{
"epoch": 0.22,
"learning_rate": 4.5996731361340994e-05,
"loss": 1.1515,
"step": 447
},
{
"epoch": 0.22,
"learning_rate": 4.597426554873037e-05,
"loss": 1.2121,
"step": 448
},
{
"epoch": 0.22,
"learning_rate": 4.595174239327862e-05,
"loss": 1.2078,
"step": 449
},
{
"epoch": 0.23,
"learning_rate": 4.592916195656322e-05,
"loss": 1.1593,
"step": 450
},
{
"epoch": 0.23,
"learning_rate": 4.59065243003182e-05,
"loss": 1.1824,
"step": 451
},
{
"epoch": 0.23,
"learning_rate": 4.588382948643406e-05,
"loss": 1.2159,
"step": 452
},
{
"epoch": 0.23,
"learning_rate": 4.586107757695755e-05,
"loss": 1.1879,
"step": 453
},
{
"epoch": 0.23,
"learning_rate": 4.5838268634091524e-05,
"loss": 1.1936,
"step": 454
},
{
"epoch": 0.23,
"learning_rate": 4.581540272019476e-05,
"loss": 1.2864,
"step": 455
},
{
"epoch": 0.23,
"learning_rate": 4.579247989778179e-05,
"loss": 1.2219,
"step": 456
},
{
"epoch": 0.23,
"learning_rate": 4.576950022952274e-05,
"loss": 1.2052,
"step": 457
},
{
"epoch": 0.23,
"learning_rate": 4.574646377824315e-05,
"loss": 1.1897,
"step": 458
},
{
"epoch": 0.23,
"learning_rate": 4.572337060692379e-05,
"loss": 1.2513,
"step": 459
},
{
"epoch": 0.23,
"learning_rate": 4.5700220778700504e-05,
"loss": 1.1641,
"step": 460
},
{
"epoch": 0.23,
"learning_rate": 4.567701435686404e-05,
"loss": 1.2097,
"step": 461
},
{
"epoch": 0.23,
"learning_rate": 4.565375140485989e-05,
"loss": 1.1747,
"step": 462
},
{
"epoch": 0.23,
"learning_rate": 4.563043198628806e-05,
"loss": 1.1827,
"step": 463
},
{
"epoch": 0.23,
"learning_rate": 4.560705616490294e-05,
"loss": 1.2405,
"step": 464
},
{
"epoch": 0.23,
"learning_rate": 4.5583624004613145e-05,
"loss": 1.1829,
"step": 465
},
{
"epoch": 0.23,
"learning_rate": 4.556013556948131e-05,
"loss": 1.1286,
"step": 466
},
{
"epoch": 0.23,
"learning_rate": 4.5536590923723906e-05,
"loss": 1.1309,
"step": 467
},
{
"epoch": 0.23,
"learning_rate": 4.551299013171111e-05,
"loss": 1.2024,
"step": 468
},
{
"epoch": 0.23,
"learning_rate": 4.548933325796658e-05,
"loss": 1.1834,
"step": 469
},
{
"epoch": 0.23,
"learning_rate": 4.546562036716732e-05,
"loss": 1.1864,
"step": 470
},
{
"epoch": 0.24,
"learning_rate": 4.544185152414343e-05,
"loss": 1.1689,
"step": 471
},
{
"epoch": 0.24,
"learning_rate": 4.541802679387806e-05,
"loss": 1.1941,
"step": 472
},
{
"epoch": 0.24,
"learning_rate": 4.539414624150708e-05,
"loss": 1.1407,
"step": 473
},
{
"epoch": 0.24,
"learning_rate": 4.537020993231904e-05,
"loss": 1.1923,
"step": 474
},
{
"epoch": 0.24,
"learning_rate": 4.534621793175487e-05,
"loss": 1.236,
"step": 475
},
{
"epoch": 0.24,
"learning_rate": 4.532217030540781e-05,
"loss": 1.2171,
"step": 476
},
{
"epoch": 0.24,
"learning_rate": 4.5298067119023114e-05,
"loss": 1.1601,
"step": 477
},
{
"epoch": 0.24,
"learning_rate": 4.5273908438498e-05,
"loss": 1.1809,
"step": 478
},
{
"epoch": 0.24,
"learning_rate": 4.524969432988138e-05,
"loss": 1.1645,
"step": 479
},
{
"epoch": 0.24,
"learning_rate": 4.522542485937369e-05,
"loss": 1.2201,
"step": 480
},
{
"epoch": 0.24,
"learning_rate": 4.520110009332674e-05,
"loss": 1.1901,
"step": 481
},
{
"epoch": 0.24,
"learning_rate": 4.517672009824351e-05,
"loss": 1.1686,
"step": 482
},
{
"epoch": 0.24,
"learning_rate": 4.515228494077798e-05,
"loss": 1.1632,
"step": 483
},
{
"epoch": 0.24,
"learning_rate": 4.512779468773494e-05,
"loss": 1.1922,
"step": 484
},
{
"epoch": 0.24,
"learning_rate": 4.510324940606979e-05,
"loss": 1.2166,
"step": 485
},
{
"epoch": 0.24,
"learning_rate": 4.5078649162888406e-05,
"loss": 1.2058,
"step": 486
},
{
"epoch": 0.24,
"learning_rate": 4.505399402544692e-05,
"loss": 1.1951,
"step": 487
},
{
"epoch": 0.24,
"learning_rate": 4.502928406115152e-05,
"loss": 1.1916,
"step": 488
},
{
"epoch": 0.24,
"learning_rate": 4.500451933755833e-05,
"loss": 1.1497,
"step": 489
},
{
"epoch": 0.24,
"learning_rate": 4.497969992237312e-05,
"loss": 1.1552,
"step": 490
},
{
"epoch": 0.25,
"learning_rate": 4.495482588345126e-05,
"loss": 1.1439,
"step": 491
},
{
"epoch": 0.25,
"learning_rate": 4.49298972887974e-05,
"loss": 1.1831,
"step": 492
},
{
"epoch": 0.25,
"learning_rate": 4.490491420656537e-05,
"loss": 1.2012,
"step": 493
},
{
"epoch": 0.25,
"learning_rate": 4.487987670505798e-05,
"loss": 1.1972,
"step": 494
},
{
"epoch": 0.25,
"learning_rate": 4.4854784852726776e-05,
"loss": 1.1618,
"step": 495
},
{
"epoch": 0.25,
"learning_rate": 4.4829638718171954e-05,
"loss": 1.2118,
"step": 496
},
{
"epoch": 0.25,
"learning_rate": 4.480443837014205e-05,
"loss": 1.1706,
"step": 497
},
{
"epoch": 0.25,
"learning_rate": 4.477918387753388e-05,
"loss": 1.1799,
"step": 498
},
{
"epoch": 0.25,
"learning_rate": 4.4753875309392266e-05,
"loss": 1.112,
"step": 499
},
{
"epoch": 0.25,
"learning_rate": 4.4728512734909844e-05,
"loss": 1.154,
"step": 500
},
{
"epoch": 0.25,
"eval_acceptance_rate": 0.7612475752830505,
"eval_acceptance_rate_1": 0.7604774236679077,
"eval_acceptance_rate_2": 0.7610577940940857,
"eval_acceptance_rate_3": 0.7606703042984009,
"eval_acceptance_rate_4": 0.7614873051643372,
"eval_avg_generation_length": 215.4989013671875,
"eval_avg_input_length": 365.6648864746094,
"eval_expected_speedup": 2.4941636620720145,
"eval_expected_tokens": 3.1177045775900183,
"eval_loss": 1.0697674751281738,
"eval_max_generation_length": 424.7227783203125,
"eval_max_input_length": 630.4989624023438,
"eval_min_generation_length": 59.60786437988281,
"eval_min_input_length": 160.72378540039062,
"eval_pulsar_acceptance_rate": 0.5294261443975046,
"eval_runtime": 120.7194,
"eval_samples_per_second": 4.1,
"eval_steps_per_second": 0.133,
"step": 500
},
{
"epoch": 0.25,
"learning_rate": 4.470309622342694e-05,
"loss": 1.1853,
"step": 501
},
{
"epoch": 0.25,
"learning_rate": 4.467762584443131e-05,
"loss": 1.1983,
"step": 502
},
{
"epoch": 0.25,
"learning_rate": 4.465210166755803e-05,
"loss": 1.2206,
"step": 503
},
{
"epoch": 0.25,
"learning_rate": 4.46265237625892e-05,
"loss": 1.211,
"step": 504
},
{
"epoch": 0.25,
"learning_rate": 4.460089219945382e-05,
"loss": 1.2069,
"step": 505
},
{
"epoch": 0.25,
"learning_rate": 4.457520704822765e-05,
"loss": 1.2131,
"step": 506
},
{
"epoch": 0.25,
"learning_rate": 4.454946837913287e-05,
"loss": 1.1619,
"step": 507
},
{
"epoch": 0.25,
"learning_rate": 4.452367626253805e-05,
"loss": 1.1813,
"step": 508
},
{
"epoch": 0.25,
"learning_rate": 4.449783076895783e-05,
"loss": 1.1799,
"step": 509
},
{
"epoch": 0.26,
"learning_rate": 4.4471931969052816e-05,
"loss": 1.1601,
"step": 510
},
{
"epoch": 0.26,
"learning_rate": 4.4445979933629324e-05,
"loss": 1.1948,
"step": 511
},
{
"epoch": 0.26,
"learning_rate": 4.4419974733639244e-05,
"loss": 1.2295,
"step": 512
},
{
"epoch": 0.26,
"learning_rate": 4.4393916440179786e-05,
"loss": 1.1814,
"step": 513
},
{
"epoch": 0.26,
"learning_rate": 4.436780512449334e-05,
"loss": 1.2158,
"step": 514
},
{
"epoch": 0.26,
"learning_rate": 4.434164085796724e-05,
"loss": 1.1483,
"step": 515
},
{
"epoch": 0.26,
"learning_rate": 4.4315423712133595e-05,
"loss": 1.1577,
"step": 516
},
{
"epoch": 0.26,
"learning_rate": 4.4289153758669075e-05,
"loss": 1.1561,
"step": 517
},
{
"epoch": 0.26,
"learning_rate": 4.426283106939474e-05,
"loss": 1.207,
"step": 518
},
{
"epoch": 0.26,
"learning_rate": 4.42364557162758e-05,
"loss": 1.1995,
"step": 519
},
{
"epoch": 0.26,
"learning_rate": 4.421002777142148e-05,
"loss": 1.2024,
"step": 520
},
{
"epoch": 0.26,
"learning_rate": 4.418354730708476e-05,
"loss": 1.1775,
"step": 521
},
{
"epoch": 0.26,
"learning_rate": 4.415701439566223e-05,
"loss": 1.1377,
"step": 522
},
{
"epoch": 0.26,
"learning_rate": 4.413042910969385e-05,
"loss": 1.1504,
"step": 523
},
{
"epoch": 0.26,
"learning_rate": 4.4103791521862784e-05,
"loss": 1.1651,
"step": 524
},
{
"epoch": 0.26,
"learning_rate": 4.4077101704995166e-05,
"loss": 1.1886,
"step": 525
},
{
"epoch": 0.26,
"learning_rate": 4.405035973205994e-05,
"loss": 1.1605,
"step": 526
},
{
"epoch": 0.26,
"learning_rate": 4.4023565676168655e-05,
"loss": 1.1652,
"step": 527
},
{
"epoch": 0.26,
"learning_rate": 4.399671961057522e-05,
"loss": 1.1908,
"step": 528
},
{
"epoch": 0.26,
"learning_rate": 4.396982160867575e-05,
"loss": 1.1913,
"step": 529
},
{
"epoch": 0.27,
"learning_rate": 4.3942871744008374e-05,
"loss": 1.1393,
"step": 530
},
{
"epoch": 0.27,
"learning_rate": 4.391587009025297e-05,
"loss": 1.1878,
"step": 531
},
{
"epoch": 0.27,
"learning_rate": 4.388881672123105e-05,
"loss": 1.1439,
"step": 532
},
{
"epoch": 0.27,
"learning_rate": 4.386171171090547e-05,
"loss": 1.1588,
"step": 533
},
{
"epoch": 0.27,
"learning_rate": 4.383455513338032e-05,
"loss": 1.1719,
"step": 534
},
{
"epoch": 0.27,
"learning_rate": 4.3807347062900624e-05,
"loss": 1.204,
"step": 535
},
{
"epoch": 0.27,
"learning_rate": 4.3780087573852213e-05,
"loss": 1.1343,
"step": 536
},
{
"epoch": 0.27,
"learning_rate": 4.375277674076149e-05,
"loss": 1.1705,
"step": 537
},
{
"epoch": 0.27,
"learning_rate": 4.3725414638295235e-05,
"loss": 1.1341,
"step": 538
},
{
"epoch": 0.27,
"learning_rate": 4.369800134126039e-05,
"loss": 1.1727,
"step": 539
},
{
"epoch": 0.27,
"learning_rate": 4.367053692460385e-05,
"loss": 1.1999,
"step": 540
},
{
"epoch": 0.27,
"learning_rate": 4.3643021463412294e-05,
"loss": 1.171,
"step": 541
},
{
"epoch": 0.27,
"learning_rate": 4.3615455032911946e-05,
"loss": 1.2348,
"step": 542
},
{
"epoch": 0.27,
"learning_rate": 4.358783770846836e-05,
"loss": 1.1795,
"step": 543
},
{
"epoch": 0.27,
"learning_rate": 4.356016956558625e-05,
"loss": 1.1452,
"step": 544
},
{
"epoch": 0.27,
"learning_rate": 4.3532450679909274e-05,
"loss": 1.2039,
"step": 545
},
{
"epoch": 0.27,
"learning_rate": 4.3504681127219793e-05,
"loss": 1.1606,
"step": 546
},
{
"epoch": 0.27,
"learning_rate": 4.3476860983438714e-05,
"loss": 1.171,
"step": 547
},
{
"epoch": 0.27,
"learning_rate": 4.3448990324625244e-05,
"loss": 1.175,
"step": 548
},
{
"epoch": 0.27,
"learning_rate": 4.342106922697669e-05,
"loss": 1.1518,
"step": 549
},
{
"epoch": 0.28,
"learning_rate": 4.3393097766828293e-05,
"loss": 1.2225,
"step": 550
},
{
"epoch": 0.28,
"learning_rate": 4.336507602065293e-05,
"loss": 1.2039,
"step": 551
},
{
"epoch": 0.28,
"learning_rate": 4.3337004065061e-05,
"loss": 1.1321,
"step": 552
},
{
"epoch": 0.28,
"learning_rate": 4.3308881976800146e-05,
"loss": 1.1624,
"step": 553
},
{
"epoch": 0.28,
"learning_rate": 4.3280709832755094e-05,
"loss": 1.152,
"step": 554
},
{
"epoch": 0.28,
"learning_rate": 4.325248770994741e-05,
"loss": 1.1204,
"step": 555
},
{
"epoch": 0.28,
"learning_rate": 4.3224215685535294e-05,
"loss": 1.1581,
"step": 556
},
{
"epoch": 0.28,
"learning_rate": 4.319589383681338e-05,
"loss": 1.1694,
"step": 557
},
{
"epoch": 0.28,
"learning_rate": 4.316752224121252e-05,
"loss": 1.1878,
"step": 558
},
{
"epoch": 0.28,
"learning_rate": 4.313910097629959e-05,
"loss": 1.1455,
"step": 559
},
{
"epoch": 0.28,
"learning_rate": 4.311063011977723e-05,
"loss": 1.1626,
"step": 560
},
{
"epoch": 0.28,
"learning_rate": 4.308210974948367e-05,
"loss": 1.1247,
"step": 561
},
{
"epoch": 0.28,
"learning_rate": 4.305353994339252e-05,
"loss": 1.1165,
"step": 562
},
{
"epoch": 0.28,
"learning_rate": 4.302492077961253e-05,
"loss": 1.159,
"step": 563
},
{
"epoch": 0.28,
"learning_rate": 4.2996252336387414e-05,
"loss": 1.1264,
"step": 564
},
{
"epoch": 0.28,
"learning_rate": 4.29675346920956e-05,
"loss": 1.1833,
"step": 565
},
{
"epoch": 0.28,
"learning_rate": 4.293876792525002e-05,
"loss": 1.1058,
"step": 566
},
{
"epoch": 0.28,
"learning_rate": 4.2909952114497925e-05,
"loss": 1.1406,
"step": 567
},
{
"epoch": 0.28,
"learning_rate": 4.288108733862064e-05,
"loss": 1.1515,
"step": 568
},
{
"epoch": 0.28,
"learning_rate": 4.2852173676533356e-05,
"loss": 1.1789,
"step": 569
},
{
"epoch": 0.28,
"learning_rate": 4.282321120728493e-05,
"loss": 1.1295,
"step": 570
},
{
"epoch": 0.29,
"learning_rate": 4.279420001005764e-05,
"loss": 1.2086,
"step": 571
},
{
"epoch": 0.29,
"learning_rate": 4.2765140164166984e-05,
"loss": 1.133,
"step": 572
},
{
"epoch": 0.29,
"learning_rate": 4.273603174906149e-05,
"loss": 1.1454,
"step": 573
},
{
"epoch": 0.29,
"learning_rate": 4.270687484432243e-05,
"loss": 1.1213,
"step": 574
},
{
"epoch": 0.29,
"learning_rate": 4.267766952966369e-05,
"loss": 1.1569,
"step": 575
},
{
"epoch": 0.29,
"learning_rate": 4.2648415884931476e-05,
"loss": 1.2088,
"step": 576
},
{
"epoch": 0.29,
"learning_rate": 4.261911399010413e-05,
"loss": 1.2193,
"step": 577
},
{
"epoch": 0.29,
"learning_rate": 4.258976392529192e-05,
"loss": 1.1351,
"step": 578
},
{
"epoch": 0.29,
"learning_rate": 4.256036577073681e-05,
"loss": 1.164,
"step": 579
},
{
"epoch": 0.29,
"learning_rate": 4.2530919606812216e-05,
"loss": 1.1552,
"step": 580
},
{
"epoch": 0.29,
"learning_rate": 4.250142551402284e-05,
"loss": 1.1633,
"step": 581
},
{
"epoch": 0.29,
"learning_rate": 4.247188357300439e-05,
"loss": 1.0952,
"step": 582
},
{
"epoch": 0.29,
"learning_rate": 4.244229386452342e-05,
"loss": 1.1708,
"step": 583
},
{
"epoch": 0.29,
"learning_rate": 4.241265646947706e-05,
"loss": 1.1491,
"step": 584
},
{
"epoch": 0.29,
"learning_rate": 4.2382971468892806e-05,
"loss": 1.0903,
"step": 585
},
{
"epoch": 0.29,
"learning_rate": 4.235323894392832e-05,
"loss": 1.169,
"step": 586
},
{
"epoch": 0.29,
"learning_rate": 4.23234589758712e-05,
"loss": 1.1661,
"step": 587
},
{
"epoch": 0.29,
"learning_rate": 4.229363164613874e-05,
"loss": 1.1544,
"step": 588
},
{
"epoch": 0.29,
"learning_rate": 4.2263757036277705e-05,
"loss": 1.2005,
"step": 589
},
{
"epoch": 0.29,
"learning_rate": 4.223383522796415e-05,
"loss": 1.1497,
"step": 590
},
{
"epoch": 0.3,
"learning_rate": 4.220386630300315e-05,
"loss": 1.1374,
"step": 591
},
{
"epoch": 0.3,
"learning_rate": 4.217385034332861e-05,
"loss": 1.1383,
"step": 592
},
{
"epoch": 0.3,
"learning_rate": 4.214378743100302e-05,
"loss": 1.1291,
"step": 593
},
{
"epoch": 0.3,
"learning_rate": 4.211367764821722e-05,
"loss": 1.1526,
"step": 594
},
{
"epoch": 0.3,
"learning_rate": 4.2083521077290213e-05,
"loss": 1.1522,
"step": 595
},
{
"epoch": 0.3,
"learning_rate": 4.205331780066892e-05,
"loss": 1.1595,
"step": 596
},
{
"epoch": 0.3,
"learning_rate": 4.202306790092792e-05,
"loss": 1.1817,
"step": 597
},
{
"epoch": 0.3,
"learning_rate": 4.199277146076933e-05,
"loss": 1.1368,
"step": 598
},
{
"epoch": 0.3,
"learning_rate": 4.1962428563022414e-05,
"loss": 1.1418,
"step": 599
},
{
"epoch": 0.3,
"learning_rate": 4.193203929064353e-05,
"loss": 1.109,
"step": 600
},
{
"epoch": 0.3,
"eval_acceptance_rate": 0.7722666263580322,
"eval_acceptance_rate_1": 0.7711466550827026,
"eval_acceptance_rate_2": 0.7723726034164429,
"eval_acceptance_rate_3": 0.7718802094459534,
"eval_acceptance_rate_4": 0.7722136378288269,
"eval_avg_generation_length": 215.4989013671875,
"eval_avg_input_length": 365.6648864746094,
"eval_expected_speedup": 2.5479414221081074,
"eval_expected_tokens": 3.184926777635134,
"eval_loss": 1.0362516641616821,
"eval_max_generation_length": 424.7227783203125,
"eval_max_input_length": 630.4989624023438,
"eval_min_generation_length": 59.60786437988281,
"eval_min_input_length": 160.72378540039062,
"eval_pulsar_acceptance_rate": 0.5462316944087835,
"eval_runtime": 115.5731,
"eval_samples_per_second": 4.283,
"eval_steps_per_second": 0.138,
"step": 600
},
{
"epoch": 0.3,
"learning_rate": 4.1901603726715766e-05,
"loss": 1.1199,
"step": 601
},
{
"epoch": 0.3,
"learning_rate": 4.18711219544488e-05,
"loss": 1.1416,
"step": 602
},
{
"epoch": 0.3,
"learning_rate": 4.184059405717863e-05,
"loss": 1.1544,
"step": 603
},
{
"epoch": 0.3,
"learning_rate": 4.181002011836737e-05,
"loss": 1.1035,
"step": 604
},
{
"epoch": 0.3,
"learning_rate": 4.177940022160299e-05,
"loss": 1.1592,
"step": 605
},
{
"epoch": 0.3,
"learning_rate": 4.174873445059913e-05,
"loss": 1.1491,
"step": 606
},
{
"epoch": 0.3,
"learning_rate": 4.171802288919482e-05,
"loss": 1.1718,
"step": 607
},
{
"epoch": 0.3,
"learning_rate": 4.1687265621354314e-05,
"loss": 1.1692,
"step": 608
},
{
"epoch": 0.3,
"learning_rate": 4.165646273116681e-05,
"loss": 1.1169,
"step": 609
},
{
"epoch": 0.3,
"learning_rate": 4.16256143028462e-05,
"loss": 1.1261,
"step": 610
},
{
"epoch": 0.31,
"learning_rate": 4.159472042073096e-05,
"loss": 1.1371,
"step": 611
},
{
"epoch": 0.31,
"learning_rate": 4.156378116928375e-05,
"loss": 1.1792,
"step": 612
},
{
"epoch": 0.31,
"learning_rate": 4.1532796633091296e-05,
"loss": 1.1732,
"step": 613
},
{
"epoch": 0.31,
"learning_rate": 4.150176689686417e-05,
"loss": 1.1309,
"step": 614
},
{
"epoch": 0.31,
"learning_rate": 4.147069204543645e-05,
"loss": 1.1263,
"step": 615
},
{
"epoch": 0.31,
"learning_rate": 4.143957216376561e-05,
"loss": 1.1767,
"step": 616
},
{
"epoch": 0.31,
"learning_rate": 4.140840733693223e-05,
"loss": 1.1351,
"step": 617
},
{
"epoch": 0.31,
"learning_rate": 4.137719765013974e-05,
"loss": 1.1663,
"step": 618
},
{
"epoch": 0.31,
"learning_rate": 4.134594318871423e-05,
"loss": 1.1027,
"step": 619
},
{
"epoch": 0.31,
"learning_rate": 4.131464403810422e-05,
"loss": 1.1137,
"step": 620
},
{
"epoch": 0.31,
"learning_rate": 4.128330028388038e-05,
"loss": 1.1601,
"step": 621
},
{
"epoch": 0.31,
"learning_rate": 4.1251912011735326e-05,
"loss": 1.1421,
"step": 622
},
{
"epoch": 0.31,
"learning_rate": 4.122047930748343e-05,
"loss": 1.0937,
"step": 623
},
{
"epoch": 0.31,
"learning_rate": 4.118900225706047e-05,
"loss": 1.1721,
"step": 624
},
{
"epoch": 0.31,
"learning_rate": 4.115748094652352e-05,
"loss": 1.1249,
"step": 625
},
{
"epoch": 0.31,
"learning_rate": 4.112591546205064e-05,
"loss": 1.1497,
"step": 626
},
{
"epoch": 0.31,
"learning_rate": 4.1094305889940646e-05,
"loss": 1.125,
"step": 627
},
{
"epoch": 0.31,
"learning_rate": 4.106265231661292e-05,
"loss": 1.146,
"step": 628
},
{
"epoch": 0.31,
"learning_rate": 4.1030954828607095e-05,
"loss": 1.095,
"step": 629
},
{
"epoch": 0.32,
"learning_rate": 4.099921351258292e-05,
"loss": 1.1813,
"step": 630
},
{
"epoch": 0.32,
"learning_rate": 4.0967428455319925e-05,
"loss": 1.1432,
"step": 631
},
{
"epoch": 0.32,
"learning_rate": 4.093559974371725e-05,
"loss": 1.1015,
"step": 632
},
{
"epoch": 0.32,
"learning_rate": 4.090372746479337e-05,
"loss": 1.1351,
"step": 633
},
{
"epoch": 0.32,
"learning_rate": 4.0871811705685883e-05,
"loss": 1.1564,
"step": 634
},
{
"epoch": 0.32,
"learning_rate": 4.0839852553651265e-05,
"loss": 1.1309,
"step": 635
},
{
"epoch": 0.32,
"learning_rate": 4.0807850096064605e-05,
"loss": 1.1105,
"step": 636
},
{
"epoch": 0.32,
"learning_rate": 4.0775804420419404e-05,
"loss": 1.1595,
"step": 637
},
{
"epoch": 0.32,
"learning_rate": 4.0743715614327317e-05,
"loss": 1.1385,
"step": 638
},
{
"epoch": 0.32,
"learning_rate": 4.07115837655179e-05,
"loss": 1.1302,
"step": 639
},
{
"epoch": 0.32,
"learning_rate": 4.067940896183843e-05,
"loss": 1.1351,
"step": 640
},
{
"epoch": 0.32,
"learning_rate": 4.064719129125356e-05,
"loss": 1.1174,
"step": 641
},
{
"epoch": 0.32,
"learning_rate": 4.06149308418452e-05,
"loss": 1.1146,
"step": 642
},
{
"epoch": 0.32,
"learning_rate": 4.058262770181217e-05,
"loss": 1.1536,
"step": 643
},
{
"epoch": 0.32,
"learning_rate": 4.0550281959470023e-05,
"loss": 1.131,
"step": 644
},
{
"epoch": 0.32,
"learning_rate": 4.051789370325078e-05,
"loss": 1.1646,
"step": 645
},
{
"epoch": 0.32,
"learning_rate": 4.048546302170271e-05,
"loss": 1.1256,
"step": 646
},
{
"epoch": 0.32,
"learning_rate": 4.0452990003490047e-05,
"loss": 1.1432,
"step": 647
},
{
"epoch": 0.32,
"learning_rate": 4.042047473739278e-05,
"loss": 1.1254,
"step": 648
},
{
"epoch": 0.32,
"learning_rate": 4.0387917312306414e-05,
"loss": 1.1108,
"step": 649
},
{
"epoch": 0.33,
"learning_rate": 4.03553178172417e-05,
"loss": 1.1011,
"step": 650
},
{
"epoch": 0.33,
"learning_rate": 4.0322676341324415e-05,
"loss": 1.1704,
"step": 651
},
{
"epoch": 0.33,
"learning_rate": 4.028999297379511e-05,
"loss": 1.1681,
"step": 652
},
{
"epoch": 0.33,
"learning_rate": 4.025726780400886e-05,
"loss": 1.1494,
"step": 653
},
{
"epoch": 0.33,
"learning_rate": 4.0224500921435035e-05,
"loss": 1.0603,
"step": 654
},
{
"epoch": 0.33,
"learning_rate": 4.019169241565703e-05,
"loss": 1.145,
"step": 655
},
{
"epoch": 0.33,
"learning_rate": 4.0158842376372064e-05,
"loss": 1.1126,
"step": 656
},
{
"epoch": 0.33,
"learning_rate": 4.0125950893390876e-05,
"loss": 1.0923,
"step": 657
},
{
"epoch": 0.33,
"learning_rate": 4.009301805663752e-05,
"loss": 1.1282,
"step": 658
},
{
"epoch": 0.33,
"learning_rate": 4.006004395614913e-05,
"loss": 1.1105,
"step": 659
},
{
"epoch": 0.33,
"learning_rate": 4.002702868207563e-05,
"loss": 1.1426,
"step": 660
},
{
"epoch": 0.33,
"learning_rate": 3.999397232467952e-05,
"loss": 1.1869,
"step": 661
},
{
"epoch": 0.33,
"learning_rate": 3.996087497433562e-05,
"loss": 1.145,
"step": 662
},
{
"epoch": 0.33,
"learning_rate": 3.9927736721530805e-05,
"loss": 1.1091,
"step": 663
},
{
"epoch": 0.33,
"learning_rate": 3.9894557656863823e-05,
"loss": 1.1219,
"step": 664
},
{
"epoch": 0.33,
"learning_rate": 3.9861337871044954e-05,
"loss": 1.1093,
"step": 665
},
{
"epoch": 0.33,
"learning_rate": 3.9828077454895815e-05,
"loss": 1.1313,
"step": 666
},
{
"epoch": 0.33,
"learning_rate": 3.9794776499349126e-05,
"loss": 1.153,
"step": 667
},
{
"epoch": 0.33,
"learning_rate": 3.976143509544843e-05,
"loss": 1.1286,
"step": 668
},
{
"epoch": 0.33,
"learning_rate": 3.972805333434784e-05,
"loss": 1.0872,
"step": 669
},
{
"epoch": 0.34,
"learning_rate": 3.969463130731183e-05,
"loss": 1.1666,
"step": 670
},
{
"epoch": 0.34,
"learning_rate": 3.966116910571494e-05,
"loss": 1.1418,
"step": 671
},
{
"epoch": 0.34,
"learning_rate": 3.9627666821041545e-05,
"loss": 1.0809,
"step": 672
},
{
"epoch": 0.34,
"learning_rate": 3.9594124544885615e-05,
"loss": 1.1292,
"step": 673
},
{
"epoch": 0.34,
"learning_rate": 3.956054236895046e-05,
"loss": 1.0884,
"step": 674
},
{
"epoch": 0.34,
"learning_rate": 3.952692038504846e-05,
"loss": 1.1141,
"step": 675
},
{
"epoch": 0.34,
"learning_rate": 3.949325868510083e-05,
"loss": 1.1354,
"step": 676
},
{
"epoch": 0.34,
"learning_rate": 3.945955736113739e-05,
"loss": 1.0922,
"step": 677
},
{
"epoch": 0.34,
"learning_rate": 3.942581650529625e-05,
"loss": 1.1635,
"step": 678
},
{
"epoch": 0.34,
"learning_rate": 3.9392036209823644e-05,
"loss": 1.1277,
"step": 679
},
{
"epoch": 0.34,
"learning_rate": 3.935821656707359e-05,
"loss": 1.1573,
"step": 680
},
{
"epoch": 0.34,
"learning_rate": 3.932435766950772e-05,
"loss": 1.1519,
"step": 681
},
{
"epoch": 0.34,
"learning_rate": 3.929045960969494e-05,
"loss": 1.1483,
"step": 682
},
{
"epoch": 0.34,
"learning_rate": 3.925652248031127e-05,
"loss": 1.1415,
"step": 683
},
{
"epoch": 0.34,
"learning_rate": 3.9222546374139533e-05,
"loss": 1.1372,
"step": 684
},
{
"epoch": 0.34,
"learning_rate": 3.9188531384069096e-05,
"loss": 1.1498,
"step": 685
},
{
"epoch": 0.34,
"learning_rate": 3.915447760309564e-05,
"loss": 1.1528,
"step": 686
},
{
"epoch": 0.34,
"learning_rate": 3.9120385124320894e-05,
"loss": 1.1207,
"step": 687
},
{
"epoch": 0.34,
"learning_rate": 3.9086254040952416e-05,
"loss": 1.1115,
"step": 688
},
{
"epoch": 0.34,
"learning_rate": 3.905208444630327e-05,
"loss": 1.1155,
"step": 689
},
{
"epoch": 0.34,
"learning_rate": 3.901787643379182e-05,
"loss": 1.1199,
"step": 690
},
{
"epoch": 0.35,
"learning_rate": 3.898363009694148e-05,
"loss": 1.1424,
"step": 691
},
{
"epoch": 0.35,
"learning_rate": 3.894934552938041e-05,
"loss": 1.0838,
"step": 692
},
{
"epoch": 0.35,
"learning_rate": 3.891502282484132e-05,
"loss": 1.2201,
"step": 693
},
{
"epoch": 0.35,
"learning_rate": 3.888066207716117e-05,
"loss": 1.1223,
"step": 694
},
{
"epoch": 0.35,
"learning_rate": 3.884626338028094e-05,
"loss": 1.1345,
"step": 695
},
{
"epoch": 0.35,
"learning_rate": 3.8811826828245334e-05,
"loss": 1.1092,
"step": 696
},
{
"epoch": 0.35,
"learning_rate": 3.877735251520258e-05,
"loss": 1.1209,
"step": 697
},
{
"epoch": 0.35,
"learning_rate": 3.874284053540416e-05,
"loss": 1.1368,
"step": 698
},
{
"epoch": 0.35,
"learning_rate": 3.870829098320446e-05,
"loss": 1.1273,
"step": 699
},
{
"epoch": 0.35,
"learning_rate": 3.867370395306068e-05,
"loss": 1.0906,
"step": 700
},
{
"epoch": 0.35,
"eval_acceptance_rate": 0.7809614539146423,
"eval_acceptance_rate_1": 0.7801964282989502,
"eval_acceptance_rate_2": 0.7812386751174927,
"eval_acceptance_rate_3": 0.7801474332809448,
"eval_acceptance_rate_4": 0.7808693647384644,
"eval_avg_generation_length": 215.4989013671875,
"eval_avg_input_length": 365.6648864746094,
"eval_expected_speedup": 2.5913201862101434,
"eval_expected_tokens": 3.2391502327626793,
"eval_loss": 1.0097919702529907,
"eval_max_generation_length": 424.7227783203125,
"eval_max_input_length": 630.4989624023438,
"eval_min_generation_length": 59.60786437988281,
"eval_min_input_length": 160.72378540039062,
"eval_pulsar_acceptance_rate": 0.5597875581906698,
"eval_runtime": 116.5895,
"eval_samples_per_second": 4.246,
"eval_steps_per_second": 0.137,
"step": 700
},
{
"epoch": 0.35,
"learning_rate": 3.8639079539532405e-05,
"loss": 1.1175,
"step": 701
},
{
"epoch": 0.35,
"learning_rate": 3.860441783728149e-05,
"loss": 1.136,
"step": 702
},
{
"epoch": 0.35,
"learning_rate": 3.8569718941071684e-05,
"loss": 1.1491,
"step": 703
},
{
"epoch": 0.35,
"learning_rate": 3.853498294576845e-05,
"loss": 1.1229,
"step": 704
},
{
"epoch": 0.35,
"learning_rate": 3.850020994633868e-05,
"loss": 1.0972,
"step": 705
},
{
"epoch": 0.35,
"learning_rate": 3.846540003785042e-05,
"loss": 1.0989,
"step": 706
},
{
"epoch": 0.35,
"learning_rate": 3.8430553315472626e-05,
"loss": 1.0977,
"step": 707
},
{
"epoch": 0.35,
"learning_rate": 3.8395669874474915e-05,
"loss": 1.1266,
"step": 708
},
{
"epoch": 0.35,
"learning_rate": 3.8360749810227286e-05,
"loss": 1.1067,
"step": 709
},
{
"epoch": 0.35,
"learning_rate": 3.832579321819985e-05,
"loss": 1.1327,
"step": 710
},
{
"epoch": 0.36,
"learning_rate": 3.829080019396259e-05,
"loss": 1.1055,
"step": 711
},
{
"epoch": 0.36,
"learning_rate": 3.825577083318512e-05,
"loss": 1.1305,
"step": 712
},
{
"epoch": 0.36,
"learning_rate": 3.822070523163636e-05,
"loss": 1.1188,
"step": 713
},
{
"epoch": 0.36,
"learning_rate": 3.818560348518433e-05,
"loss": 1.1313,
"step": 714
},
{
"epoch": 0.36,
"learning_rate": 3.8150465689795854e-05,
"loss": 1.1256,
"step": 715
},
{
"epoch": 0.36,
"learning_rate": 3.8115291941536345e-05,
"loss": 1.0979,
"step": 716
},
{
"epoch": 0.36,
"learning_rate": 3.8080082336569455e-05,
"loss": 1.1199,
"step": 717
},
{
"epoch": 0.36,
"learning_rate": 3.8044836971156936e-05,
"loss": 1.1361,
"step": 718
},
{
"epoch": 0.36,
"learning_rate": 3.800955594165825e-05,
"loss": 1.1286,
"step": 719
},
{
"epoch": 0.36,
"learning_rate": 3.797423934453038e-05,
"loss": 1.0763,
"step": 720
},
{
"epoch": 0.36,
"learning_rate": 3.793888727632757e-05,
"loss": 1.143,
"step": 721
},
{
"epoch": 0.36,
"learning_rate": 3.7903499833701006e-05,
"loss": 1.1314,
"step": 722
},
{
"epoch": 0.36,
"learning_rate": 3.786807711339863e-05,
"loss": 1.1358,
"step": 723
},
{
"epoch": 0.36,
"learning_rate": 3.783261921226479e-05,
"loss": 1.1273,
"step": 724
},
{
"epoch": 0.36,
"learning_rate": 3.779712622724003e-05,
"loss": 1.1515,
"step": 725
},
{
"epoch": 0.36,
"learning_rate": 3.776159825536082e-05,
"loss": 1.1058,
"step": 726
},
{
"epoch": 0.36,
"learning_rate": 3.7726035393759285e-05,
"loss": 1.0671,
"step": 727
},
{
"epoch": 0.36,
"learning_rate": 3.7690437739662924e-05,
"loss": 1.0752,
"step": 728
},
{
"epoch": 0.36,
"learning_rate": 3.7654805390394366e-05,
"loss": 1.0842,
"step": 729
},
{
"epoch": 0.36,
"learning_rate": 3.76191384433711e-05,
"loss": 1.117,
"step": 730
},
{
"epoch": 0.37,
"learning_rate": 3.7583436996105185e-05,
"loss": 1.0982,
"step": 731
},
{
"epoch": 0.37,
"learning_rate": 3.7547701146203005e-05,
"loss": 1.1008,
"step": 732
},
{
"epoch": 0.37,
"learning_rate": 3.751193099136505e-05,
"loss": 1.1313,
"step": 733
},
{
"epoch": 0.37,
"learning_rate": 3.747612662938552e-05,
"loss": 1.1067,
"step": 734
},
{
"epoch": 0.37,
"learning_rate": 3.7440288158152187e-05,
"loss": 1.1422,
"step": 735
},
{
"epoch": 0.37,
"learning_rate": 3.7404415675646054e-05,
"loss": 1.0987,
"step": 736
},
{
"epoch": 0.37,
"learning_rate": 3.7368509279941134e-05,
"loss": 1.1174,
"step": 737
},
{
"epoch": 0.37,
"learning_rate": 3.733256906920413e-05,
"loss": 1.1447,
"step": 738
},
{
"epoch": 0.37,
"learning_rate": 3.72965951416942e-05,
"loss": 1.1129,
"step": 739
},
{
"epoch": 0.37,
"learning_rate": 3.726058759576271e-05,
"loss": 1.1683,
"step": 740
},
{
"epoch": 0.37,
"learning_rate": 3.722454652985289e-05,
"loss": 1.141,
"step": 741
},
{
"epoch": 0.37,
"learning_rate": 3.718847204249966e-05,
"loss": 1.1411,
"step": 742
},
{
"epoch": 0.37,
"learning_rate": 3.715236423232928e-05,
"loss": 1.1417,
"step": 743
},
{
"epoch": 0.37,
"learning_rate": 3.711622319805913e-05,
"loss": 1.1284,
"step": 744
},
{
"epoch": 0.37,
"learning_rate": 3.708004903849741e-05,
"loss": 1.158,
"step": 745
},
{
"epoch": 0.37,
"learning_rate": 3.704384185254288e-05,
"loss": 1.1625,
"step": 746
},
{
"epoch": 0.37,
"learning_rate": 3.700760173918463e-05,
"loss": 1.0862,
"step": 747
},
{
"epoch": 0.37,
"learning_rate": 3.697132879750174e-05,
"loss": 1.151,
"step": 748
},
{
"epoch": 0.37,
"learning_rate": 3.693502312666304e-05,
"loss": 1.0727,
"step": 749
},
{
"epoch": 0.38,
"learning_rate": 3.689868482592684e-05,
"loss": 1.0726,
"step": 750
},
{
"epoch": 0.38,
"learning_rate": 3.6862313994640686e-05,
"loss": 1.1454,
"step": 751
},
{
"epoch": 0.38,
"learning_rate": 3.6825910732241026e-05,
"loss": 1.1174,
"step": 752
},
{
"epoch": 0.38,
"learning_rate": 3.678947513825299e-05,
"loss": 1.1179,
"step": 753
},
{
"epoch": 0.38,
"learning_rate": 3.675300731229012e-05,
"loss": 1.0917,
"step": 754
},
{
"epoch": 0.38,
"learning_rate": 3.671650735405404e-05,
"loss": 1.0923,
"step": 755
},
{
"epoch": 0.38,
"learning_rate": 3.667997536333424e-05,
"loss": 1.1101,
"step": 756
},
{
"epoch": 0.38,
"learning_rate": 3.6643411440007804e-05,
"loss": 1.0979,
"step": 757
},
{
"epoch": 0.38,
"learning_rate": 3.66068156840391e-05,
"loss": 1.0926,
"step": 758
},
{
"epoch": 0.38,
"learning_rate": 3.65701881954795e-05,
"loss": 1.0618,
"step": 759
},
{
"epoch": 0.38,
"learning_rate": 3.65335290744672e-05,
"loss": 1.1073,
"step": 760
},
{
"epoch": 0.38,
"learning_rate": 3.649683842122681e-05,
"loss": 1.1045,
"step": 761
},
{
"epoch": 0.38,
"learning_rate": 3.6460116336069176e-05,
"loss": 1.1433,
"step": 762
},
{
"epoch": 0.38,
"learning_rate": 3.642336291939109e-05,
"loss": 1.1615,
"step": 763
},
{
"epoch": 0.38,
"learning_rate": 3.6386578271674984e-05,
"loss": 1.1436,
"step": 764
},
{
"epoch": 0.38,
"learning_rate": 3.634976249348867e-05,
"loss": 1.14,
"step": 765
},
{
"epoch": 0.38,
"learning_rate": 3.631291568548509e-05,
"loss": 1.1574,
"step": 766
},
{
"epoch": 0.38,
"learning_rate": 3.627603794840201e-05,
"loss": 1.1343,
"step": 767
},
{
"epoch": 0.38,
"learning_rate": 3.623912938306176e-05,
"loss": 1.0863,
"step": 768
},
{
"epoch": 0.38,
"learning_rate": 3.6202190090370944e-05,
"loss": 1.1124,
"step": 769
},
{
"epoch": 0.39,
"learning_rate": 3.616522017132017e-05,
"loss": 1.1027,
"step": 770
},
{
"epoch": 0.39,
"learning_rate": 3.6128219726983784e-05,
"loss": 1.0998,
"step": 771
},
{
"epoch": 0.39,
"learning_rate": 3.6091188858519607e-05,
"loss": 1.0973,
"step": 772
},
{
"epoch": 0.39,
"learning_rate": 3.6054127667168596e-05,
"loss": 1.1419,
"step": 773
},
{
"epoch": 0.39,
"learning_rate": 3.601703625425464e-05,
"loss": 1.1095,
"step": 774
},
{
"epoch": 0.39,
"learning_rate": 3.597991472118426e-05,
"loss": 1.0861,
"step": 775
},
{
"epoch": 0.39,
"learning_rate": 3.5942763169446295e-05,
"loss": 1.0426,
"step": 776
},
{
"epoch": 0.39,
"learning_rate": 3.590558170061168e-05,
"loss": 1.1339,
"step": 777
},
{
"epoch": 0.39,
"learning_rate": 3.586837041633312e-05,
"loss": 1.1308,
"step": 778
},
{
"epoch": 0.39,
"learning_rate": 3.5831129418344845e-05,
"loss": 1.1134,
"step": 779
},
{
"epoch": 0.39,
"learning_rate": 3.579385880846232e-05,
"loss": 1.1633,
"step": 780
},
{
"epoch": 0.39,
"learning_rate": 3.5756558688581985e-05,
"loss": 1.101,
"step": 781
},
{
"epoch": 0.39,
"learning_rate": 3.571922916068094e-05,
"loss": 1.1154,
"step": 782
},
{
"epoch": 0.39,
"learning_rate": 3.568187032681667e-05,
"loss": 1.0932,
"step": 783
},
{
"epoch": 0.39,
"learning_rate": 3.564448228912682e-05,
"loss": 1.0782,
"step": 784
},
{
"epoch": 0.39,
"learning_rate": 3.5607065149828843e-05,
"loss": 1.0861,
"step": 785
},
{
"epoch": 0.39,
"learning_rate": 3.556961901121978e-05,
"loss": 1.1374,
"step": 786
},
{
"epoch": 0.39,
"learning_rate": 3.553214397567595e-05,
"loss": 1.0708,
"step": 787
},
{
"epoch": 0.39,
"learning_rate": 3.549464014565265e-05,
"loss": 1.0721,
"step": 788
},
{
"epoch": 0.39,
"learning_rate": 3.545710762368392e-05,
"loss": 1.1101,
"step": 789
},
{
"epoch": 0.4,
"learning_rate": 3.5419546512382266e-05,
"loss": 1.1279,
"step": 790
},
{
"epoch": 0.4,
"learning_rate": 3.5381956914438305e-05,
"loss": 1.1433,
"step": 791
},
{
"epoch": 0.4,
"learning_rate": 3.534433893262058e-05,
"loss": 1.0469,
"step": 792
},
{
"epoch": 0.4,
"learning_rate": 3.530669266977521e-05,
"loss": 1.0985,
"step": 793
},
{
"epoch": 0.4,
"learning_rate": 3.526901822882564e-05,
"loss": 1.1376,
"step": 794
},
{
"epoch": 0.4,
"learning_rate": 3.523131571277235e-05,
"loss": 1.1125,
"step": 795
},
{
"epoch": 0.4,
"learning_rate": 3.519358522469259e-05,
"loss": 1.1431,
"step": 796
},
{
"epoch": 0.4,
"learning_rate": 3.515582686774007e-05,
"loss": 1.1175,
"step": 797
},
{
"epoch": 0.4,
"learning_rate": 3.511804074514468e-05,
"loss": 1.0595,
"step": 798
},
{
"epoch": 0.4,
"learning_rate": 3.508022696021226e-05,
"loss": 1.0998,
"step": 799
},
{
"epoch": 0.4,
"learning_rate": 3.504238561632424e-05,
"loss": 1.139,
"step": 800
},
{
"epoch": 0.4,
"eval_acceptance_rate": 0.7878292202949524,
"eval_acceptance_rate_1": 0.7870592474937439,
"eval_acceptance_rate_2": 0.7882823348045349,
"eval_acceptance_rate_3": 0.7869051098823547,
"eval_acceptance_rate_4": 0.7877203822135925,
"eval_avg_generation_length": 215.4989013671875,
"eval_avg_input_length": 365.6648864746094,
"eval_expected_speedup": 2.6261817718295335,
"eval_expected_tokens": 3.282727214786917,
"eval_loss": 0.9893876910209656,
"eval_max_generation_length": 424.7227783203125,
"eval_max_input_length": 630.4989624023438,
"eval_min_generation_length": 59.60786437988281,
"eval_min_input_length": 160.72378540039062,
"eval_pulsar_acceptance_rate": 0.5706818036967292,
"eval_runtime": 123.1218,
"eval_samples_per_second": 4.02,
"eval_steps_per_second": 0.13,
"step": 800
},
{
"epoch": 0.4,
"learning_rate": 3.500451681693741e-05,
"loss": 1.1112,
"step": 801
},
{
"epoch": 0.4,
"learning_rate": 3.496662066558363e-05,
"loss": 1.1186,
"step": 802
},
{
"epoch": 0.4,
"learning_rate": 3.4928697265869515e-05,
"loss": 1.1539,
"step": 803
},
{
"epoch": 0.4,
"learning_rate": 3.489074672147621e-05,
"loss": 1.1095,
"step": 804
},
{
"epoch": 0.4,
"learning_rate": 3.485276913615905e-05,
"loss": 1.1393,
"step": 805
},
{
"epoch": 0.4,
"learning_rate": 3.4814764613747286e-05,
"loss": 1.1348,
"step": 806
},
{
"epoch": 0.4,
"learning_rate": 3.4776733258143854e-05,
"loss": 1.1079,
"step": 807
},
{
"epoch": 0.4,
"learning_rate": 3.473867517332501e-05,
"loss": 1.115,
"step": 808
},
{
"epoch": 0.4,
"learning_rate": 3.470059046334011e-05,
"loss": 1.0997,
"step": 809
},
{
"epoch": 0.41,
"learning_rate": 3.4662479232311306e-05,
"loss": 1.1019,
"step": 810
},
{
"epoch": 0.41,
"learning_rate": 3.4624341584433244e-05,
"loss": 1.1012,
"step": 811
},
{
"epoch": 0.41,
"learning_rate": 3.458617762397279e-05,
"loss": 1.11,
"step": 812
},
{
"epoch": 0.41,
"learning_rate": 3.454798745526876e-05,
"loss": 1.0878,
"step": 813
},
{
"epoch": 0.41,
"learning_rate": 3.450977118273162e-05,
"loss": 1.1063,
"step": 814
},
{
"epoch": 0.41,
"learning_rate": 3.447152891084319e-05,
"loss": 1.062,
"step": 815
},
{
"epoch": 0.41,
"learning_rate": 3.4433260744156396e-05,
"loss": 1.0861,
"step": 816
},
{
"epoch": 0.41,
"learning_rate": 3.439496678729493e-05,
"loss": 1.1144,
"step": 817
},
{
"epoch": 0.41,
"learning_rate": 3.435664714495301e-05,
"loss": 1.1233,
"step": 818
},
{
"epoch": 0.41,
"learning_rate": 3.4318301921895084e-05,
"loss": 1.076,
"step": 819
},
{
"epoch": 0.41,
"learning_rate": 3.427993122295552e-05,
"loss": 1.0954,
"step": 820
},
{
"epoch": 0.41,
"learning_rate": 3.424153515303835e-05,
"loss": 1.1487,
"step": 821
},
{
"epoch": 0.41,
"learning_rate": 3.4203113817116957e-05,
"loss": 1.1348,
"step": 822
},
{
"epoch": 0.41,
"learning_rate": 3.41646673202338e-05,
"loss": 1.0922,
"step": 823
},
{
"epoch": 0.41,
"learning_rate": 3.412619576750014e-05,
"loss": 1.0858,
"step": 824
},
{
"epoch": 0.41,
"learning_rate": 3.4087699264095745e-05,
"loss": 1.1319,
"step": 825
},
{
"epoch": 0.41,
"learning_rate": 3.4049177915268576e-05,
"loss": 1.0388,
"step": 826
},
{
"epoch": 0.41,
"learning_rate": 3.4010631826334526e-05,
"loss": 1.0843,
"step": 827
},
{
"epoch": 0.41,
"learning_rate": 3.397206110267713e-05,
"loss": 1.1308,
"step": 828
},
{
"epoch": 0.41,
"learning_rate": 3.3933465849747275e-05,
"loss": 1.0888,
"step": 829
},
{
"epoch": 0.41,
"learning_rate": 3.389484617306292e-05,
"loss": 1.0859,
"step": 830
},
{
"epoch": 0.42,
"learning_rate": 3.385620217820877e-05,
"loss": 1.1502,
"step": 831
},
{
"epoch": 0.42,
"learning_rate": 3.381753397083604e-05,
"loss": 1.0971,
"step": 832
},
{
"epoch": 0.42,
"learning_rate": 3.377884165666212e-05,
"loss": 1.1102,
"step": 833
},
{
"epoch": 0.42,
"learning_rate": 3.3740125341470336e-05,
"loss": 1.0611,
"step": 834
},
{
"epoch": 0.42,
"learning_rate": 3.3701385131109616e-05,
"loss": 1.1284,
"step": 835
},
{
"epoch": 0.42,
"learning_rate": 3.3662621131494204e-05,
"loss": 1.0967,
"step": 836
},
{
"epoch": 0.42,
"learning_rate": 3.36238334486034e-05,
"loss": 1.1029,
"step": 837
},
{
"epoch": 0.42,
"learning_rate": 3.358502218848125e-05,
"loss": 1.0541,
"step": 838
},
{
"epoch": 0.42,
"learning_rate": 3.3546187457236244e-05,
"loss": 1.1006,
"step": 839
},
{
"epoch": 0.42,
"learning_rate": 3.350732936104108e-05,
"loss": 1.119,
"step": 840
},
{
"epoch": 0.42,
"learning_rate": 3.346844800613229e-05,
"loss": 1.0804,
"step": 841
},
{
"epoch": 0.42,
"learning_rate": 3.342954349881001e-05,
"loss": 1.1324,
"step": 842
},
{
"epoch": 0.42,
"learning_rate": 3.33906159454377e-05,
"loss": 1.116,
"step": 843
},
{
"epoch": 0.42,
"learning_rate": 3.335166545244178e-05,
"loss": 1.0748,
"step": 844
},
{
"epoch": 0.42,
"learning_rate": 3.3312692126311425e-05,
"loss": 1.0553,
"step": 845
},
{
"epoch": 0.42,
"learning_rate": 3.327369607359821e-05,
"loss": 1.0633,
"step": 846
},
{
"epoch": 0.42,
"learning_rate": 3.3234677400915865e-05,
"loss": 1.0415,
"step": 847
},
{
"epoch": 0.42,
"learning_rate": 3.319563621493994e-05,
"loss": 1.0678,
"step": 848
},
{
"epoch": 0.42,
"learning_rate": 3.3156572622407565e-05,
"loss": 1.0809,
"step": 849
},
{
"epoch": 0.42,
"learning_rate": 3.311748673011709e-05,
"loss": 1.0527,
"step": 850
},
{
"epoch": 0.43,
"learning_rate": 3.307837864492786e-05,
"loss": 1.1094,
"step": 851
},
{
"epoch": 0.43,
"learning_rate": 3.3039248473759885e-05,
"loss": 1.0822,
"step": 852
},
{
"epoch": 0.43,
"learning_rate": 3.300009632359357e-05,
"loss": 1.1206,
"step": 853
},
{
"epoch": 0.43,
"learning_rate": 3.2960922301469385e-05,
"loss": 1.0881,
"step": 854
},
{
"epoch": 0.43,
"learning_rate": 3.2921726514487614e-05,
"loss": 1.1375,
"step": 855
},
{
"epoch": 0.43,
"learning_rate": 3.2882509069808044e-05,
"loss": 1.0995,
"step": 856
},
{
"epoch": 0.43,
"learning_rate": 3.284327007464966e-05,
"loss": 1.0886,
"step": 857
},
{
"epoch": 0.43,
"learning_rate": 3.28040096362904e-05,
"loss": 1.0945,
"step": 858
},
{
"epoch": 0.43,
"learning_rate": 3.276472786206679e-05,
"loss": 1.119,
"step": 859
},
{
"epoch": 0.43,
"learning_rate": 3.272542485937369e-05,
"loss": 1.0826,
"step": 860
},
{
"epoch": 0.43,
"learning_rate": 3.2686100735664024e-05,
"loss": 1.0542,
"step": 861
},
{
"epoch": 0.43,
"learning_rate": 3.264675559844844e-05,
"loss": 1.09,
"step": 862
},
{
"epoch": 0.43,
"learning_rate": 3.260738955529504e-05,
"loss": 1.0797,
"step": 863
},
{
"epoch": 0.43,
"learning_rate": 3.2568002713829084e-05,
"loss": 1.0397,
"step": 864
},
{
"epoch": 0.43,
"learning_rate": 3.252859518173269e-05,
"loss": 1.1289,
"step": 865
},
{
"epoch": 0.43,
"learning_rate": 3.2489167066744547e-05,
"loss": 1.0691,
"step": 866
},
{
"epoch": 0.43,
"learning_rate": 3.244971847665962e-05,
"loss": 1.0645,
"step": 867
},
{
"epoch": 0.43,
"learning_rate": 3.241024951932885e-05,
"loss": 1.1049,
"step": 868
},
{
"epoch": 0.43,
"learning_rate": 3.237076030265884e-05,
"loss": 1.0716,
"step": 869
},
{
"epoch": 0.43,
"learning_rate": 3.2331250934611624e-05,
"loss": 1.0983,
"step": 870
},
{
"epoch": 0.44,
"learning_rate": 3.229172152320429e-05,
"loss": 1.0458,
"step": 871
},
{
"epoch": 0.44,
"learning_rate": 3.225217217650876e-05,
"loss": 1.0907,
"step": 872
},
{
"epoch": 0.44,
"learning_rate": 3.22126030026514e-05,
"loss": 1.083,
"step": 873
},
{
"epoch": 0.44,
"learning_rate": 3.217301410981285e-05,
"loss": 1.0795,
"step": 874
},
{
"epoch": 0.44,
"learning_rate": 3.213340560622763e-05,
"loss": 1.0982,
"step": 875
},
{
"epoch": 0.44,
"learning_rate": 3.2093777600183875e-05,
"loss": 1.0506,
"step": 876
},
{
"epoch": 0.44,
"learning_rate": 3.205413020002303e-05,
"loss": 1.1047,
"step": 877
},
{
"epoch": 0.44,
"learning_rate": 3.201446351413958e-05,
"loss": 1.0556,
"step": 878
},
{
"epoch": 0.44,
"learning_rate": 3.1974777650980735e-05,
"loss": 1.1373,
"step": 879
},
{
"epoch": 0.44,
"learning_rate": 3.1935072719046115e-05,
"loss": 1.0722,
"step": 880
},
{
"epoch": 0.44,
"learning_rate": 3.18953488268875e-05,
"loss": 1.0447,
"step": 881
},
{
"epoch": 0.44,
"learning_rate": 3.185560608310849e-05,
"loss": 1.07,
"step": 882
},
{
"epoch": 0.44,
"learning_rate": 3.181584459636423e-05,
"loss": 1.0674,
"step": 883
},
{
"epoch": 0.44,
"learning_rate": 3.1776064475361114e-05,
"loss": 1.0579,
"step": 884
},
{
"epoch": 0.44,
"learning_rate": 3.173626582885645e-05,
"loss": 1.0901,
"step": 885
},
{
"epoch": 0.44,
"learning_rate": 3.169644876565824e-05,
"loss": 1.0997,
"step": 886
},
{
"epoch": 0.44,
"learning_rate": 3.165661339462481e-05,
"loss": 1.054,
"step": 887
},
{
"epoch": 0.44,
"learning_rate": 3.161675982466454e-05,
"loss": 1.09,
"step": 888
},
{
"epoch": 0.44,
"learning_rate": 3.1576888164735575e-05,
"loss": 1.1098,
"step": 889
},
{
"epoch": 0.45,
"learning_rate": 3.1536998523845494e-05,
"loss": 1.0551,
"step": 890
},
{
"epoch": 0.45,
"learning_rate": 3.149709101105107e-05,
"loss": 1.1155,
"step": 891
},
{
"epoch": 0.45,
"learning_rate": 3.145716573545792e-05,
"loss": 1.1034,
"step": 892
},
{
"epoch": 0.45,
"learning_rate": 3.141722280622021e-05,
"loss": 1.103,
"step": 893
},
{
"epoch": 0.45,
"learning_rate": 3.137726233254038e-05,
"loss": 1.1217,
"step": 894
},
{
"epoch": 0.45,
"learning_rate": 3.133728442366885e-05,
"loss": 1.0947,
"step": 895
},
{
"epoch": 0.45,
"learning_rate": 3.129728918890371e-05,
"loss": 1.1059,
"step": 896
},
{
"epoch": 0.45,
"learning_rate": 3.1257276737590365e-05,
"loss": 1.1023,
"step": 897
},
{
"epoch": 0.45,
"learning_rate": 3.121724717912138e-05,
"loss": 1.0828,
"step": 898
},
{
"epoch": 0.45,
"learning_rate": 3.117720062293599e-05,
"loss": 1.0746,
"step": 899
},
{
"epoch": 0.45,
"learning_rate": 3.1137137178519985e-05,
"loss": 1.089,
"step": 900
},
{
"epoch": 0.45,
"eval_acceptance_rate": 0.7923885583877563,
"eval_acceptance_rate_1": 0.791546106338501,
"eval_acceptance_rate_2": 0.7925243973731995,
"eval_acceptance_rate_3": 0.7917414307594299,
"eval_acceptance_rate_4": 0.7924862504005432,
"eval_avg_generation_length": 215.4989013671875,
"eval_avg_input_length": 365.6648864746094,
"eval_expected_speedup": 2.6496205162262405,
"eval_expected_tokens": 3.3120256452828007,
"eval_loss": 0.9742996692657471,
"eval_max_generation_length": 424.7227783203125,
"eval_max_input_length": 630.4989624023438,
"eval_min_generation_length": 59.60786437988281,
"eval_min_input_length": 160.72378540039062,
"eval_pulsar_acceptance_rate": 0.5780064113207001,
"eval_runtime": 122.5543,
"eval_samples_per_second": 4.039,
"eval_steps_per_second": 0.131,
"step": 900
},
{
"epoch": 0.45,
"learning_rate": 3.1097056955405274e-05,
"loss": 1.0531,
"step": 901
},
{
"epoch": 0.45,
"learning_rate": 3.105696006316966e-05,
"loss": 1.1154,
"step": 902
},
{
"epoch": 0.45,
"learning_rate": 3.101684661143653e-05,
"loss": 1.122,
"step": 903
},
{
"epoch": 0.45,
"learning_rate": 3.0976716709874496e-05,
"loss": 1.0532,
"step": 904
},
{
"epoch": 0.45,
"learning_rate": 3.093657046819722e-05,
"loss": 1.0593,
"step": 905
},
{
"epoch": 0.45,
"learning_rate": 3.0896407996162954e-05,
"loss": 1.0343,
"step": 906
},
{
"epoch": 0.45,
"learning_rate": 3.08562294035744e-05,
"loss": 1.0496,
"step": 907
},
{
"epoch": 0.45,
"learning_rate": 3.081603480027826e-05,
"loss": 1.0795,
"step": 908
},
{
"epoch": 0.45,
"learning_rate": 3.077582429616506e-05,
"loss": 1.0683,
"step": 909
},
{
"epoch": 0.46,
"learning_rate": 3.073559800116879e-05,
"loss": 1.1216,
"step": 910
},
{
"epoch": 0.46,
"learning_rate": 3.06953560252666e-05,
"loss": 1.0314,
"step": 911
},
{
"epoch": 0.46,
"learning_rate": 3.065509847847851e-05,
"loss": 1.0508,
"step": 912
},
{
"epoch": 0.46,
"learning_rate": 3.061482547086712e-05,
"loss": 1.0959,
"step": 913
},
{
"epoch": 0.46,
"learning_rate": 3.05745371125373e-05,
"loss": 1.11,
"step": 914
},
{
"epoch": 0.46,
"learning_rate": 3.053423351363586e-05,
"loss": 1.1324,
"step": 915
},
{
"epoch": 0.46,
"learning_rate": 3.0493914784351328e-05,
"loss": 1.0934,
"step": 916
},
{
"epoch": 0.46,
"learning_rate": 3.045358103491357e-05,
"loss": 1.058,
"step": 917
},
{
"epoch": 0.46,
"learning_rate": 3.0413232375593497e-05,
"loss": 1.0738,
"step": 918
},
{
"epoch": 0.46,
"learning_rate": 3.037286891670281e-05,
"loss": 1.0387,
"step": 919
},
{
"epoch": 0.46,
"learning_rate": 3.0332490768593675e-05,
"loss": 1.1324,
"step": 920
},
{
"epoch": 0.46,
"learning_rate": 3.0292098041658397e-05,
"loss": 1.0277,
"step": 921
},
{
"epoch": 0.46,
"learning_rate": 3.025169084632915e-05,
"loss": 1.0649,
"step": 922
},
{
"epoch": 0.46,
"learning_rate": 3.021126929307766e-05,
"loss": 1.0698,
"step": 923
},
{
"epoch": 0.46,
"learning_rate": 3.017083349241492e-05,
"loss": 1.0828,
"step": 924
},
{
"epoch": 0.46,
"learning_rate": 3.0130383554890856e-05,
"loss": 1.0868,
"step": 925
},
{
"epoch": 0.46,
"learning_rate": 3.008991959109406e-05,
"loss": 1.1356,
"step": 926
},
{
"epoch": 0.46,
"learning_rate": 3.004944171165146e-05,
"loss": 1.079,
"step": 927
},
{
"epoch": 0.46,
"learning_rate": 3.0008950027228033e-05,
"loss": 1.0532,
"step": 928
},
{
"epoch": 0.46,
"learning_rate": 2.9968444648526493e-05,
"loss": 1.0455,
"step": 929
},
{
"epoch": 0.47,
"learning_rate": 2.9927925686287006e-05,
"loss": 1.0617,
"step": 930
},
{
"epoch": 0.47,
"learning_rate": 2.988739325128687e-05,
"loss": 1.0588,
"step": 931
},
{
"epoch": 0.47,
"learning_rate": 2.984684745434021e-05,
"loss": 1.0922,
"step": 932
},
{
"epoch": 0.47,
"learning_rate": 2.9806288406297676e-05,
"loss": 1.0807,
"step": 933
},
{
"epoch": 0.47,
"learning_rate": 2.9765716218046175e-05,
"loss": 1.0941,
"step": 934
},
{
"epoch": 0.47,
"learning_rate": 2.972513100050851e-05,
"loss": 1.0444,
"step": 935
},
{
"epoch": 0.47,
"learning_rate": 2.9684532864643122e-05,
"loss": 1.06,
"step": 936
},
{
"epoch": 0.47,
"learning_rate": 2.964392192144375e-05,
"loss": 1.0972,
"step": 937
},
{
"epoch": 0.47,
"learning_rate": 2.9603298281939178e-05,
"loss": 1.0861,
"step": 938
},
{
"epoch": 0.47,
"learning_rate": 2.956266205719288e-05,
"loss": 1.0482,
"step": 939
},
{
"epoch": 0.47,
"learning_rate": 2.952201335830275e-05,
"loss": 1.0848,
"step": 940
},
{
"epoch": 0.47,
"learning_rate": 2.9481352296400766e-05,
"loss": 1.0709,
"step": 941
},
{
"epoch": 0.47,
"learning_rate": 2.944067898265272e-05,
"loss": 1.0574,
"step": 942
},
{
"epoch": 0.47,
"learning_rate": 2.9399993528257902e-05,
"loss": 1.0338,
"step": 943
},
{
"epoch": 0.47,
"learning_rate": 2.9359296044448794e-05,
"loss": 1.0411,
"step": 944
},
{
"epoch": 0.47,
"learning_rate": 2.9318586642490763e-05,
"loss": 1.0973,
"step": 945
},
{
"epoch": 0.47,
"learning_rate": 2.927786543368175e-05,
"loss": 1.1025,
"step": 946
},
{
"epoch": 0.47,
"learning_rate": 2.9237132529351996e-05,
"loss": 1.0788,
"step": 947
},
{
"epoch": 0.47,
"learning_rate": 2.9196388040863693e-05,
"loss": 1.0344,
"step": 948
},
{
"epoch": 0.47,
"learning_rate": 2.915563207961074e-05,
"loss": 1.1088,
"step": 949
},
{
"epoch": 0.47,
"learning_rate": 2.9114864757018352e-05,
"loss": 1.08,
"step": 950
},
{
"epoch": 0.48,
"learning_rate": 2.9074086184542843e-05,
"loss": 1.1063,
"step": 951
},
{
"epoch": 0.48,
"learning_rate": 2.9033296473671278e-05,
"loss": 1.1185,
"step": 952
},
{
"epoch": 0.48,
"learning_rate": 2.8992495735921165e-05,
"loss": 1.0879,
"step": 953
},
{
"epoch": 0.48,
"learning_rate": 2.8951684082840163e-05,
"loss": 1.1085,
"step": 954
},
{
"epoch": 0.48,
"learning_rate": 2.8910861626005776e-05,
"loss": 1.1001,
"step": 955
},
{
"epoch": 0.48,
"learning_rate": 2.8870028477025042e-05,
"loss": 1.0557,
"step": 956
},
{
"epoch": 0.48,
"learning_rate": 2.8829184747534214e-05,
"loss": 1.0364,
"step": 957
},
{
"epoch": 0.48,
"learning_rate": 2.878833054919851e-05,
"loss": 1.0747,
"step": 958
},
{
"epoch": 0.48,
"learning_rate": 2.874746599371175e-05,
"loss": 1.0824,
"step": 959
},
{
"epoch": 0.48,
"learning_rate": 2.870659119279605e-05,
"loss": 1.0787,
"step": 960
},
{
"epoch": 0.48,
"learning_rate": 2.8665706258201574e-05,
"loss": 1.0667,
"step": 961
},
{
"epoch": 0.48,
"learning_rate": 2.862481130170615e-05,
"loss": 1.0696,
"step": 962
},
{
"epoch": 0.48,
"learning_rate": 2.8583906435115047e-05,
"loss": 1.0751,
"step": 963
},
{
"epoch": 0.48,
"learning_rate": 2.8542991770260608e-05,
"loss": 1.0816,
"step": 964
},
{
"epoch": 0.48,
"learning_rate": 2.850206741900195e-05,
"loss": 1.0741,
"step": 965
},
{
"epoch": 0.48,
"learning_rate": 2.846113349322469e-05,
"loss": 1.1111,
"step": 966
},
{
"epoch": 0.48,
"learning_rate": 2.8420190104840628e-05,
"loss": 1.0798,
"step": 967
},
{
"epoch": 0.48,
"learning_rate": 2.8379237365787426e-05,
"loss": 1.0841,
"step": 968
},
{
"epoch": 0.48,
"learning_rate": 2.8338275388028295e-05,
"loss": 1.075,
"step": 969
},
{
"epoch": 0.48,
"learning_rate": 2.8297304283551728e-05,
"loss": 1.0658,
"step": 970
},
{
"epoch": 0.49,
"learning_rate": 2.825632416437115e-05,
"loss": 1.0991,
"step": 971
},
{
"epoch": 0.49,
"learning_rate": 2.8215335142524657e-05,
"loss": 1.0533,
"step": 972
},
{
"epoch": 0.49,
"learning_rate": 2.817433733007466e-05,
"loss": 1.0409,
"step": 973
},
{
"epoch": 0.49,
"learning_rate": 2.8133330839107608e-05,
"loss": 1.0434,
"step": 974
},
{
"epoch": 0.49,
"learning_rate": 2.8092315781733696e-05,
"loss": 1.0965,
"step": 975
},
{
"epoch": 0.49,
"learning_rate": 2.8051292270086503e-05,
"loss": 1.0593,
"step": 976
},
{
"epoch": 0.49,
"learning_rate": 2.8010260416322774e-05,
"loss": 1.0305,
"step": 977
},
{
"epoch": 0.49,
"learning_rate": 2.7969220332622003e-05,
"loss": 1.1153,
"step": 978
},
{
"epoch": 0.49,
"learning_rate": 2.792817213118623e-05,
"loss": 1.0635,
"step": 979
},
{
"epoch": 0.49,
"learning_rate": 2.788711592423966e-05,
"loss": 1.0628,
"step": 980
},
{
"epoch": 0.49,
"learning_rate": 2.7846051824028386e-05,
"loss": 1.0642,
"step": 981
},
{
"epoch": 0.49,
"learning_rate": 2.7804979942820113e-05,
"loss": 1.0516,
"step": 982
},
{
"epoch": 0.49,
"learning_rate": 2.776390039290378e-05,
"loss": 1.0787,
"step": 983
},
{
"epoch": 0.49,
"learning_rate": 2.7722813286589316e-05,
"loss": 1.0905,
"step": 984
},
{
"epoch": 0.49,
"learning_rate": 2.7681718736207298e-05,
"loss": 1.0554,
"step": 985
},
{
"epoch": 0.49,
"learning_rate": 2.764061685410865e-05,
"loss": 1.0823,
"step": 986
},
{
"epoch": 0.49,
"learning_rate": 2.7599507752664354e-05,
"loss": 0.9882,
"step": 987
},
{
"epoch": 0.49,
"learning_rate": 2.755839154426513e-05,
"loss": 1.0977,
"step": 988
},
{
"epoch": 0.49,
"learning_rate": 2.7517268341321112e-05,
"loss": 1.0862,
"step": 989
},
{
"epoch": 0.49,
"learning_rate": 2.7476138256261575e-05,
"loss": 1.0822,
"step": 990
},
{
"epoch": 0.5,
"learning_rate": 2.7435001401534586e-05,
"loss": 1.0608,
"step": 991
},
{
"epoch": 0.5,
"learning_rate": 2.7393857889606756e-05,
"loss": 1.0532,
"step": 992
},
{
"epoch": 0.5,
"learning_rate": 2.7352707832962865e-05,
"loss": 1.0531,
"step": 993
},
{
"epoch": 0.5,
"learning_rate": 2.731155134410559e-05,
"loss": 1.0396,
"step": 994
},
{
"epoch": 0.5,
"learning_rate": 2.727038853555521e-05,
"loss": 1.0583,
"step": 995
},
{
"epoch": 0.5,
"learning_rate": 2.722921951984927e-05,
"loss": 1.0939,
"step": 996
},
{
"epoch": 0.5,
"learning_rate": 2.7188044409542278e-05,
"loss": 1.0158,
"step": 997
},
{
"epoch": 0.5,
"learning_rate": 2.7146863317205427e-05,
"loss": 1.0531,
"step": 998
},
{
"epoch": 0.5,
"learning_rate": 2.7105676355426248e-05,
"loss": 1.0627,
"step": 999
},
{
"epoch": 0.5,
"learning_rate": 2.7064483636808313e-05,
"loss": 1.1029,
"step": 1000
},
{
"epoch": 0.5,
"eval_acceptance_rate": 0.7963886857032776,
"eval_acceptance_rate_1": 0.7957209944725037,
"eval_acceptance_rate_2": 0.7964814901351929,
"eval_acceptance_rate_3": 0.7954385876655579,
"eval_acceptance_rate_4": 0.7965638041496277,
"eval_avg_generation_length": 215.4989013671875,
"eval_avg_input_length": 365.6648864746094,
"eval_expected_speedup": 2.6703800889432756,
"eval_expected_tokens": 3.3379751111790945,
"eval_loss": 0.9622464776039124,
"eval_max_generation_length": 424.7227783203125,
"eval_max_input_length": 630.4989624023438,
"eval_min_generation_length": 59.60786437988281,
"eval_min_input_length": 160.72378540039062,
"eval_pulsar_acceptance_rate": 0.5844937777947735,
"eval_runtime": 117.5443,
"eval_samples_per_second": 4.211,
"eval_steps_per_second": 0.136,
"step": 1000
},
{
"epoch": 0.5,
"learning_rate": 2.7023285273970945e-05,
"loss": 1.0857,
"step": 1001
},
{
"epoch": 0.5,
"learning_rate": 2.6982081379548896e-05,
"loss": 1.0756,
"step": 1002
},
{
"epoch": 0.5,
"learning_rate": 2.6940872066192052e-05,
"loss": 1.024,
"step": 1003
},
{
"epoch": 0.5,
"learning_rate": 2.689965744656508e-05,
"loss": 1.0708,
"step": 1004
},
{
"epoch": 0.5,
"learning_rate": 2.6858437633347194e-05,
"loss": 1.0548,
"step": 1005
},
{
"epoch": 0.5,
"learning_rate": 2.681721273923178e-05,
"loss": 1.0898,
"step": 1006
},
{
"epoch": 0.5,
"learning_rate": 2.6775982876926126e-05,
"loss": 1.1023,
"step": 1007
},
{
"epoch": 0.5,
"learning_rate": 2.6734748159151102e-05,
"loss": 1.0682,
"step": 1008
},
{
"epoch": 0.5,
"learning_rate": 2.6693508698640852e-05,
"loss": 1.0774,
"step": 1009
},
{
"epoch": 0.51,
"learning_rate": 2.6652264608142484e-05,
"loss": 1.0695,
"step": 1010
},
{
"epoch": 0.51,
"learning_rate": 2.661101600041577e-05,
"loss": 1.0483,
"step": 1011
},
{
"epoch": 0.51,
"learning_rate": 2.656976298823284e-05,
"loss": 1.0645,
"step": 1012
},
{
"epoch": 0.51,
"learning_rate": 2.652850568437783e-05,
"loss": 1.0311,
"step": 1013
},
{
"epoch": 0.51,
"learning_rate": 2.6487244201646645e-05,
"loss": 1.0826,
"step": 1014
},
{
"epoch": 0.51,
"learning_rate": 2.6445978652846602e-05,
"loss": 1.0644,
"step": 1015
},
{
"epoch": 0.51,
"learning_rate": 2.6404709150796137e-05,
"loss": 1.0863,
"step": 1016
},
{
"epoch": 0.51,
"learning_rate": 2.63634358083245e-05,
"loss": 1.0993,
"step": 1017
},
{
"epoch": 0.51,
"learning_rate": 2.632215873827142e-05,
"loss": 1.0334,
"step": 1018
},
{
"epoch": 0.51,
"learning_rate": 2.6280878053486828e-05,
"loss": 1.0576,
"step": 1019
},
{
"epoch": 0.51,
"learning_rate": 2.623959386683056e-05,
"loss": 1.1187,
"step": 1020
},
{
"epoch": 0.51,
"learning_rate": 2.6198306291171993e-05,
"loss": 1.0788,
"step": 1021
},
{
"epoch": 0.51,
"learning_rate": 2.6157015439389774e-05,
"loss": 1.0775,
"step": 1022
},
{
"epoch": 0.51,
"learning_rate": 2.6115721424371532e-05,
"loss": 1.1013,
"step": 1023
},
{
"epoch": 0.51,
"learning_rate": 2.6074424359013517e-05,
"loss": 1.066,
"step": 1024
},
{
"epoch": 0.51,
"learning_rate": 2.6033124356220328e-05,
"loss": 1.0695,
"step": 1025
},
{
"epoch": 0.51,
"learning_rate": 2.599182152890461e-05,
"loss": 1.068,
"step": 1026
},
{
"epoch": 0.51,
"learning_rate": 2.5950515989986697e-05,
"loss": 1.0418,
"step": 1027
},
{
"epoch": 0.51,
"learning_rate": 2.5909207852394363e-05,
"loss": 1.0706,
"step": 1028
},
{
"epoch": 0.51,
"learning_rate": 2.5867897229062478e-05,
"loss": 1.0937,
"step": 1029
},
{
"epoch": 0.52,
"learning_rate": 2.5826584232932706e-05,
"loss": 1.1007,
"step": 1030
},
{
"epoch": 0.52,
"learning_rate": 2.578526897695321e-05,
"loss": 1.0272,
"step": 1031
},
{
"epoch": 0.52,
"learning_rate": 2.5743951574078314e-05,
"loss": 1.0467,
"step": 1032
},
{
"epoch": 0.52,
"learning_rate": 2.5702632137268223e-05,
"loss": 1.0459,
"step": 1033
},
{
"epoch": 0.52,
"learning_rate": 2.5661310779488695e-05,
"loss": 1.0844,
"step": 1034
},
{
"epoch": 0.52,
"learning_rate": 2.5619987613710756e-05,
"loss": 1.0614,
"step": 1035
},
{
"epoch": 0.52,
"learning_rate": 2.5578662752910347e-05,
"loss": 1.0805,
"step": 1036
},
{
"epoch": 0.52,
"learning_rate": 2.553733631006807e-05,
"loss": 1.0927,
"step": 1037
},
{
"epoch": 0.52,
"learning_rate": 2.5496008398168843e-05,
"loss": 1.0433,
"step": 1038
},
{
"epoch": 0.52,
"learning_rate": 2.5454679130201593e-05,
"loss": 1.0463,
"step": 1039
},
{
"epoch": 0.52,
"learning_rate": 2.5413348619158967e-05,
"loss": 1.1019,
"step": 1040
},
{
"epoch": 0.52,
"learning_rate": 2.5372016978036995e-05,
"loss": 1.0926,
"step": 1041
},
{
"epoch": 0.52,
"learning_rate": 2.5330684319834814e-05,
"loss": 1.0412,
"step": 1042
},
{
"epoch": 0.52,
"learning_rate": 2.528935075755432e-05,
"loss": 1.0451,
"step": 1043
},
{
"epoch": 0.52,
"learning_rate": 2.5248016404199908e-05,
"loss": 1.0732,
"step": 1044
},
{
"epoch": 0.52,
"learning_rate": 2.5206681372778124e-05,
"loss": 1.0617,
"step": 1045
},
{
"epoch": 0.52,
"learning_rate": 2.5165345776297355e-05,
"loss": 1.0603,
"step": 1046
},
{
"epoch": 0.52,
"learning_rate": 2.512400972776755e-05,
"loss": 1.0688,
"step": 1047
},
{
"epoch": 0.52,
"learning_rate": 2.508267334019988e-05,
"loss": 1.0741,
"step": 1048
},
{
"epoch": 0.52,
"learning_rate": 2.5041336726606457e-05,
"loss": 1.0621,
"step": 1049
},
{
"epoch": 0.53,
"learning_rate": 2.5e-05,
"loss": 1.0775,
"step": 1050
},
{
"epoch": 0.53,
"learning_rate": 2.4958663273393546e-05,
"loss": 1.0548,
"step": 1051
},
{
"epoch": 0.53,
"learning_rate": 2.4917326659800123e-05,
"loss": 1.0343,
"step": 1052
},
{
"epoch": 0.53,
"learning_rate": 2.487599027223246e-05,
"loss": 1.0501,
"step": 1053
},
{
"epoch": 0.53,
"learning_rate": 2.483465422370265e-05,
"loss": 1.1063,
"step": 1054
},
{
"epoch": 0.53,
"learning_rate": 2.4793318627221878e-05,
"loss": 1.0481,
"step": 1055
},
{
"epoch": 0.53,
"learning_rate": 2.475198359580009e-05,
"loss": 1.0129,
"step": 1056
},
{
"epoch": 0.53,
"learning_rate": 2.4710649242445686e-05,
"loss": 1.0714,
"step": 1057
},
{
"epoch": 0.53,
"learning_rate": 2.46693156801652e-05,
"loss": 1.071,
"step": 1058
},
{
"epoch": 0.53,
"learning_rate": 2.4627983021963014e-05,
"loss": 1.0617,
"step": 1059
},
{
"epoch": 0.53,
"learning_rate": 2.458665138084104e-05,
"loss": 1.0716,
"step": 1060
},
{
"epoch": 0.53,
"learning_rate": 2.4545320869798406e-05,
"loss": 1.0762,
"step": 1061
},
{
"epoch": 0.53,
"learning_rate": 2.4503991601831163e-05,
"loss": 1.078,
"step": 1062
},
{
"epoch": 0.53,
"learning_rate": 2.4462663689931935e-05,
"loss": 1.0802,
"step": 1063
},
{
"epoch": 0.53,
"learning_rate": 2.4421337247089655e-05,
"loss": 1.0348,
"step": 1064
},
{
"epoch": 0.53,
"learning_rate": 2.438001238628925e-05,
"loss": 1.0986,
"step": 1065
},
{
"epoch": 0.53,
"learning_rate": 2.4338689220511304e-05,
"loss": 1.0836,
"step": 1066
},
{
"epoch": 0.53,
"learning_rate": 2.4297367862731783e-05,
"loss": 1.0953,
"step": 1067
},
{
"epoch": 0.53,
"learning_rate": 2.425604842592169e-05,
"loss": 1.082,
"step": 1068
},
{
"epoch": 0.53,
"learning_rate": 2.4214731023046793e-05,
"loss": 1.0634,
"step": 1069
},
{
"epoch": 0.54,
"learning_rate": 2.4173415767067297e-05,
"loss": 1.0953,
"step": 1070
},
{
"epoch": 0.54,
"learning_rate": 2.413210277093753e-05,
"loss": 1.1048,
"step": 1071
},
{
"epoch": 0.54,
"learning_rate": 2.4090792147605647e-05,
"loss": 1.0331,
"step": 1072
},
{
"epoch": 0.54,
"learning_rate": 2.404948401001331e-05,
"loss": 1.0468,
"step": 1073
},
{
"epoch": 0.54,
"learning_rate": 2.4008178471095397e-05,
"loss": 1.0733,
"step": 1074
},
{
"epoch": 0.54,
"learning_rate": 2.3966875643779667e-05,
"loss": 1.067,
"step": 1075
},
{
"epoch": 0.54,
"learning_rate": 2.392557564098649e-05,
"loss": 1.0541,
"step": 1076
},
{
"epoch": 0.54,
"learning_rate": 2.388427857562847e-05,
"loss": 1.0561,
"step": 1077
},
{
"epoch": 0.54,
"learning_rate": 2.3842984560610228e-05,
"loss": 1.0801,
"step": 1078
},
{
"epoch": 0.54,
"learning_rate": 2.3801693708828013e-05,
"loss": 1.108,
"step": 1079
},
{
"epoch": 0.54,
"learning_rate": 2.3760406133169443e-05,
"loss": 1.0695,
"step": 1080
},
{
"epoch": 0.54,
"learning_rate": 2.3719121946513174e-05,
"loss": 1.0585,
"step": 1081
},
{
"epoch": 0.54,
"learning_rate": 2.367784126172859e-05,
"loss": 1.0683,
"step": 1082
},
{
"epoch": 0.54,
"learning_rate": 2.3636564191675507e-05,
"loss": 1.0416,
"step": 1083
},
{
"epoch": 0.54,
"learning_rate": 2.3595290849203862e-05,
"loss": 1.0576,
"step": 1084
},
{
"epoch": 0.54,
"learning_rate": 2.3554021347153403e-05,
"loss": 1.0409,
"step": 1085
},
{
"epoch": 0.54,
"learning_rate": 2.351275579835336e-05,
"loss": 1.0671,
"step": 1086
},
{
"epoch": 0.54,
"learning_rate": 2.3471494315622177e-05,
"loss": 1.0665,
"step": 1087
},
{
"epoch": 0.54,
"learning_rate": 2.3430237011767167e-05,
"loss": 1.0606,
"step": 1088
},
{
"epoch": 0.54,
"learning_rate": 2.3388983999584224e-05,
"loss": 1.0902,
"step": 1089
},
{
"epoch": 0.55,
"learning_rate": 2.334773539185752e-05,
"loss": 1.0823,
"step": 1090
},
{
"epoch": 0.55,
"learning_rate": 2.330649130135915e-05,
"loss": 1.0549,
"step": 1091
},
{
"epoch": 0.55,
"learning_rate": 2.32652518408489e-05,
"loss": 1.0574,
"step": 1092
},
{
"epoch": 0.55,
"learning_rate": 2.3224017123073877e-05,
"loss": 1.067,
"step": 1093
},
{
"epoch": 0.55,
"learning_rate": 2.3182787260768223e-05,
"loss": 1.0771,
"step": 1094
},
{
"epoch": 0.55,
"learning_rate": 2.3141562366652812e-05,
"loss": 1.0409,
"step": 1095
},
{
"epoch": 0.55,
"learning_rate": 2.3100342553434924e-05,
"loss": 1.0722,
"step": 1096
},
{
"epoch": 0.55,
"learning_rate": 2.3059127933807954e-05,
"loss": 1.0269,
"step": 1097
},
{
"epoch": 0.55,
"learning_rate": 2.30179186204511e-05,
"loss": 1.103,
"step": 1098
},
{
"epoch": 0.55,
"learning_rate": 2.2976714726029068e-05,
"loss": 1.0702,
"step": 1099
},
{
"epoch": 0.55,
"learning_rate": 2.2935516363191693e-05,
"loss": 1.0696,
"step": 1100
},
{
"epoch": 0.55,
"eval_acceptance_rate": 0.7997491955757141,
"eval_acceptance_rate_1": 0.7989146113395691,
"eval_acceptance_rate_2": 0.8000128865242004,
"eval_acceptance_rate_3": 0.7989808917045593,
"eval_acceptance_rate_4": 0.7997729778289795,
"eval_avg_generation_length": 215.4989013671875,
"eval_avg_input_length": 365.6648864746094,
"eval_expected_speedup": 2.6879625375133878,
"eval_expected_tokens": 3.3599531718917346,
"eval_loss": 0.9537034630775452,
"eval_max_generation_length": 424.7227783203125,
"eval_max_input_length": 630.4989624023438,
"eval_min_generation_length": 59.60786437988281,
"eval_min_input_length": 160.72378540039062,
"eval_pulsar_acceptance_rate": 0.5899882929729336,
"eval_runtime": 125.036,
"eval_samples_per_second": 3.959,
"eval_steps_per_second": 0.128,
"step": 1100
},
{
"epoch": 0.55,
"learning_rate": 2.2894323644573758e-05,
"loss": 1.0623,
"step": 1101
},
{
"epoch": 0.55,
"learning_rate": 2.2853136682794575e-05,
"loss": 1.0298,
"step": 1102
},
{
"epoch": 0.55,
"learning_rate": 2.281195559045772e-05,
"loss": 1.0234,
"step": 1103
},
{
"epoch": 0.55,
"learning_rate": 2.2770780480150744e-05,
"loss": 1.0601,
"step": 1104
},
{
"epoch": 0.55,
"learning_rate": 2.2729611464444794e-05,
"loss": 1.071,
"step": 1105
},
{
"epoch": 0.55,
"learning_rate": 2.2688448655894415e-05,
"loss": 1.1104,
"step": 1106
},
{
"epoch": 0.55,
"learning_rate": 2.2647292167037144e-05,
"loss": 1.0179,
"step": 1107
},
{
"epoch": 0.55,
"learning_rate": 2.2606142110393247e-05,
"loss": 1.0864,
"step": 1108
},
{
"epoch": 0.55,
"learning_rate": 2.2564998598465423e-05,
"loss": 1.0685,
"step": 1109
},
{
"epoch": 0.56,
"learning_rate": 2.2523861743738434e-05,
"loss": 1.0662,
"step": 1110
},
{
"epoch": 0.56,
"learning_rate": 2.248273165867889e-05,
"loss": 1.0783,
"step": 1111
},
{
"epoch": 0.56,
"learning_rate": 2.2441608455734873e-05,
"loss": 1.0326,
"step": 1112
},
{
"epoch": 0.56,
"learning_rate": 2.240049224733566e-05,
"loss": 1.0112,
"step": 1113
},
{
"epoch": 0.56,
"learning_rate": 2.2359383145891364e-05,
"loss": 1.0759,
"step": 1114
},
{
"epoch": 0.56,
"learning_rate": 2.231828126379271e-05,
"loss": 1.0492,
"step": 1115
},
{
"epoch": 0.56,
"learning_rate": 2.2277186713410687e-05,
"loss": 1.0746,
"step": 1116
},
{
"epoch": 0.56,
"learning_rate": 2.223609960709622e-05,
"loss": 1.0831,
"step": 1117
},
{
"epoch": 0.56,
"learning_rate": 2.2195020057179896e-05,
"loss": 1.0267,
"step": 1118
},
{
"epoch": 0.56,
"learning_rate": 2.215394817597162e-05,
"loss": 1.1172,
"step": 1119
},
{
"epoch": 0.56,
"learning_rate": 2.2112884075760347e-05,
"loss": 1.0778,
"step": 1120
},
{
"epoch": 0.56,
"learning_rate": 2.2071827868813774e-05,
"loss": 1.0356,
"step": 1121
},
{
"epoch": 0.56,
"learning_rate": 2.2030779667377996e-05,
"loss": 1.0894,
"step": 1122
},
{
"epoch": 0.56,
"learning_rate": 2.1989739583677238e-05,
"loss": 1.0672,
"step": 1123
},
{
"epoch": 0.56,
"learning_rate": 2.19487077299135e-05,
"loss": 0.9969,
"step": 1124
},
{
"epoch": 0.56,
"learning_rate": 2.190768421826631e-05,
"loss": 1.0739,
"step": 1125
},
{
"epoch": 0.56,
"learning_rate": 2.186666916089239e-05,
"loss": 1.0887,
"step": 1126
},
{
"epoch": 0.56,
"learning_rate": 2.1825662669925354e-05,
"loss": 1.0433,
"step": 1127
},
{
"epoch": 0.56,
"learning_rate": 2.1784664857475352e-05,
"loss": 1.0055,
"step": 1128
},
{
"epoch": 0.56,
"learning_rate": 2.1743675835628856e-05,
"loss": 1.0935,
"step": 1129
},
{
"epoch": 0.56,
"learning_rate": 2.1702695716448278e-05,
"loss": 1.0663,
"step": 1130
},
{
"epoch": 0.57,
"learning_rate": 2.1661724611971708e-05,
"loss": 1.0657,
"step": 1131
},
{
"epoch": 0.57,
"learning_rate": 2.1620762634212586e-05,
"loss": 1.0548,
"step": 1132
},
{
"epoch": 0.57,
"learning_rate": 2.1579809895159375e-05,
"loss": 1.0691,
"step": 1133
},
{
"epoch": 0.57,
"learning_rate": 2.153886650677531e-05,
"loss": 1.0483,
"step": 1134
},
{
"epoch": 0.57,
"learning_rate": 2.1497932580998053e-05,
"loss": 1.0064,
"step": 1135
},
{
"epoch": 0.57,
"learning_rate": 2.1457008229739394e-05,
"loss": 1.0678,
"step": 1136
},
{
"epoch": 0.57,
"learning_rate": 2.141609356488496e-05,
"loss": 1.0393,
"step": 1137
},
{
"epoch": 0.57,
"learning_rate": 2.1375188698293855e-05,
"loss": 1.0745,
"step": 1138
},
{
"epoch": 0.57,
"learning_rate": 2.1334293741798432e-05,
"loss": 1.0285,
"step": 1139
},
{
"epoch": 0.57,
"learning_rate": 2.1293408807203947e-05,
"loss": 1.0815,
"step": 1140
},
{
"epoch": 0.57,
"learning_rate": 2.1252534006288264e-05,
"loss": 1.0833,
"step": 1141
},
{
"epoch": 0.57,
"learning_rate": 2.1211669450801493e-05,
"loss": 1.1144,
"step": 1142
},
{
"epoch": 0.57,
"learning_rate": 2.117081525246579e-05,
"loss": 1.0715,
"step": 1143
},
{
"epoch": 0.57,
"learning_rate": 2.1129971522974967e-05,
"loss": 0.9991,
"step": 1144
},
{
"epoch": 0.57,
"learning_rate": 2.1089138373994223e-05,
"loss": 1.0727,
"step": 1145
},
{
"epoch": 0.57,
"learning_rate": 2.1048315917159846e-05,
"loss": 1.0665,
"step": 1146
},
{
"epoch": 0.57,
"learning_rate": 2.100750426407884e-05,
"loss": 1.0329,
"step": 1147
},
{
"epoch": 0.57,
"learning_rate": 2.0966703526328728e-05,
"loss": 1.0818,
"step": 1148
},
{
"epoch": 0.57,
"learning_rate": 2.0925913815457153e-05,
"loss": 1.0683,
"step": 1149
},
{
"epoch": 0.57,
"learning_rate": 2.088513524298165e-05,
"loss": 1.0691,
"step": 1150
},
{
"epoch": 0.58,
"learning_rate": 2.0844367920389273e-05,
"loss": 1.0868,
"step": 1151
},
{
"epoch": 0.58,
"learning_rate": 2.080361195913631e-05,
"loss": 1.0347,
"step": 1152
},
{
"epoch": 0.58,
"learning_rate": 2.0762867470648013e-05,
"loss": 1.0336,
"step": 1153
},
{
"epoch": 0.58,
"learning_rate": 2.072213456631825e-05,
"loss": 1.0622,
"step": 1154
},
{
"epoch": 0.58,
"learning_rate": 2.068141335750925e-05,
"loss": 1.0389,
"step": 1155
},
{
"epoch": 0.58,
"learning_rate": 2.0640703955551212e-05,
"loss": 1.021,
"step": 1156
},
{
"epoch": 0.58,
"learning_rate": 2.0600006471742104e-05,
"loss": 1.0481,
"step": 1157
},
{
"epoch": 0.58,
"learning_rate": 2.0559321017347285e-05,
"loss": 1.0539,
"step": 1158
},
{
"epoch": 0.58,
"learning_rate": 2.0518647703599236e-05,
"loss": 1.0805,
"step": 1159
},
{
"epoch": 0.58,
"learning_rate": 2.047798664169726e-05,
"loss": 1.0347,
"step": 1160
},
{
"epoch": 0.58,
"learning_rate": 2.043733794280712e-05,
"loss": 1.0473,
"step": 1161
},
{
"epoch": 0.58,
"learning_rate": 2.0396701718060824e-05,
"loss": 1.0746,
"step": 1162
},
{
"epoch": 0.58,
"learning_rate": 2.035607807855625e-05,
"loss": 1.0589,
"step": 1163
},
{
"epoch": 0.58,
"learning_rate": 2.031546713535688e-05,
"loss": 1.0741,
"step": 1164
},
{
"epoch": 0.58,
"learning_rate": 2.02748689994915e-05,
"loss": 1.0455,
"step": 1165
},
{
"epoch": 0.58,
"learning_rate": 2.0234283781953834e-05,
"loss": 1.0328,
"step": 1166
},
{
"epoch": 0.58,
"learning_rate": 2.0193711593702326e-05,
"loss": 1.0738,
"step": 1167
},
{
"epoch": 0.58,
"learning_rate": 2.0153152545659798e-05,
"loss": 1.0643,
"step": 1168
},
{
"epoch": 0.58,
"learning_rate": 2.0112606748713138e-05,
"loss": 1.077,
"step": 1169
},
{
"epoch": 0.58,
"learning_rate": 2.0072074313712997e-05,
"loss": 1.0995,
"step": 1170
},
{
"epoch": 0.59,
"learning_rate": 2.003155535147351e-05,
"loss": 1.0389,
"step": 1171
},
{
"epoch": 0.59,
"learning_rate": 1.9991049972771972e-05,
"loss": 1.0526,
"step": 1172
},
{
"epoch": 0.59,
"learning_rate": 1.9950558288348542e-05,
"loss": 1.0182,
"step": 1173
},
{
"epoch": 0.59,
"learning_rate": 1.9910080408905944e-05,
"loss": 1.0181,
"step": 1174
},
{
"epoch": 0.59,
"learning_rate": 1.9869616445109147e-05,
"loss": 1.072,
"step": 1175
},
{
"epoch": 0.59,
"learning_rate": 1.9829166507585083e-05,
"loss": 1.0588,
"step": 1176
},
{
"epoch": 0.59,
"learning_rate": 1.9788730706922342e-05,
"loss": 1.0817,
"step": 1177
},
{
"epoch": 0.59,
"learning_rate": 1.974830915367086e-05,
"loss": 1.0473,
"step": 1178
},
{
"epoch": 0.59,
"learning_rate": 1.9707901958341612e-05,
"loss": 1.0257,
"step": 1179
},
{
"epoch": 0.59,
"learning_rate": 1.9667509231406334e-05,
"loss": 1.0812,
"step": 1180
},
{
"epoch": 0.59,
"learning_rate": 1.962713108329719e-05,
"loss": 1.0369,
"step": 1181
},
{
"epoch": 0.59,
"learning_rate": 1.958676762440651e-05,
"loss": 1.0576,
"step": 1182
},
{
"epoch": 0.59,
"learning_rate": 1.9546418965086442e-05,
"loss": 1.0561,
"step": 1183
},
{
"epoch": 0.59,
"learning_rate": 1.9506085215648675e-05,
"loss": 1.0647,
"step": 1184
},
{
"epoch": 0.59,
"learning_rate": 1.9465766486364143e-05,
"loss": 1.0738,
"step": 1185
},
{
"epoch": 0.59,
"learning_rate": 1.9425462887462712e-05,
"loss": 1.0721,
"step": 1186
},
{
"epoch": 0.59,
"learning_rate": 1.9385174529132884e-05,
"loss": 1.0369,
"step": 1187
},
{
"epoch": 0.59,
"learning_rate": 1.93449015215215e-05,
"loss": 1.0593,
"step": 1188
},
{
"epoch": 0.59,
"learning_rate": 1.930464397473341e-05,
"loss": 1.0127,
"step": 1189
},
{
"epoch": 0.59,
"learning_rate": 1.9264401998831213e-05,
"loss": 1.0549,
"step": 1190
},
{
"epoch": 0.6,
"learning_rate": 1.922417570383494e-05,
"loss": 1.0684,
"step": 1191
},
{
"epoch": 0.6,
"learning_rate": 1.9183965199721745e-05,
"loss": 1.0996,
"step": 1192
},
{
"epoch": 0.6,
"learning_rate": 1.9143770596425615e-05,
"loss": 1.0973,
"step": 1193
},
{
"epoch": 0.6,
"learning_rate": 1.9103592003837045e-05,
"loss": 1.0669,
"step": 1194
},
{
"epoch": 0.6,
"learning_rate": 1.9063429531802786e-05,
"loss": 1.0473,
"step": 1195
},
{
"epoch": 0.6,
"learning_rate": 1.90232832901255e-05,
"loss": 1.0443,
"step": 1196
},
{
"epoch": 0.6,
"learning_rate": 1.8983153388563486e-05,
"loss": 1.0229,
"step": 1197
},
{
"epoch": 0.6,
"learning_rate": 1.8943039936830344e-05,
"loss": 1.0435,
"step": 1198
},
{
"epoch": 0.6,
"learning_rate": 1.8902943044594735e-05,
"loss": 1.033,
"step": 1199
},
{
"epoch": 0.6,
"learning_rate": 1.8862862821480025e-05,
"loss": 1.0436,
"step": 1200
},
{
"epoch": 0.6,
"eval_acceptance_rate": 0.8013065457344055,
"eval_acceptance_rate_1": 0.8006454110145569,
"eval_acceptance_rate_2": 0.801694929599762,
"eval_acceptance_rate_3": 0.8004037737846375,
"eval_acceptance_rate_4": 0.8012261986732483,
"eval_avg_generation_length": 215.4989013671875,
"eval_avg_input_length": 365.6648864746094,
"eval_expected_speedup": 2.6961550087048733,
"eval_expected_tokens": 3.3701937608810915,
"eval_loss": 0.9475569128990173,
"eval_max_generation_length": 424.7227783203125,
"eval_max_input_length": 630.4989624023438,
"eval_min_generation_length": 59.60786437988281,
"eval_min_input_length": 160.72378540039062,
"eval_pulsar_acceptance_rate": 0.592548440220273,
"eval_runtime": 115.6606,
"eval_samples_per_second": 4.28,
"eval_steps_per_second": 0.138,
"step": 1200
},
{
"epoch": 0.6,
"learning_rate": 1.8822799377064014e-05,
"loss": 1.0242,
"step": 1201
},
{
"epoch": 0.6,
"learning_rate": 1.8782752820878634e-05,
"loss": 1.0798,
"step": 1202
},
{
"epoch": 0.6,
"learning_rate": 1.8742723262409634e-05,
"loss": 1.0341,
"step": 1203
},
{
"epoch": 0.6,
"learning_rate": 1.87027108110963e-05,
"loss": 1.061,
"step": 1204
},
{
"epoch": 0.6,
"learning_rate": 1.866271557633115e-05,
"loss": 1.0348,
"step": 1205
},
{
"epoch": 0.6,
"learning_rate": 1.8622737667459623e-05,
"loss": 1.0868,
"step": 1206
},
{
"epoch": 0.6,
"learning_rate": 1.8582777193779805e-05,
"loss": 1.0313,
"step": 1207
},
{
"epoch": 0.6,
"learning_rate": 1.8542834264542092e-05,
"loss": 1.0401,
"step": 1208
},
{
"epoch": 0.6,
"learning_rate": 1.850290898894893e-05,
"loss": 1.057,
"step": 1209
},
{
"epoch": 0.6,
"learning_rate": 1.8463001476154508e-05,
"loss": 1.0521,
"step": 1210
},
{
"epoch": 0.61,
"learning_rate": 1.8423111835264434e-05,
"loss": 1.0316,
"step": 1211
},
{
"epoch": 0.61,
"learning_rate": 1.8383240175335464e-05,
"loss": 1.1016,
"step": 1212
},
{
"epoch": 0.61,
"learning_rate": 1.8343386605375192e-05,
"loss": 1.0224,
"step": 1213
},
{
"epoch": 0.61,
"learning_rate": 1.8303551234341763e-05,
"loss": 1.0724,
"step": 1214
},
{
"epoch": 0.61,
"learning_rate": 1.826373417114355e-05,
"loss": 1.0664,
"step": 1215
},
{
"epoch": 0.61,
"learning_rate": 1.8223935524638898e-05,
"loss": 1.0341,
"step": 1216
},
{
"epoch": 0.61,
"learning_rate": 1.818415540363577e-05,
"loss": 1.1081,
"step": 1217
},
{
"epoch": 0.61,
"learning_rate": 1.814439391689151e-05,
"loss": 1.0463,
"step": 1218
},
{
"epoch": 0.61,
"learning_rate": 1.8104651173112503e-05,
"loss": 1.0667,
"step": 1219
},
{
"epoch": 0.61,
"learning_rate": 1.806492728095389e-05,
"loss": 1.0352,
"step": 1220
},
{
"epoch": 0.61,
"learning_rate": 1.802522234901927e-05,
"loss": 1.0096,
"step": 1221
},
{
"epoch": 0.61,
"learning_rate": 1.798553648586042e-05,
"loss": 1.0155,
"step": 1222
},
{
"epoch": 0.61,
"learning_rate": 1.7945869799976973e-05,
"loss": 1.0623,
"step": 1223
},
{
"epoch": 0.61,
"learning_rate": 1.7906222399816124e-05,
"loss": 0.9985,
"step": 1224
},
{
"epoch": 0.61,
"learning_rate": 1.7866594393772373e-05,
"loss": 1.066,
"step": 1225
},
{
"epoch": 0.61,
"learning_rate": 1.7826985890187148e-05,
"loss": 1.0624,
"step": 1226
},
{
"epoch": 0.61,
"learning_rate": 1.7787396997348604e-05,
"loss": 1.0523,
"step": 1227
},
{
"epoch": 0.61,
"learning_rate": 1.7747827823491252e-05,
"loss": 1.0252,
"step": 1228
},
{
"epoch": 0.61,
"learning_rate": 1.770827847679571e-05,
"loss": 1.0527,
"step": 1229
},
{
"epoch": 0.61,
"learning_rate": 1.7668749065388385e-05,
"loss": 1.0294,
"step": 1230
},
{
"epoch": 0.62,
"learning_rate": 1.7629239697341165e-05,
"loss": 1.0102,
"step": 1231
},
{
"epoch": 0.62,
"learning_rate": 1.758975048067116e-05,
"loss": 1.0569,
"step": 1232
},
{
"epoch": 0.62,
"learning_rate": 1.7550281523340382e-05,
"loss": 1.0251,
"step": 1233
},
{
"epoch": 0.62,
"learning_rate": 1.7510832933255463e-05,
"loss": 1.0192,
"step": 1234
},
{
"epoch": 0.62,
"learning_rate": 1.7471404818267316e-05,
"loss": 1.0713,
"step": 1235
},
{
"epoch": 0.62,
"learning_rate": 1.7431997286170922e-05,
"loss": 1.043,
"step": 1236
},
{
"epoch": 0.62,
"learning_rate": 1.7392610444704966e-05,
"loss": 1.0782,
"step": 1237
},
{
"epoch": 0.62,
"learning_rate": 1.7353244401551566e-05,
"loss": 1.0445,
"step": 1238
},
{
"epoch": 0.62,
"learning_rate": 1.7313899264335985e-05,
"loss": 1.0483,
"step": 1239
},
{
"epoch": 0.62,
"learning_rate": 1.7274575140626318e-05,
"loss": 1.0852,
"step": 1240
},
{
"epoch": 0.62,
"learning_rate": 1.723527213793322e-05,
"loss": 1.0298,
"step": 1241
},
{
"epoch": 0.62,
"learning_rate": 1.7195990363709604e-05,
"loss": 1.0794,
"step": 1242
},
{
"epoch": 0.62,
"learning_rate": 1.7156729925350336e-05,
"loss": 1.0514,
"step": 1243
},
{
"epoch": 0.62,
"learning_rate": 1.7117490930191965e-05,
"loss": 1.0075,
"step": 1244
},
{
"epoch": 0.62,
"learning_rate": 1.7078273485512392e-05,
"loss": 1.0509,
"step": 1245
},
{
"epoch": 0.62,
"learning_rate": 1.703907769853062e-05,
"loss": 1.1256,
"step": 1246
},
{
"epoch": 0.62,
"learning_rate": 1.699990367640643e-05,
"loss": 1.0699,
"step": 1247
},
{
"epoch": 0.62,
"learning_rate": 1.696075152624012e-05,
"loss": 1.0245,
"step": 1248
},
{
"epoch": 0.62,
"learning_rate": 1.6921621355072144e-05,
"loss": 1.0455,
"step": 1249
},
{
"epoch": 0.62,
"learning_rate": 1.6882513269882917e-05,
"loss": 1.051,
"step": 1250
},
{
"epoch": 0.63,
"learning_rate": 1.684342737759244e-05,
"loss": 1.0808,
"step": 1251
},
{
"epoch": 0.63,
"learning_rate": 1.6804363785060056e-05,
"loss": 1.065,
"step": 1252
},
{
"epoch": 0.63,
"learning_rate": 1.6765322599084147e-05,
"loss": 1.0606,
"step": 1253
},
{
"epoch": 0.63,
"learning_rate": 1.6726303926401792e-05,
"loss": 1.0661,
"step": 1254
},
{
"epoch": 0.63,
"learning_rate": 1.668730787368858e-05,
"loss": 1.0482,
"step": 1255
},
{
"epoch": 0.63,
"learning_rate": 1.6648334547558226e-05,
"loss": 1.0518,
"step": 1256
},
{
"epoch": 0.63,
"learning_rate": 1.6609384054562305e-05,
"loss": 1.0909,
"step": 1257
},
{
"epoch": 0.63,
"learning_rate": 1.6570456501189998e-05,
"loss": 1.0817,
"step": 1258
},
{
"epoch": 0.63,
"learning_rate": 1.6531551993867717e-05,
"loss": 1.0685,
"step": 1259
},
{
"epoch": 0.63,
"learning_rate": 1.6492670638958924e-05,
"loss": 1.0819,
"step": 1260
},
{
"epoch": 0.63,
"learning_rate": 1.6453812542763752e-05,
"loss": 1.067,
"step": 1261
},
{
"epoch": 0.63,
"learning_rate": 1.641497781151877e-05,
"loss": 1.0387,
"step": 1262
},
{
"epoch": 0.63,
"learning_rate": 1.6376166551396607e-05,
"loss": 1.0154,
"step": 1263
},
{
"epoch": 0.63,
"learning_rate": 1.6337378868505805e-05,
"loss": 1.0503,
"step": 1264
},
{
"epoch": 0.63,
"learning_rate": 1.6298614868890387e-05,
"loss": 1.0694,
"step": 1265
},
{
"epoch": 0.63,
"learning_rate": 1.6259874658529663e-05,
"loss": 1.0069,
"step": 1266
},
{
"epoch": 0.63,
"learning_rate": 1.622115834333789e-05,
"loss": 1.0337,
"step": 1267
},
{
"epoch": 0.63,
"learning_rate": 1.6182466029163975e-05,
"loss": 1.0248,
"step": 1268
},
{
"epoch": 0.63,
"learning_rate": 1.614379782179124e-05,
"loss": 1.0355,
"step": 1269
},
{
"epoch": 0.64,
"learning_rate": 1.6105153826937085e-05,
"loss": 1.0345,
"step": 1270
},
{
"epoch": 0.64,
"learning_rate": 1.6066534150252727e-05,
"loss": 1.0513,
"step": 1271
},
{
"epoch": 0.64,
"learning_rate": 1.602793889732288e-05,
"loss": 1.018,
"step": 1272
},
{
"epoch": 0.64,
"learning_rate": 1.598936817366548e-05,
"loss": 1.0813,
"step": 1273
},
{
"epoch": 0.64,
"learning_rate": 1.5950822084731426e-05,
"loss": 1.0543,
"step": 1274
},
{
"epoch": 0.64,
"learning_rate": 1.591230073590425e-05,
"loss": 1.0384,
"step": 1275
},
{
"epoch": 0.64,
"learning_rate": 1.5873804232499863e-05,
"loss": 1.0532,
"step": 1276
},
{
"epoch": 0.64,
"learning_rate": 1.583533267976621e-05,
"loss": 1.0492,
"step": 1277
},
{
"epoch": 0.64,
"learning_rate": 1.5796886182883053e-05,
"loss": 1.0111,
"step": 1278
},
{
"epoch": 0.64,
"learning_rate": 1.5758464846961657e-05,
"loss": 1.052,
"step": 1279
},
{
"epoch": 0.64,
"learning_rate": 1.5720068777044476e-05,
"loss": 1.0407,
"step": 1280
},
{
"epoch": 0.64,
"learning_rate": 1.5681698078104926e-05,
"loss": 1.039,
"step": 1281
},
{
"epoch": 0.64,
"learning_rate": 1.5643352855046996e-05,
"loss": 1.0545,
"step": 1282
},
{
"epoch": 0.64,
"learning_rate": 1.560503321270507e-05,
"loss": 1.0622,
"step": 1283
},
{
"epoch": 0.64,
"learning_rate": 1.5566739255843606e-05,
"loss": 1.0539,
"step": 1284
},
{
"epoch": 0.64,
"learning_rate": 1.5528471089156804e-05,
"loss": 1.0274,
"step": 1285
},
{
"epoch": 0.64,
"learning_rate": 1.549022881726839e-05,
"loss": 1.0528,
"step": 1286
},
{
"epoch": 0.64,
"learning_rate": 1.5452012544731245e-05,
"loss": 1.0162,
"step": 1287
},
{
"epoch": 0.64,
"learning_rate": 1.541382237602721e-05,
"loss": 1.0221,
"step": 1288
},
{
"epoch": 0.64,
"learning_rate": 1.537565841556676e-05,
"loss": 1.0372,
"step": 1289
},
{
"epoch": 0.65,
"learning_rate": 1.5337520767688703e-05,
"loss": 1.047,
"step": 1290
},
{
"epoch": 0.65,
"learning_rate": 1.5299409536659895e-05,
"loss": 1.0357,
"step": 1291
},
{
"epoch": 0.65,
"learning_rate": 1.5261324826675e-05,
"loss": 1.0522,
"step": 1292
},
{
"epoch": 0.65,
"learning_rate": 1.5223266741856152e-05,
"loss": 1.0266,
"step": 1293
},
{
"epoch": 0.65,
"learning_rate": 1.5185235386252717e-05,
"loss": 1.063,
"step": 1294
},
{
"epoch": 0.65,
"learning_rate": 1.5147230863840966e-05,
"loss": 1.0149,
"step": 1295
},
{
"epoch": 0.65,
"learning_rate": 1.5109253278523799e-05,
"loss": 1.0265,
"step": 1296
},
{
"epoch": 0.65,
"learning_rate": 1.5071302734130489e-05,
"loss": 1.0762,
"step": 1297
},
{
"epoch": 0.65,
"learning_rate": 1.5033379334416375e-05,
"loss": 1.0528,
"step": 1298
},
{
"epoch": 0.65,
"learning_rate": 1.499548318306259e-05,
"loss": 1.0235,
"step": 1299
},
{
"epoch": 0.65,
"learning_rate": 1.495761438367577e-05,
"loss": 1.0173,
"step": 1300
},
{
"epoch": 0.65,
"eval_acceptance_rate": 0.8032432198524475,
"eval_acceptance_rate_1": 0.8025344014167786,
"eval_acceptance_rate_2": 0.8036085367202759,
"eval_acceptance_rate_3": 0.802588701248169,
"eval_acceptance_rate_4": 0.8030030727386475,
"eval_avg_generation_length": 215.4989013671875,
"eval_avg_input_length": 365.6648864746094,
"eval_expected_speedup": 2.706382212110599,
"eval_expected_tokens": 3.3829777651382487,
"eval_loss": 0.9430884122848511,
"eval_max_generation_length": 424.7227783203125,
"eval_max_input_length": 630.4989624023438,
"eval_min_generation_length": 59.60786437988281,
"eval_min_input_length": 160.72378540039062,
"eval_pulsar_acceptance_rate": 0.5957444412845622,
"eval_runtime": 115.4683,
"eval_samples_per_second": 4.287,
"eval_steps_per_second": 0.139,
"step": 1300
},
{
"epoch": 0.65,
"learning_rate": 1.4919773039787748e-05,
"loss": 1.0457,
"step": 1301
},
{
"epoch": 0.65,
"learning_rate": 1.4881959254855324e-05,
"loss": 1.0641,
"step": 1302
},
{
"epoch": 0.65,
"learning_rate": 1.4844173132259933e-05,
"loss": 0.9923,
"step": 1303
},
{
"epoch": 0.65,
"learning_rate": 1.4806414775307418e-05,
"loss": 1.0533,
"step": 1304
},
{
"epoch": 0.65,
"learning_rate": 1.4768684287227652e-05,
"loss": 1.0136,
"step": 1305
},
{
"epoch": 0.65,
"learning_rate": 1.4730981771174369e-05,
"loss": 1.0206,
"step": 1306
},
{
"epoch": 0.65,
"learning_rate": 1.4693307330224798e-05,
"loss": 1.0024,
"step": 1307
},
{
"epoch": 0.65,
"learning_rate": 1.465566106737942e-05,
"loss": 1.047,
"step": 1308
},
{
"epoch": 0.65,
"learning_rate": 1.4618043085561702e-05,
"loss": 0.9761,
"step": 1309
},
{
"epoch": 0.66,
"learning_rate": 1.4580453487617745e-05,
"loss": 1.0514,
"step": 1310
},
{
"epoch": 0.66,
"learning_rate": 1.4542892376316078e-05,
"loss": 1.059,
"step": 1311
},
{
"epoch": 0.66,
"learning_rate": 1.4505359854347361e-05,
"loss": 1.0643,
"step": 1312
},
{
"epoch": 0.66,
"learning_rate": 1.4467856024324056e-05,
"loss": 1.0965,
"step": 1313
},
{
"epoch": 0.66,
"learning_rate": 1.4430380988780218e-05,
"loss": 1.1107,
"step": 1314
},
{
"epoch": 0.66,
"learning_rate": 1.439293485017116e-05,
"loss": 1.0131,
"step": 1315
},
{
"epoch": 0.66,
"learning_rate": 1.4355517710873184e-05,
"loss": 1.0364,
"step": 1316
},
{
"epoch": 0.66,
"learning_rate": 1.4318129673183333e-05,
"loss": 1.0873,
"step": 1317
},
{
"epoch": 0.66,
"learning_rate": 1.4280770839319071e-05,
"loss": 1.03,
"step": 1318
},
{
"epoch": 0.66,
"learning_rate": 1.4243441311418013e-05,
"loss": 1.0264,
"step": 1319
},
{
"epoch": 0.66,
"learning_rate": 1.4206141191537682e-05,
"loss": 1.0223,
"step": 1320
},
{
"epoch": 0.66,
"learning_rate": 1.4168870581655159e-05,
"loss": 1.022,
"step": 1321
},
{
"epoch": 0.66,
"learning_rate": 1.4131629583666888e-05,
"loss": 1.0765,
"step": 1322
},
{
"epoch": 0.66,
"learning_rate": 1.4094418299388331e-05,
"loss": 1.0306,
"step": 1323
},
{
"epoch": 0.66,
"learning_rate": 1.4057236830553704e-05,
"loss": 1.0315,
"step": 1324
},
{
"epoch": 0.66,
"learning_rate": 1.4020085278815745e-05,
"loss": 1.0226,
"step": 1325
},
{
"epoch": 0.66,
"learning_rate": 1.3982963745745352e-05,
"loss": 1.0242,
"step": 1326
},
{
"epoch": 0.66,
"learning_rate": 1.3945872332831412e-05,
"loss": 1.0382,
"step": 1327
},
{
"epoch": 0.66,
"learning_rate": 1.3908811141480408e-05,
"loss": 1.0098,
"step": 1328
},
{
"epoch": 0.66,
"learning_rate": 1.3871780273016215e-05,
"loss": 1.0667,
"step": 1329
},
{
"epoch": 0.67,
"learning_rate": 1.383477982867984e-05,
"loss": 1.0581,
"step": 1330
},
{
"epoch": 0.67,
"learning_rate": 1.3797809909629058e-05,
"loss": 1.0291,
"step": 1331
},
{
"epoch": 0.67,
"learning_rate": 1.3760870616938248e-05,
"loss": 1.0711,
"step": 1332
},
{
"epoch": 0.67,
"learning_rate": 1.3723962051597988e-05,
"loss": 1.0387,
"step": 1333
},
{
"epoch": 0.67,
"learning_rate": 1.3687084314514907e-05,
"loss": 1.097,
"step": 1334
},
{
"epoch": 0.67,
"learning_rate": 1.3650237506511331e-05,
"loss": 1.0019,
"step": 1335
},
{
"epoch": 0.67,
"learning_rate": 1.3613421728325018e-05,
"loss": 1.0201,
"step": 1336
},
{
"epoch": 0.67,
"learning_rate": 1.3576637080608923e-05,
"loss": 1.0126,
"step": 1337
},
{
"epoch": 0.67,
"learning_rate": 1.353988366393083e-05,
"loss": 1.0421,
"step": 1338
},
{
"epoch": 0.67,
"learning_rate": 1.3503161578773193e-05,
"loss": 1.0552,
"step": 1339
},
{
"epoch": 0.67,
"learning_rate": 1.346647092553281e-05,
"loss": 1.0585,
"step": 1340
},
{
"epoch": 0.67,
"learning_rate": 1.3429811804520492e-05,
"loss": 1.0129,
"step": 1341
},
{
"epoch": 0.67,
"learning_rate": 1.3393184315960918e-05,
"loss": 1.0877,
"step": 1342
},
{
"epoch": 0.67,
"learning_rate": 1.33565885599922e-05,
"loss": 1.0335,
"step": 1343
},
{
"epoch": 0.67,
"learning_rate": 1.3320024636665757e-05,
"loss": 1.0139,
"step": 1344
},
{
"epoch": 0.67,
"learning_rate": 1.3283492645945966e-05,
"loss": 1.0425,
"step": 1345
},
{
"epoch": 0.67,
"learning_rate": 1.3246992687709889e-05,
"loss": 1.0431,
"step": 1346
},
{
"epoch": 0.67,
"learning_rate": 1.3210524861747015e-05,
"loss": 1.0346,
"step": 1347
},
{
"epoch": 0.67,
"learning_rate": 1.3174089267758983e-05,
"loss": 1.0184,
"step": 1348
},
{
"epoch": 0.67,
"learning_rate": 1.3137686005359318e-05,
"loss": 1.0431,
"step": 1349
},
{
"epoch": 0.68,
"learning_rate": 1.3101315174073162e-05,
"loss": 1.0161,
"step": 1350
},
{
"epoch": 0.68,
"learning_rate": 1.306497687333697e-05,
"loss": 1.0257,
"step": 1351
},
{
"epoch": 0.68,
"learning_rate": 1.3028671202498261e-05,
"loss": 1.0514,
"step": 1352
},
{
"epoch": 0.68,
"learning_rate": 1.2992398260815369e-05,
"loss": 1.0614,
"step": 1353
},
{
"epoch": 0.68,
"learning_rate": 1.2956158147457115e-05,
"loss": 1.0438,
"step": 1354
},
{
"epoch": 0.68,
"learning_rate": 1.2919950961502603e-05,
"loss": 1.0316,
"step": 1355
},
{
"epoch": 0.68,
"learning_rate": 1.2883776801940884e-05,
"loss": 1.066,
"step": 1356
},
{
"epoch": 0.68,
"learning_rate": 1.2847635767670722e-05,
"loss": 1.0638,
"step": 1357
},
{
"epoch": 0.68,
"learning_rate": 1.2811527957500345e-05,
"loss": 1.0276,
"step": 1358
},
{
"epoch": 0.68,
"learning_rate": 1.2775453470147106e-05,
"loss": 1.0446,
"step": 1359
},
{
"epoch": 0.68,
"learning_rate": 1.2739412404237306e-05,
"loss": 1.0407,
"step": 1360
},
{
"epoch": 0.68,
"learning_rate": 1.2703404858305806e-05,
"loss": 1.0204,
"step": 1361
},
{
"epoch": 0.68,
"learning_rate": 1.2667430930795877e-05,
"loss": 1.0152,
"step": 1362
},
{
"epoch": 0.68,
"learning_rate": 1.2631490720058875e-05,
"loss": 1.0735,
"step": 1363
},
{
"epoch": 0.68,
"learning_rate": 1.2595584324353943e-05,
"loss": 1.0463,
"step": 1364
},
{
"epoch": 0.68,
"learning_rate": 1.255971184184783e-05,
"loss": 1.0835,
"step": 1365
},
{
"epoch": 0.68,
"learning_rate": 1.2523873370614489e-05,
"loss": 1.0654,
"step": 1366
},
{
"epoch": 0.68,
"learning_rate": 1.2488069008634954e-05,
"loss": 1.014,
"step": 1367
},
{
"epoch": 0.68,
"learning_rate": 1.245229885379699e-05,
"loss": 1.0756,
"step": 1368
},
{
"epoch": 0.68,
"learning_rate": 1.2416563003894826e-05,
"loss": 1.0627,
"step": 1369
},
{
"epoch": 0.69,
"learning_rate": 1.2380861556628915e-05,
"loss": 1.0369,
"step": 1370
},
{
"epoch": 0.69,
"learning_rate": 1.2345194609605636e-05,
"loss": 1.1073,
"step": 1371
},
{
"epoch": 0.69,
"learning_rate": 1.2309562260337073e-05,
"loss": 1.0814,
"step": 1372
},
{
"epoch": 0.69,
"learning_rate": 1.2273964606240718e-05,
"loss": 1.0399,
"step": 1373
},
{
"epoch": 0.69,
"learning_rate": 1.2238401744639185e-05,
"loss": 1.0231,
"step": 1374
},
{
"epoch": 0.69,
"learning_rate": 1.2202873772759981e-05,
"loss": 0.9938,
"step": 1375
},
{
"epoch": 0.69,
"learning_rate": 1.216738078773522e-05,
"loss": 1.0156,
"step": 1376
},
{
"epoch": 0.69,
"learning_rate": 1.2131922886601368e-05,
"loss": 1.0335,
"step": 1377
},
{
"epoch": 0.69,
"learning_rate": 1.209650016629899e-05,
"loss": 1.0594,
"step": 1378
},
{
"epoch": 0.69,
"learning_rate": 1.2061112723672438e-05,
"loss": 1.0885,
"step": 1379
},
{
"epoch": 0.69,
"learning_rate": 1.202576065546963e-05,
"loss": 1.0391,
"step": 1380
},
{
"epoch": 0.69,
"learning_rate": 1.1990444058341757e-05,
"loss": 1.0322,
"step": 1381
},
{
"epoch": 0.69,
"learning_rate": 1.1955163028843063e-05,
"loss": 1.0315,
"step": 1382
},
{
"epoch": 0.69,
"learning_rate": 1.1919917663430552e-05,
"loss": 0.9996,
"step": 1383
},
{
"epoch": 0.69,
"learning_rate": 1.1884708058463668e-05,
"loss": 1.01,
"step": 1384
},
{
"epoch": 0.69,
"learning_rate": 1.1849534310204152e-05,
"loss": 1.0394,
"step": 1385
},
{
"epoch": 0.69,
"learning_rate": 1.1814396514815676e-05,
"loss": 1.0346,
"step": 1386
},
{
"epoch": 0.69,
"learning_rate": 1.1779294768363639e-05,
"loss": 1.0432,
"step": 1387
},
{
"epoch": 0.69,
"learning_rate": 1.1744229166814888e-05,
"loss": 1.088,
"step": 1388
},
{
"epoch": 0.69,
"learning_rate": 1.170919980603741e-05,
"loss": 1.0689,
"step": 1389
},
{
"epoch": 0.69,
"learning_rate": 1.1674206781800162e-05,
"loss": 1.0677,
"step": 1390
},
{
"epoch": 0.7,
"learning_rate": 1.1639250189772718e-05,
"loss": 1.042,
"step": 1391
},
{
"epoch": 0.7,
"learning_rate": 1.1604330125525079e-05,
"loss": 1.0653,
"step": 1392
},
{
"epoch": 0.7,
"learning_rate": 1.1569446684527383e-05,
"loss": 1.0066,
"step": 1393
},
{
"epoch": 0.7,
"learning_rate": 1.1534599962149587e-05,
"loss": 1.0591,
"step": 1394
},
{
"epoch": 0.7,
"learning_rate": 1.1499790053661327e-05,
"loss": 1.0406,
"step": 1395
},
{
"epoch": 0.7,
"learning_rate": 1.146501705423155e-05,
"loss": 1.0649,
"step": 1396
},
{
"epoch": 0.7,
"learning_rate": 1.1430281058928324e-05,
"loss": 1.02,
"step": 1397
},
{
"epoch": 0.7,
"learning_rate": 1.1395582162718525e-05,
"loss": 1.0406,
"step": 1398
},
{
"epoch": 0.7,
"learning_rate": 1.1360920460467598e-05,
"loss": 1.0342,
"step": 1399
},
{
"epoch": 0.7,
"learning_rate": 1.1326296046939333e-05,
"loss": 1.0464,
"step": 1400
},
{
"epoch": 0.7,
"eval_acceptance_rate": 0.8042372465133667,
"eval_acceptance_rate_1": 0.8035486340522766,
"eval_acceptance_rate_2": 0.8043899536132812,
"eval_acceptance_rate_3": 0.8036088347434998,
"eval_acceptance_rate_4": 0.8041674494743347,
"eval_avg_generation_length": 215.4989013671875,
"eval_avg_input_length": 365.6648864746094,
"eval_expected_speedup": 2.711648435017575,
"eval_expected_tokens": 3.389560543771969,
"eval_loss": 0.940096914768219,
"eval_max_generation_length": 424.7227783203125,
"eval_max_input_length": 630.4989624023438,
"eval_min_generation_length": 59.60786437988281,
"eval_min_input_length": 160.72378540039062,
"eval_pulsar_acceptance_rate": 0.5973901359429922,
"eval_runtime": 123.2454,
"eval_samples_per_second": 4.016,
"eval_steps_per_second": 0.13,
"step": 1400
},
{
"epoch": 0.7,
"learning_rate": 1.129170901679554e-05,
"loss": 1.042,
"step": 1401
},
{
"epoch": 0.7,
"learning_rate": 1.1257159464595855e-05,
"loss": 1.0294,
"step": 1402
},
{
"epoch": 0.7,
"learning_rate": 1.1222647484797422e-05,
"loss": 1.0263,
"step": 1403
},
{
"epoch": 0.7,
"learning_rate": 1.118817317175467e-05,
"loss": 1.0384,
"step": 1404
},
{
"epoch": 0.7,
"learning_rate": 1.1153736619719077e-05,
"loss": 1.0327,
"step": 1405
},
{
"epoch": 0.7,
"learning_rate": 1.1119337922838832e-05,
"loss": 1.0409,
"step": 1406
},
{
"epoch": 0.7,
"learning_rate": 1.1084977175158687e-05,
"loss": 1.0396,
"step": 1407
},
{
"epoch": 0.7,
"learning_rate": 1.1050654470619601e-05,
"loss": 1.1275,
"step": 1408
},
{
"epoch": 0.7,
"learning_rate": 1.1016369903058529e-05,
"loss": 1.0538,
"step": 1409
},
{
"epoch": 0.7,
"learning_rate": 1.0982123566208185e-05,
"loss": 1.0855,
"step": 1410
},
{
"epoch": 0.71,
"learning_rate": 1.0947915553696742e-05,
"loss": 0.9956,
"step": 1411
},
{
"epoch": 0.71,
"learning_rate": 1.091374595904759e-05,
"loss": 1.0467,
"step": 1412
},
{
"epoch": 0.71,
"learning_rate": 1.0879614875679109e-05,
"loss": 1.0299,
"step": 1413
},
{
"epoch": 0.71,
"learning_rate": 1.0845522396904367e-05,
"loss": 1.0187,
"step": 1414
},
{
"epoch": 0.71,
"learning_rate": 1.0811468615930911e-05,
"loss": 1.0109,
"step": 1415
},
{
"epoch": 0.71,
"learning_rate": 1.0777453625860472e-05,
"loss": 1.0297,
"step": 1416
},
{
"epoch": 0.71,
"learning_rate": 1.0743477519688725e-05,
"loss": 1.0798,
"step": 1417
},
{
"epoch": 0.71,
"learning_rate": 1.0709540390305061e-05,
"loss": 1.0548,
"step": 1418
},
{
"epoch": 0.71,
"learning_rate": 1.0675642330492286e-05,
"loss": 1.0965,
"step": 1419
},
{
"epoch": 0.71,
"learning_rate": 1.064178343292641e-05,
"loss": 1.0221,
"step": 1420
},
{
"epoch": 0.71,
"learning_rate": 1.0607963790176365e-05,
"loss": 1.0737,
"step": 1421
},
{
"epoch": 0.71,
"learning_rate": 1.0574183494703748e-05,
"loss": 1.0514,
"step": 1422
},
{
"epoch": 0.71,
"learning_rate": 1.0540442638862618e-05,
"loss": 1.096,
"step": 1423
},
{
"epoch": 0.71,
"learning_rate": 1.0506741314899166e-05,
"loss": 1.0143,
"step": 1424
},
{
"epoch": 0.71,
"learning_rate": 1.0473079614951545e-05,
"loss": 1.0495,
"step": 1425
},
{
"epoch": 0.71,
"learning_rate": 1.0439457631049549e-05,
"loss": 1.0386,
"step": 1426
},
{
"epoch": 0.71,
"learning_rate": 1.0405875455114387e-05,
"loss": 1.1013,
"step": 1427
},
{
"epoch": 0.71,
"learning_rate": 1.0372333178958462e-05,
"loss": 1.0383,
"step": 1428
},
{
"epoch": 0.71,
"learning_rate": 1.0338830894285065e-05,
"loss": 1.0655,
"step": 1429
},
{
"epoch": 0.71,
"learning_rate": 1.0305368692688174e-05,
"loss": 1.0374,
"step": 1430
},
{
"epoch": 0.72,
"learning_rate": 1.0271946665652166e-05,
"loss": 1.088,
"step": 1431
},
{
"epoch": 0.72,
"learning_rate": 1.0238564904551574e-05,
"loss": 1.0178,
"step": 1432
},
{
"epoch": 0.72,
"learning_rate": 1.0205223500650876e-05,
"loss": 1.0377,
"step": 1433
},
{
"epoch": 0.72,
"learning_rate": 1.0171922545104184e-05,
"loss": 1.0714,
"step": 1434
},
{
"epoch": 0.72,
"learning_rate": 1.0138662128955053e-05,
"loss": 1.0635,
"step": 1435
},
{
"epoch": 0.72,
"learning_rate": 1.0105442343136184e-05,
"loss": 1.0433,
"step": 1436
},
{
"epoch": 0.72,
"learning_rate": 1.0072263278469194e-05,
"loss": 0.9967,
"step": 1437
},
{
"epoch": 0.72,
"learning_rate": 1.0039125025664392e-05,
"loss": 1.0536,
"step": 1438
},
{
"epoch": 0.72,
"learning_rate": 1.0006027675320493e-05,
"loss": 1.0511,
"step": 1439
},
{
"epoch": 0.72,
"learning_rate": 9.972971317924374e-06,
"loss": 1.0873,
"step": 1440
},
{
"epoch": 0.72,
"learning_rate": 9.939956043850876e-06,
"loss": 1.0639,
"step": 1441
},
{
"epoch": 0.72,
"learning_rate": 9.90698194336248e-06,
"loss": 1.0858,
"step": 1442
},
{
"epoch": 0.72,
"learning_rate": 9.874049106609135e-06,
"loss": 1.0387,
"step": 1443
},
{
"epoch": 0.72,
"learning_rate": 9.841157623627947e-06,
"loss": 1.0666,
"step": 1444
},
{
"epoch": 0.72,
"learning_rate": 9.808307584342971e-06,
"loss": 1.0573,
"step": 1445
},
{
"epoch": 0.72,
"learning_rate": 9.775499078564973e-06,
"loss": 1.0618,
"step": 1446
},
{
"epoch": 0.72,
"learning_rate": 9.742732195991142e-06,
"loss": 1.0088,
"step": 1447
},
{
"epoch": 0.72,
"learning_rate": 9.710007026204895e-06,
"loss": 1.0615,
"step": 1448
},
{
"epoch": 0.72,
"learning_rate": 9.677323658675594e-06,
"loss": 1.0663,
"step": 1449
},
{
"epoch": 0.72,
"learning_rate": 9.644682182758306e-06,
"loss": 1.0105,
"step": 1450
},
{
"epoch": 0.73,
"learning_rate": 9.612082687693597e-06,
"loss": 1.0695,
"step": 1451
},
{
"epoch": 0.73,
"learning_rate": 9.579525262607226e-06,
"loss": 1.0658,
"step": 1452
},
{
"epoch": 0.73,
"learning_rate": 9.547009996509964e-06,
"loss": 1.0246,
"step": 1453
},
{
"epoch": 0.73,
"learning_rate": 9.514536978297303e-06,
"loss": 0.9943,
"step": 1454
},
{
"epoch": 0.73,
"learning_rate": 9.482106296749221e-06,
"loss": 1.0844,
"step": 1455
},
{
"epoch": 0.73,
"learning_rate": 9.449718040529987e-06,
"loss": 1.1073,
"step": 1456
},
{
"epoch": 0.73,
"learning_rate": 9.417372298187833e-06,
"loss": 1.0455,
"step": 1457
},
{
"epoch": 0.73,
"learning_rate": 9.385069158154805e-06,
"loss": 1.0369,
"step": 1458
},
{
"epoch": 0.73,
"learning_rate": 9.352808708746441e-06,
"loss": 1.0404,
"step": 1459
},
{
"epoch": 0.73,
"learning_rate": 9.320591038161574e-06,
"loss": 1.0611,
"step": 1460
},
{
"epoch": 0.73,
"learning_rate": 9.2884162344821e-06,
"loss": 1.0744,
"step": 1461
},
{
"epoch": 0.73,
"learning_rate": 9.25628438567269e-06,
"loss": 1.0264,
"step": 1462
},
{
"epoch": 0.73,
"learning_rate": 9.224195579580602e-06,
"loss": 1.0279,
"step": 1463
},
{
"epoch": 0.73,
"learning_rate": 9.192149903935405e-06,
"loss": 1.0402,
"step": 1464
},
{
"epoch": 0.73,
"learning_rate": 9.160147446348739e-06,
"loss": 1.0319,
"step": 1465
},
{
"epoch": 0.73,
"learning_rate": 9.128188294314119e-06,
"loss": 1.0546,
"step": 1466
},
{
"epoch": 0.73,
"learning_rate": 9.096272535206641e-06,
"loss": 1.0245,
"step": 1467
},
{
"epoch": 0.73,
"learning_rate": 9.064400256282757e-06,
"loss": 1.0581,
"step": 1468
},
{
"epoch": 0.73,
"learning_rate": 9.032571544680086e-06,
"loss": 1.0473,
"step": 1469
},
{
"epoch": 0.73,
"learning_rate": 9.000786487417085e-06,
"loss": 1.1013,
"step": 1470
},
{
"epoch": 0.74,
"learning_rate": 8.969045171392909e-06,
"loss": 1.0264,
"step": 1471
},
{
"epoch": 0.74,
"learning_rate": 8.937347683387095e-06,
"loss": 1.0607,
"step": 1472
},
{
"epoch": 0.74,
"learning_rate": 8.905694110059353e-06,
"loss": 1.0356,
"step": 1473
},
{
"epoch": 0.74,
"learning_rate": 8.874084537949364e-06,
"loss": 1.0532,
"step": 1474
},
{
"epoch": 0.74,
"learning_rate": 8.842519053476476e-06,
"loss": 1.0576,
"step": 1475
},
{
"epoch": 0.74,
"learning_rate": 8.810997742939531e-06,
"loss": 1.0658,
"step": 1476
},
{
"epoch": 0.74,
"learning_rate": 8.77952069251658e-06,
"loss": 1.0597,
"step": 1477
},
{
"epoch": 0.74,
"learning_rate": 8.74808798826467e-06,
"loss": 1.0723,
"step": 1478
},
{
"epoch": 0.74,
"learning_rate": 8.71669971611963e-06,
"loss": 1.0384,
"step": 1479
},
{
"epoch": 0.74,
"learning_rate": 8.685355961895784e-06,
"loss": 1.0108,
"step": 1480
},
{
"epoch": 0.74,
"learning_rate": 8.654056811285772e-06,
"loss": 1.0575,
"step": 1481
},
{
"epoch": 0.74,
"learning_rate": 8.622802349860268e-06,
"loss": 1.0468,
"step": 1482
},
{
"epoch": 0.74,
"learning_rate": 8.591592663067771e-06,
"loss": 1.0431,
"step": 1483
},
{
"epoch": 0.74,
"learning_rate": 8.56042783623439e-06,
"loss": 1.0367,
"step": 1484
},
{
"epoch": 0.74,
"learning_rate": 8.52930795456355e-06,
"loss": 1.0272,
"step": 1485
},
{
"epoch": 0.74,
"learning_rate": 8.49823310313584e-06,
"loss": 1.0274,
"step": 1486
},
{
"epoch": 0.74,
"learning_rate": 8.467203366908707e-06,
"loss": 1.0355,
"step": 1487
},
{
"epoch": 0.74,
"learning_rate": 8.436218830716258e-06,
"loss": 1.0592,
"step": 1488
},
{
"epoch": 0.74,
"learning_rate": 8.405279579269046e-06,
"loss": 1.0564,
"step": 1489
},
{
"epoch": 0.74,
"learning_rate": 8.374385697153792e-06,
"loss": 1.0333,
"step": 1490
},
{
"epoch": 0.75,
"learning_rate": 8.343537268833199e-06,
"loss": 1.0266,
"step": 1491
},
{
"epoch": 0.75,
"learning_rate": 8.31273437864569e-06,
"loss": 1.0348,
"step": 1492
},
{
"epoch": 0.75,
"learning_rate": 8.281977110805177e-06,
"loss": 1.0625,
"step": 1493
},
{
"epoch": 0.75,
"learning_rate": 8.251265549400877e-06,
"loss": 1.0341,
"step": 1494
},
{
"epoch": 0.75,
"learning_rate": 8.220599778397017e-06,
"loss": 1.0775,
"step": 1495
},
{
"epoch": 0.75,
"learning_rate": 8.189979881632634e-06,
"loss": 1.0235,
"step": 1496
},
{
"epoch": 0.75,
"learning_rate": 8.159405942821375e-06,
"loss": 1.0356,
"step": 1497
},
{
"epoch": 0.75,
"learning_rate": 8.1288780455512e-06,
"loss": 1.032,
"step": 1498
},
{
"epoch": 0.75,
"learning_rate": 8.098396273284236e-06,
"loss": 1.0012,
"step": 1499
},
{
"epoch": 0.75,
"learning_rate": 8.067960709356478e-06,
"loss": 1.0359,
"step": 1500
},
{
"epoch": 0.75,
"eval_acceptance_rate": 0.8049868941307068,
"eval_acceptance_rate_1": 0.8042713403701782,
"eval_acceptance_rate_2": 0.8051593899726868,
"eval_acceptance_rate_3": 0.8043556213378906,
"eval_acceptance_rate_4": 0.8048893213272095,
"eval_avg_generation_length": 215.4989013671875,
"eval_avg_input_length": 365.6648864746094,
"eval_expected_speedup": 2.7156275956240408,
"eval_expected_tokens": 3.3945344945300513,
"eval_loss": 0.938217043876648,
"eval_max_generation_length": 424.7227783203125,
"eval_max_input_length": 630.4989624023438,
"eval_min_generation_length": 59.60786437988281,
"eval_min_input_length": 160.72378540039062,
"eval_pulsar_acceptance_rate": 0.5986336236325127,
"eval_runtime": 120.8862,
"eval_samples_per_second": 4.095,
"eval_steps_per_second": 0.132,
"step": 1500
},
{
"epoch": 0.75,
"learning_rate": 8.037571436977582e-06,
"loss": 1.0471,
"step": 1501
},
{
"epoch": 0.75,
"learning_rate": 8.007228539230676e-06,
"loss": 0.9997,
"step": 1502
},
{
"epoch": 0.75,
"learning_rate": 7.976932099072068e-06,
"loss": 1.0484,
"step": 1503
},
{
"epoch": 0.75,
"learning_rate": 7.946682199331088e-06,
"loss": 1.0266,
"step": 1504
},
{
"epoch": 0.75,
"learning_rate": 7.91647892270979e-06,
"loss": 1.0376,
"step": 1505
},
{
"epoch": 0.75,
"learning_rate": 7.886322351782783e-06,
"loss": 1.0515,
"step": 1506
},
{
"epoch": 0.75,
"learning_rate": 7.856212568996987e-06,
"loss": 1.0684,
"step": 1507
},
{
"epoch": 0.75,
"learning_rate": 7.826149656671386e-06,
"loss": 1.0676,
"step": 1508
},
{
"epoch": 0.75,
"learning_rate": 7.796133696996858e-06,
"loss": 1.0405,
"step": 1509
},
{
"epoch": 0.76,
"learning_rate": 7.766164772035856e-06,
"loss": 1.0213,
"step": 1510
},
{
"epoch": 0.76,
"learning_rate": 7.736242963722299e-06,
"loss": 1.0327,
"step": 1511
},
{
"epoch": 0.76,
"learning_rate": 7.706368353861269e-06,
"loss": 1.0809,
"step": 1512
},
{
"epoch": 0.76,
"learning_rate": 7.676541024128798e-06,
"loss": 1.0638,
"step": 1513
},
{
"epoch": 0.76,
"learning_rate": 7.646761056071686e-06,
"loss": 1.0103,
"step": 1514
},
{
"epoch": 0.76,
"learning_rate": 7.617028531107201e-06,
"loss": 1.0343,
"step": 1515
},
{
"epoch": 0.76,
"learning_rate": 7.587343530522945e-06,
"loss": 1.0095,
"step": 1516
},
{
"epoch": 0.76,
"learning_rate": 7.5577061354765835e-06,
"loss": 1.0968,
"step": 1517
},
{
"epoch": 0.76,
"learning_rate": 7.528116426995604e-06,
"loss": 1.0377,
"step": 1518
},
{
"epoch": 0.76,
"learning_rate": 7.498574485977172e-06,
"loss": 1.0598,
"step": 1519
},
{
"epoch": 0.76,
"learning_rate": 7.469080393187786e-06,
"loss": 1.0418,
"step": 1520
},
{
"epoch": 0.76,
"learning_rate": 7.4396342292631895e-06,
"loss": 1.014,
"step": 1521
},
{
"epoch": 0.76,
"learning_rate": 7.410236074708077e-06,
"loss": 1.0752,
"step": 1522
},
{
"epoch": 0.76,
"learning_rate": 7.380886009895874e-06,
"loss": 1.0723,
"step": 1523
},
{
"epoch": 0.76,
"learning_rate": 7.351584115068535e-06,
"loss": 1.0302,
"step": 1524
},
{
"epoch": 0.76,
"learning_rate": 7.3223304703363135e-06,
"loss": 1.0601,
"step": 1525
},
{
"epoch": 0.76,
"learning_rate": 7.293125155677566e-06,
"loss": 1.0232,
"step": 1526
},
{
"epoch": 0.76,
"learning_rate": 7.263968250938516e-06,
"loss": 1.0432,
"step": 1527
},
{
"epoch": 0.76,
"learning_rate": 7.234859835833021e-06,
"loss": 1.0607,
"step": 1528
},
{
"epoch": 0.76,
"learning_rate": 7.205799989942372e-06,
"loss": 1.073,
"step": 1529
},
{
"epoch": 0.77,
"learning_rate": 7.176788792715075e-06,
"loss": 1.047,
"step": 1530
},
{
"epoch": 0.77,
"learning_rate": 7.147826323466638e-06,
"loss": 1.0289,
"step": 1531
},
{
"epoch": 0.77,
"learning_rate": 7.118912661379368e-06,
"loss": 1.0914,
"step": 1532
},
{
"epoch": 0.77,
"learning_rate": 7.090047885502077e-06,
"loss": 1.012,
"step": 1533
},
{
"epoch": 0.77,
"learning_rate": 7.061232074749985e-06,
"loss": 1.0762,
"step": 1534
},
{
"epoch": 0.77,
"learning_rate": 7.032465307904404e-06,
"loss": 1.029,
"step": 1535
},
{
"epoch": 0.77,
"learning_rate": 7.003747663612581e-06,
"loss": 1.0638,
"step": 1536
},
{
"epoch": 0.77,
"learning_rate": 6.9750792203874785e-06,
"loss": 1.0181,
"step": 1537
},
{
"epoch": 0.77,
"learning_rate": 6.94646005660749e-06,
"loss": 1.0649,
"step": 1538
},
{
"epoch": 0.77,
"learning_rate": 6.91789025051634e-06,
"loss": 1.0724,
"step": 1539
},
{
"epoch": 0.77,
"learning_rate": 6.889369880222776e-06,
"loss": 1.0353,
"step": 1540
},
{
"epoch": 0.77,
"learning_rate": 6.860899023700407e-06,
"loss": 1.0466,
"step": 1541
},
{
"epoch": 0.77,
"learning_rate": 6.832477758787484e-06,
"loss": 1.0394,
"step": 1542
},
{
"epoch": 0.77,
"learning_rate": 6.8041061631866245e-06,
"loss": 0.9814,
"step": 1543
},
{
"epoch": 0.77,
"learning_rate": 6.775784314464717e-06,
"loss": 1.0304,
"step": 1544
},
{
"epoch": 0.77,
"learning_rate": 6.747512290052596e-06,
"loss": 1.0376,
"step": 1545
},
{
"epoch": 0.77,
"learning_rate": 6.71929016724491e-06,
"loss": 1.0842,
"step": 1546
},
{
"epoch": 0.77,
"learning_rate": 6.691118023199861e-06,
"loss": 1.0342,
"step": 1547
},
{
"epoch": 0.77,
"learning_rate": 6.662995934939007e-06,
"loss": 1.0171,
"step": 1548
},
{
"epoch": 0.77,
"learning_rate": 6.634923979347074e-06,
"loss": 1.0236,
"step": 1549
},
{
"epoch": 0.78,
"learning_rate": 6.606902233171711e-06,
"loss": 1.0852,
"step": 1550
},
{
"epoch": 0.78,
"learning_rate": 6.5789307730233065e-06,
"loss": 1.0514,
"step": 1551
},
{
"epoch": 0.78,
"learning_rate": 6.551009675374764e-06,
"loss": 1.0344,
"step": 1552
},
{
"epoch": 0.78,
"learning_rate": 6.5231390165612884e-06,
"loss": 1.0183,
"step": 1553
},
{
"epoch": 0.78,
"learning_rate": 6.49531887278021e-06,
"loss": 1.0296,
"step": 1554
},
{
"epoch": 0.78,
"learning_rate": 6.46754932009073e-06,
"loss": 1.0207,
"step": 1555
},
{
"epoch": 0.78,
"learning_rate": 6.439830434413754e-06,
"loss": 1.0031,
"step": 1556
},
{
"epoch": 0.78,
"learning_rate": 6.412162291531651e-06,
"loss": 1.0529,
"step": 1557
},
{
"epoch": 0.78,
"learning_rate": 6.384544967088063e-06,
"loss": 1.0185,
"step": 1558
},
{
"epoch": 0.78,
"learning_rate": 6.3569785365877125e-06,
"loss": 1.0528,
"step": 1559
},
{
"epoch": 0.78,
"learning_rate": 6.329463075396161e-06,
"loss": 1.0372,
"step": 1560
},
{
"epoch": 0.78,
"learning_rate": 6.301998658739619e-06,
"loss": 1.0403,
"step": 1561
},
{
"epoch": 0.78,
"learning_rate": 6.274585361704774e-06,
"loss": 1.0474,
"step": 1562
},
{
"epoch": 0.78,
"learning_rate": 6.247223259238511e-06,
"loss": 1.0359,
"step": 1563
},
{
"epoch": 0.78,
"learning_rate": 6.219912426147795e-06,
"loss": 1.0536,
"step": 1564
},
{
"epoch": 0.78,
"learning_rate": 6.192652937099388e-06,
"loss": 1.0763,
"step": 1565
},
{
"epoch": 0.78,
"learning_rate": 6.165444866619685e-06,
"loss": 1.0357,
"step": 1566
},
{
"epoch": 0.78,
"learning_rate": 6.138288289094532e-06,
"loss": 1.0304,
"step": 1567
},
{
"epoch": 0.78,
"learning_rate": 6.111183278768956e-06,
"loss": 1.0305,
"step": 1568
},
{
"epoch": 0.78,
"learning_rate": 6.084129909747033e-06,
"loss": 1.0493,
"step": 1569
},
{
"epoch": 0.79,
"learning_rate": 6.057128255991637e-06,
"loss": 1.0175,
"step": 1570
},
{
"epoch": 0.79,
"learning_rate": 6.030178391324251e-06,
"loss": 1.038,
"step": 1571
},
{
"epoch": 0.79,
"learning_rate": 6.003280389424789e-06,
"loss": 1.0722,
"step": 1572
},
{
"epoch": 0.79,
"learning_rate": 5.976434323831348e-06,
"loss": 0.9955,
"step": 1573
},
{
"epoch": 0.79,
"learning_rate": 5.9496402679400594e-06,
"loss": 1.0745,
"step": 1574
},
{
"epoch": 0.79,
"learning_rate": 5.9228982950048416e-06,
"loss": 1.0231,
"step": 1575
},
{
"epoch": 0.79,
"learning_rate": 5.896208478137222e-06,
"loss": 1.0219,
"step": 1576
},
{
"epoch": 0.79,
"learning_rate": 5.869570890306153e-06,
"loss": 1.0268,
"step": 1577
},
{
"epoch": 0.79,
"learning_rate": 5.842985604337769e-06,
"loss": 1.0198,
"step": 1578
},
{
"epoch": 0.79,
"learning_rate": 5.816452692915242e-06,
"loss": 1.0343,
"step": 1579
},
{
"epoch": 0.79,
"learning_rate": 5.78997222857853e-06,
"loss": 1.0505,
"step": 1580
},
{
"epoch": 0.79,
"learning_rate": 5.763544283724204e-06,
"loss": 1.0467,
"step": 1581
},
{
"epoch": 0.79,
"learning_rate": 5.737168930605272e-06,
"loss": 1.0451,
"step": 1582
},
{
"epoch": 0.79,
"learning_rate": 5.710846241330928e-06,
"loss": 1.0099,
"step": 1583
},
{
"epoch": 0.79,
"learning_rate": 5.684576287866411e-06,
"loss": 1.099,
"step": 1584
},
{
"epoch": 0.79,
"learning_rate": 5.6583591420327684e-06,
"loss": 1.0455,
"step": 1585
},
{
"epoch": 0.79,
"learning_rate": 5.632194875506663e-06,
"loss": 1.0892,
"step": 1586
},
{
"epoch": 0.79,
"learning_rate": 5.606083559820219e-06,
"loss": 1.019,
"step": 1587
},
{
"epoch": 0.79,
"learning_rate": 5.580025266360764e-06,
"loss": 1.0292,
"step": 1588
},
{
"epoch": 0.79,
"learning_rate": 5.554020066370677e-06,
"loss": 1.0548,
"step": 1589
},
{
"epoch": 0.8,
"learning_rate": 5.528068030947192e-06,
"loss": 1.0244,
"step": 1590
},
{
"epoch": 0.8,
"learning_rate": 5.50216923104217e-06,
"loss": 1.0455,
"step": 1591
},
{
"epoch": 0.8,
"learning_rate": 5.476323737461955e-06,
"loss": 1.0269,
"step": 1592
},
{
"epoch": 0.8,
"learning_rate": 5.45053162086713e-06,
"loss": 1.0354,
"step": 1593
},
{
"epoch": 0.8,
"learning_rate": 5.424792951772353e-06,
"loss": 1.0091,
"step": 1594
},
{
"epoch": 0.8,
"learning_rate": 5.399107800546177e-06,
"loss": 1.0317,
"step": 1595
},
{
"epoch": 0.8,
"learning_rate": 5.373476237410807e-06,
"loss": 1.0673,
"step": 1596
},
{
"epoch": 0.8,
"learning_rate": 5.347898332441975e-06,
"loss": 1.0433,
"step": 1597
},
{
"epoch": 0.8,
"learning_rate": 5.322374155568688e-06,
"loss": 1.043,
"step": 1598
},
{
"epoch": 0.8,
"learning_rate": 5.296903776573065e-06,
"loss": 1.0033,
"step": 1599
},
{
"epoch": 0.8,
"learning_rate": 5.271487265090163e-06,
"loss": 1.0322,
"step": 1600
},
{
"epoch": 0.8,
"eval_acceptance_rate": 0.8054376244544983,
"eval_acceptance_rate_1": 0.8046924471855164,
"eval_acceptance_rate_2": 0.805618166923523,
"eval_acceptance_rate_3": 0.804777204990387,
"eval_acceptance_rate_4": 0.8054026961326599,
"eval_avg_generation_length": 215.4989013671875,
"eval_avg_input_length": 365.6648864746094,
"eval_expected_speedup": 2.7180232513126814,
"eval_expected_tokens": 3.3975290641408518,
"eval_loss": 0.9369726181030273,
"eval_max_generation_length": 424.7227783203125,
"eval_max_input_length": 630.4989624023438,
"eval_min_generation_length": 59.60786437988281,
"eval_min_input_length": 160.72378540039062,
"eval_pulsar_acceptance_rate": 0.5993822660352129,
"eval_runtime": 116.5819,
"eval_samples_per_second": 4.246,
"eval_steps_per_second": 0.137,
"step": 1600
},
{
"epoch": 0.8,
"learning_rate": 5.24612469060774e-06,
"loss": 1.0502,
"step": 1601
},
{
"epoch": 0.8,
"learning_rate": 5.220816122466119e-06,
"loss": 1.0557,
"step": 1602
},
{
"epoch": 0.8,
"learning_rate": 5.195561629857953e-06,
"loss": 1.0019,
"step": 1603
},
{
"epoch": 0.8,
"learning_rate": 5.170361281828054e-06,
"loss": 0.987,
"step": 1604
},
{
"epoch": 0.8,
"learning_rate": 5.145215147273224e-06,
"loss": 1.0385,
"step": 1605
},
{
"epoch": 0.8,
"learning_rate": 5.120123294942022e-06,
"loss": 1.0036,
"step": 1606
},
{
"epoch": 0.8,
"learning_rate": 5.095085793434629e-06,
"loss": 1.0538,
"step": 1607
},
{
"epoch": 0.8,
"learning_rate": 5.070102711202607e-06,
"loss": 1.0107,
"step": 1608
},
{
"epoch": 0.8,
"learning_rate": 5.045174116548745e-06,
"loss": 1.0824,
"step": 1609
},
{
"epoch": 0.81,
"learning_rate": 5.0203000776268825e-06,
"loss": 1.0556,
"step": 1610
},
{
"epoch": 0.81,
"learning_rate": 4.995480662441679e-06,
"loss": 1.0486,
"step": 1611
},
{
"epoch": 0.81,
"learning_rate": 4.970715938848478e-06,
"loss": 1.0595,
"step": 1612
},
{
"epoch": 0.81,
"learning_rate": 4.946005974553086e-06,
"loss": 1.0412,
"step": 1613
},
{
"epoch": 0.81,
"learning_rate": 4.9213508371115935e-06,
"loss": 1.0516,
"step": 1614
},
{
"epoch": 0.81,
"learning_rate": 4.896750593930216e-06,
"loss": 1.0574,
"step": 1615
},
{
"epoch": 0.81,
"learning_rate": 4.872205312265074e-06,
"loss": 1.054,
"step": 1616
},
{
"epoch": 0.81,
"learning_rate": 4.847715059222024e-06,
"loss": 1.0704,
"step": 1617
},
{
"epoch": 0.81,
"learning_rate": 4.823279901756497e-06,
"loss": 1.0938,
"step": 1618
},
{
"epoch": 0.81,
"learning_rate": 4.798899906673263e-06,
"loss": 1.0153,
"step": 1619
},
{
"epoch": 0.81,
"learning_rate": 4.7745751406263165e-06,
"loss": 1.0324,
"step": 1620
},
{
"epoch": 0.81,
"learning_rate": 4.750305670118629e-06,
"loss": 1.0414,
"step": 1621
},
{
"epoch": 0.81,
"learning_rate": 4.726091561502e-06,
"loss": 1.0737,
"step": 1622
},
{
"epoch": 0.81,
"learning_rate": 4.7019328809768895e-06,
"loss": 1.0583,
"step": 1623
},
{
"epoch": 0.81,
"learning_rate": 4.677829694592198e-06,
"loss": 1.0573,
"step": 1624
},
{
"epoch": 0.81,
"learning_rate": 4.653782068245127e-06,
"loss": 1.0613,
"step": 1625
},
{
"epoch": 0.81,
"learning_rate": 4.629790067680964e-06,
"loss": 1.0058,
"step": 1626
},
{
"epoch": 0.81,
"learning_rate": 4.605853758492914e-06,
"loss": 1.0071,
"step": 1627
},
{
"epoch": 0.81,
"learning_rate": 4.581973206121948e-06,
"loss": 1.0141,
"step": 1628
},
{
"epoch": 0.81,
"learning_rate": 4.5581484758565665e-06,
"loss": 1.0345,
"step": 1629
},
{
"epoch": 0.81,
"learning_rate": 4.534379632832692e-06,
"loss": 1.0637,
"step": 1630
},
{
"epoch": 0.82,
"learning_rate": 4.51066674203342e-06,
"loss": 1.015,
"step": 1631
},
{
"epoch": 0.82,
"learning_rate": 4.487009868288888e-06,
"loss": 1.0177,
"step": 1632
},
{
"epoch": 0.82,
"learning_rate": 4.463409076276095e-06,
"loss": 1.0255,
"step": 1633
},
{
"epoch": 0.82,
"learning_rate": 4.439864430518692e-06,
"loss": 0.9817,
"step": 1634
},
{
"epoch": 0.82,
"learning_rate": 4.416375995386857e-06,
"loss": 1.0544,
"step": 1635
},
{
"epoch": 0.82,
"learning_rate": 4.392943835097069e-06,
"loss": 1.0541,
"step": 1636
},
{
"epoch": 0.82,
"learning_rate": 4.369568013711947e-06,
"loss": 1.0618,
"step": 1637
},
{
"epoch": 0.82,
"learning_rate": 4.346248595140112e-06,
"loss": 1.0363,
"step": 1638
},
{
"epoch": 0.82,
"learning_rate": 4.322985643135952e-06,
"loss": 1.0105,
"step": 1639
},
{
"epoch": 0.82,
"learning_rate": 4.299779221299499e-06,
"loss": 1.0419,
"step": 1640
},
{
"epoch": 0.82,
"learning_rate": 4.27662939307622e-06,
"loss": 1.0737,
"step": 1641
},
{
"epoch": 0.82,
"learning_rate": 4.253536221756851e-06,
"loss": 1.0044,
"step": 1642
},
{
"epoch": 0.82,
"learning_rate": 4.230499770477258e-06,
"loss": 1.0848,
"step": 1643
},
{
"epoch": 0.82,
"learning_rate": 4.207520102218213e-06,
"loss": 1.062,
"step": 1644
},
{
"epoch": 0.82,
"learning_rate": 4.184597279805241e-06,
"loss": 1.0315,
"step": 1645
},
{
"epoch": 0.82,
"learning_rate": 4.161731365908481e-06,
"loss": 1.0672,
"step": 1646
},
{
"epoch": 0.82,
"learning_rate": 4.138922423042449e-06,
"loss": 1.0524,
"step": 1647
},
{
"epoch": 0.82,
"learning_rate": 4.116170513565942e-06,
"loss": 1.0456,
"step": 1648
},
{
"epoch": 0.82,
"learning_rate": 4.093475699681806e-06,
"loss": 1.0835,
"step": 1649
},
{
"epoch": 0.82,
"learning_rate": 4.070838043436786e-06,
"loss": 1.0587,
"step": 1650
},
{
"epoch": 0.83,
"learning_rate": 4.048257606721381e-06,
"loss": 1.0059,
"step": 1651
},
{
"epoch": 0.83,
"learning_rate": 4.025734451269636e-06,
"loss": 1.0314,
"step": 1652
},
{
"epoch": 0.83,
"learning_rate": 4.003268638659005e-06,
"loss": 1.067,
"step": 1653
},
{
"epoch": 0.83,
"learning_rate": 3.98086023031016e-06,
"loss": 1.01,
"step": 1654
},
{
"epoch": 0.83,
"learning_rate": 3.958509287486823e-06,
"loss": 1.0388,
"step": 1655
},
{
"epoch": 0.83,
"learning_rate": 3.936215871295634e-06,
"loss": 1.0556,
"step": 1656
},
{
"epoch": 0.83,
"learning_rate": 3.913980042685928e-06,
"loss": 0.9993,
"step": 1657
},
{
"epoch": 0.83,
"learning_rate": 3.891801862449629e-06,
"loss": 1.0657,
"step": 1658
},
{
"epoch": 0.83,
"learning_rate": 3.869681391221011e-06,
"loss": 1.0398,
"step": 1659
},
{
"epoch": 0.83,
"learning_rate": 3.847618689476612e-06,
"loss": 1.0507,
"step": 1660
},
{
"epoch": 0.83,
"learning_rate": 3.825613817535021e-06,
"loss": 1.0495,
"step": 1661
},
{
"epoch": 0.83,
"learning_rate": 3.8036668355567045e-06,
"loss": 1.0608,
"step": 1662
},
{
"epoch": 0.83,
"learning_rate": 3.78177780354389e-06,
"loss": 1.0432,
"step": 1663
},
{
"epoch": 0.83,
"learning_rate": 3.7599467813403315e-06,
"loss": 1.0121,
"step": 1664
},
{
"epoch": 0.83,
"learning_rate": 3.738173828631228e-06,
"loss": 1.0254,
"step": 1665
},
{
"epoch": 0.83,
"learning_rate": 3.7164590049429987e-06,
"loss": 1.0461,
"step": 1666
},
{
"epoch": 0.83,
"learning_rate": 3.6948023696431354e-06,
"loss": 1.0759,
"step": 1667
},
{
"epoch": 0.83,
"learning_rate": 3.6732039819400683e-06,
"loss": 1.0132,
"step": 1668
},
{
"epoch": 0.83,
"learning_rate": 3.65166390088294e-06,
"loss": 1.0444,
"step": 1669
},
{
"epoch": 0.83,
"learning_rate": 3.630182185361522e-06,
"loss": 1.0453,
"step": 1670
},
{
"epoch": 0.84,
"learning_rate": 3.6087588941060124e-06,
"loss": 1.046,
"step": 1671
},
{
"epoch": 0.84,
"learning_rate": 3.5873940856868656e-06,
"loss": 1.0085,
"step": 1672
},
{
"epoch": 0.84,
"learning_rate": 3.5660878185146463e-06,
"loss": 1.0241,
"step": 1673
},
{
"epoch": 0.84,
"learning_rate": 3.544840150839876e-06,
"loss": 1.0226,
"step": 1674
},
{
"epoch": 0.84,
"learning_rate": 3.523651140752868e-06,
"loss": 1.0297,
"step": 1675
},
{
"epoch": 0.84,
"learning_rate": 3.502520846183577e-06,
"loss": 1.0479,
"step": 1676
},
{
"epoch": 0.84,
"learning_rate": 3.4814493249014116e-06,
"loss": 1.0554,
"step": 1677
},
{
"epoch": 0.84,
"learning_rate": 3.46043663451511e-06,
"loss": 1.0514,
"step": 1678
},
{
"epoch": 0.84,
"learning_rate": 3.439482832472554e-06,
"loss": 0.9779,
"step": 1679
},
{
"epoch": 0.84,
"learning_rate": 3.418587976060653e-06,
"loss": 1.022,
"step": 1680
},
{
"epoch": 0.84,
"learning_rate": 3.3977521224051427e-06,
"loss": 1.0822,
"step": 1681
},
{
"epoch": 0.84,
"learning_rate": 3.3769753284704526e-06,
"loss": 1.0158,
"step": 1682
},
{
"epoch": 0.84,
"learning_rate": 3.3562576510595385e-06,
"loss": 1.0438,
"step": 1683
},
{
"epoch": 0.84,
"learning_rate": 3.3355991468137394e-06,
"loss": 1.0723,
"step": 1684
},
{
"epoch": 0.84,
"learning_rate": 3.314999872212618e-06,
"loss": 1.0183,
"step": 1685
},
{
"epoch": 0.84,
"learning_rate": 3.2944598835738193e-06,
"loss": 0.9958,
"step": 1686
},
{
"epoch": 0.84,
"learning_rate": 3.2739792370528628e-06,
"loss": 1.0672,
"step": 1687
},
{
"epoch": 0.84,
"learning_rate": 3.2535579886430718e-06,
"loss": 1.0308,
"step": 1688
},
{
"epoch": 0.84,
"learning_rate": 3.2331961941753474e-06,
"loss": 1.0232,
"step": 1689
},
{
"epoch": 0.84,
"learning_rate": 3.2128939093180655e-06,
"loss": 1.0111,
"step": 1690
},
{
"epoch": 0.85,
"learning_rate": 3.192651189576909e-06,
"loss": 1.0072,
"step": 1691
},
{
"epoch": 0.85,
"learning_rate": 3.1724680902946753e-06,
"loss": 1.0377,
"step": 1692
},
{
"epoch": 0.85,
"learning_rate": 3.152344666651208e-06,
"loss": 1.043,
"step": 1693
},
{
"epoch": 0.85,
"learning_rate": 3.1322809736631654e-06,
"loss": 1.0183,
"step": 1694
},
{
"epoch": 0.85,
"learning_rate": 3.11227706618393e-06,
"loss": 1.0676,
"step": 1695
},
{
"epoch": 0.85,
"learning_rate": 3.092332998903416e-06,
"loss": 1.0629,
"step": 1696
},
{
"epoch": 0.85,
"learning_rate": 3.072448826347932e-06,
"loss": 1.0335,
"step": 1697
},
{
"epoch": 0.85,
"learning_rate": 3.0526246028800637e-06,
"loss": 1.0138,
"step": 1698
},
{
"epoch": 0.85,
"learning_rate": 3.0328603826984658e-06,
"loss": 1.1068,
"step": 1699
},
{
"epoch": 0.85,
"learning_rate": 3.013156219837776e-06,
"loss": 1.0675,
"step": 1700
},
{
"epoch": 0.85,
"eval_acceptance_rate": 0.805456280708313,
"eval_acceptance_rate_1": 0.8047510981559753,
"eval_acceptance_rate_2": 0.8056121468544006,
"eval_acceptance_rate_3": 0.8047822713851929,
"eval_acceptance_rate_4": 0.8054250478744507,
"eval_avg_generation_length": 215.4989013671875,
"eval_avg_input_length": 365.6648864746094,
"eval_expected_speedup": 2.71812246148479,
"eval_expected_tokens": 3.3976530768559874,
"eval_loss": 0.9364066123962402,
"eval_max_generation_length": 424.7227783203125,
"eval_max_input_length": 630.4989624023438,
"eval_min_generation_length": 59.60786437988281,
"eval_min_input_length": 160.72378540039062,
"eval_pulsar_acceptance_rate": 0.5994132692139968,
"eval_runtime": 120.9013,
"eval_samples_per_second": 4.094,
"eval_steps_per_second": 0.132,
"step": 1700
},
{
"epoch": 0.85,
"learning_rate": 2.9935121681684138e-06,
"loss": 1.0321,
"step": 1701
},
{
"epoch": 0.85,
"learning_rate": 2.9739282813964627e-06,
"loss": 0.995,
"step": 1702
},
{
"epoch": 0.85,
"learning_rate": 2.954404613063527e-06,
"loss": 1.0246,
"step": 1703
},
{
"epoch": 0.85,
"learning_rate": 2.9349412165465773e-06,
"loss": 1.0467,
"step": 1704
},
{
"epoch": 0.85,
"learning_rate": 2.9155381450577863e-06,
"loss": 1.058,
"step": 1705
},
{
"epoch": 0.85,
"learning_rate": 2.896195451644415e-06,
"loss": 1.0161,
"step": 1706
},
{
"epoch": 0.85,
"learning_rate": 2.876913189188643e-06,
"loss": 1.0368,
"step": 1707
},
{
"epoch": 0.85,
"learning_rate": 2.8576914104074425e-06,
"loss": 1.0646,
"step": 1708
},
{
"epoch": 0.85,
"learning_rate": 2.838530167852435e-06,
"loss": 1.0333,
"step": 1709
},
{
"epoch": 0.85,
"learning_rate": 2.8194295139097048e-06,
"loss": 1.0371,
"step": 1710
},
{
"epoch": 0.86,
"learning_rate": 2.8003895007997274e-06,
"loss": 1.0161,
"step": 1711
},
{
"epoch": 0.86,
"learning_rate": 2.781410180577157e-06,
"loss": 1.0142,
"step": 1712
},
{
"epoch": 0.86,
"learning_rate": 2.7624916051307405e-06,
"loss": 0.9961,
"step": 1713
},
{
"epoch": 0.86,
"learning_rate": 2.743633826183145e-06,
"loss": 1.0144,
"step": 1714
},
{
"epoch": 0.86,
"learning_rate": 2.7248368952908053e-06,
"loss": 1.0487,
"step": 1715
},
{
"epoch": 0.86,
"learning_rate": 2.706100863843822e-06,
"loss": 1.0585,
"step": 1716
},
{
"epoch": 0.86,
"learning_rate": 2.6874257830657805e-06,
"loss": 1.039,
"step": 1717
},
{
"epoch": 0.86,
"learning_rate": 2.6688117040136464e-06,
"loss": 1.0201,
"step": 1718
},
{
"epoch": 0.86,
"learning_rate": 2.6502586775776076e-06,
"loss": 1.034,
"step": 1719
},
{
"epoch": 0.86,
"learning_rate": 2.6317667544809134e-06,
"loss": 1.0182,
"step": 1720
},
{
"epoch": 0.86,
"learning_rate": 2.6133359852797886e-06,
"loss": 1.0833,
"step": 1721
},
{
"epoch": 0.86,
"learning_rate": 2.5949664203632428e-06,
"loss": 1.0108,
"step": 1722
},
{
"epoch": 0.86,
"learning_rate": 2.576658109952973e-06,
"loss": 1.0262,
"step": 1723
},
{
"epoch": 0.86,
"learning_rate": 2.558411104103198e-06,
"loss": 1.057,
"step": 1724
},
{
"epoch": 0.86,
"learning_rate": 2.5402254527005287e-06,
"loss": 1.0137,
"step": 1725
},
{
"epoch": 0.86,
"learning_rate": 2.522101205463853e-06,
"loss": 1.0776,
"step": 1726
},
{
"epoch": 0.86,
"learning_rate": 2.5040384119441594e-06,
"loss": 1.0798,
"step": 1727
},
{
"epoch": 0.86,
"learning_rate": 2.4860371215244484e-06,
"loss": 1.0189,
"step": 1728
},
{
"epoch": 0.86,
"learning_rate": 2.4680973834195516e-06,
"loss": 1.0754,
"step": 1729
},
{
"epoch": 0.86,
"learning_rate": 2.4502192466760276e-06,
"loss": 0.9813,
"step": 1730
},
{
"epoch": 0.87,
"learning_rate": 2.4324027601720257e-06,
"loss": 1.0602,
"step": 1731
},
{
"epoch": 0.87,
"learning_rate": 2.414647972617129e-06,
"loss": 1.0331,
"step": 1732
},
{
"epoch": 0.87,
"learning_rate": 2.3969549325522618e-06,
"loss": 1.0336,
"step": 1733
},
{
"epoch": 0.87,
"learning_rate": 2.379323688349516e-06,
"loss": 1.0352,
"step": 1734
},
{
"epoch": 0.87,
"learning_rate": 2.361754288212031e-06,
"loss": 1.0061,
"step": 1735
},
{
"epoch": 0.87,
"learning_rate": 2.3442467801738863e-06,
"loss": 1.0297,
"step": 1736
},
{
"epoch": 0.87,
"learning_rate": 2.326801212099938e-06,
"loss": 1.0246,
"step": 1737
},
{
"epoch": 0.87,
"learning_rate": 2.309417631685698e-06,
"loss": 1.0362,
"step": 1738
},
{
"epoch": 0.87,
"learning_rate": 2.2920960864572212e-06,
"loss": 1.0604,
"step": 1739
},
{
"epoch": 0.87,
"learning_rate": 2.2748366237709374e-06,
"loss": 1.0378,
"step": 1740
},
{
"epoch": 0.87,
"learning_rate": 2.2576392908135707e-06,
"loss": 1.0502,
"step": 1741
},
{
"epoch": 0.87,
"learning_rate": 2.24050413460197e-06,
"loss": 1.0255,
"step": 1742
},
{
"epoch": 0.87,
"learning_rate": 2.223431201982992e-06,
"loss": 1.0185,
"step": 1743
},
{
"epoch": 0.87,
"learning_rate": 2.2064205396333886e-06,
"loss": 1.0011,
"step": 1744
},
{
"epoch": 0.87,
"learning_rate": 2.1894721940596554e-06,
"loss": 1.0398,
"step": 1745
},
{
"epoch": 0.87,
"learning_rate": 2.1725862115979322e-06,
"loss": 1.0836,
"step": 1746
},
{
"epoch": 0.87,
"learning_rate": 2.155762638413841e-06,
"loss": 1.0202,
"step": 1747
},
{
"epoch": 0.87,
"learning_rate": 2.13900152050239e-06,
"loss": 1.0487,
"step": 1748
},
{
"epoch": 0.87,
"learning_rate": 2.1223029036878395e-06,
"loss": 1.0326,
"step": 1749
},
{
"epoch": 0.88,
"learning_rate": 2.1056668336235622e-06,
"loss": 1.0501,
"step": 1750
},
{
"epoch": 0.88,
"learning_rate": 2.089093355791952e-06,
"loss": 1.0302,
"step": 1751
},
{
"epoch": 0.88,
"learning_rate": 2.072582515504254e-06,
"loss": 1.0516,
"step": 1752
},
{
"epoch": 0.88,
"learning_rate": 2.0561343579004715e-06,
"loss": 0.9906,
"step": 1753
},
{
"epoch": 0.88,
"learning_rate": 2.0397489279492506e-06,
"loss": 1.1006,
"step": 1754
},
{
"epoch": 0.88,
"learning_rate": 2.023426270447723e-06,
"loss": 1.0594,
"step": 1755
},
{
"epoch": 0.88,
"learning_rate": 2.007166430021415e-06,
"loss": 1.0731,
"step": 1756
},
{
"epoch": 0.88,
"learning_rate": 1.990969451124111e-06,
"loss": 1.0098,
"step": 1757
},
{
"epoch": 0.88,
"learning_rate": 1.9748353780377234e-06,
"loss": 1.0514,
"step": 1758
},
{
"epoch": 0.88,
"learning_rate": 1.958764254872206e-06,
"loss": 1.0095,
"step": 1759
},
{
"epoch": 0.88,
"learning_rate": 1.9427561255653816e-06,
"loss": 1.0192,
"step": 1760
},
{
"epoch": 0.88,
"learning_rate": 1.9268110338828747e-06,
"loss": 1.0065,
"step": 1761
},
{
"epoch": 0.88,
"learning_rate": 1.9109290234179567e-06,
"loss": 1.0526,
"step": 1762
},
{
"epoch": 0.88,
"learning_rate": 1.895110137591427e-06,
"loss": 1.046,
"step": 1763
},
{
"epoch": 0.88,
"learning_rate": 1.87935441965153e-06,
"loss": 1.012,
"step": 1764
},
{
"epoch": 0.88,
"learning_rate": 1.8636619126737892e-06,
"loss": 1.0432,
"step": 1765
},
{
"epoch": 0.88,
"learning_rate": 1.8480326595609166e-06,
"loss": 1.0238,
"step": 1766
},
{
"epoch": 0.88,
"learning_rate": 1.832466703042701e-06,
"loss": 1.054,
"step": 1767
},
{
"epoch": 0.88,
"learning_rate": 1.8169640856758651e-06,
"loss": 1.0777,
"step": 1768
},
{
"epoch": 0.88,
"learning_rate": 1.80152484984398e-06,
"loss": 1.041,
"step": 1769
},
{
"epoch": 0.89,
"learning_rate": 1.7861490377573258e-06,
"loss": 1.0392,
"step": 1770
},
{
"epoch": 0.89,
"learning_rate": 1.770836691452779e-06,
"loss": 0.9898,
"step": 1771
},
{
"epoch": 0.89,
"learning_rate": 1.7555878527937164e-06,
"loss": 1.0174,
"step": 1772
},
{
"epoch": 0.89,
"learning_rate": 1.7404025634698756e-06,
"loss": 1.0932,
"step": 1773
},
{
"epoch": 0.89,
"learning_rate": 1.7252808649972668e-06,
"loss": 1.0655,
"step": 1774
},
{
"epoch": 0.89,
"learning_rate": 1.710222798718028e-06,
"loss": 1.0313,
"step": 1775
},
{
"epoch": 0.89,
"learning_rate": 1.6952284058003366e-06,
"loss": 1.0293,
"step": 1776
},
{
"epoch": 0.89,
"learning_rate": 1.6802977272382974e-06,
"loss": 1.0253,
"step": 1777
},
{
"epoch": 0.89,
"learning_rate": 1.6654308038518058e-06,
"loss": 1.0673,
"step": 1778
},
{
"epoch": 0.89,
"learning_rate": 1.650627676286473e-06,
"loss": 1.0366,
"step": 1779
},
{
"epoch": 0.89,
"learning_rate": 1.6358883850134816e-06,
"loss": 1.0027,
"step": 1780
},
{
"epoch": 0.89,
"learning_rate": 1.6212129703294832e-06,
"loss": 1.0071,
"step": 1781
},
{
"epoch": 0.89,
"learning_rate": 1.6066014723565116e-06,
"loss": 1.0011,
"step": 1782
},
{
"epoch": 0.89,
"learning_rate": 1.5920539310418341e-06,
"loss": 1.0222,
"step": 1783
},
{
"epoch": 0.89,
"learning_rate": 1.5775703861578866e-06,
"loss": 1.0125,
"step": 1784
},
{
"epoch": 0.89,
"learning_rate": 1.5631508773021165e-06,
"loss": 1.0607,
"step": 1785
},
{
"epoch": 0.89,
"learning_rate": 1.5487954438969139e-06,
"loss": 1.0145,
"step": 1786
},
{
"epoch": 0.89,
"learning_rate": 1.5345041251894871e-06,
"loss": 1.0147,
"step": 1787
},
{
"epoch": 0.89,
"learning_rate": 1.5202769602517515e-06,
"loss": 1.0362,
"step": 1788
},
{
"epoch": 0.89,
"learning_rate": 1.5061139879802372e-06,
"loss": 1.0256,
"step": 1789
},
{
"epoch": 0.9,
"learning_rate": 1.4920152470959707e-06,
"loss": 1.0002,
"step": 1790
},
{
"epoch": 0.9,
"learning_rate": 1.4779807761443636e-06,
"loss": 1.068,
"step": 1791
},
{
"epoch": 0.9,
"learning_rate": 1.4640106134951316e-06,
"loss": 1.0429,
"step": 1792
},
{
"epoch": 0.9,
"learning_rate": 1.4501047973421639e-06,
"loss": 1.0779,
"step": 1793
},
{
"epoch": 0.9,
"learning_rate": 1.4362633657034248e-06,
"loss": 1.0725,
"step": 1794
},
{
"epoch": 0.9,
"learning_rate": 1.4224863564208684e-06,
"loss": 1.086,
"step": 1795
},
{
"epoch": 0.9,
"learning_rate": 1.4087738071603075e-06,
"loss": 1.0511,
"step": 1796
},
{
"epoch": 0.9,
"learning_rate": 1.3951257554113306e-06,
"loss": 1.0371,
"step": 1797
},
{
"epoch": 0.9,
"learning_rate": 1.3815422384871879e-06,
"loss": 1.0318,
"step": 1798
},
{
"epoch": 0.9,
"learning_rate": 1.368023293524695e-06,
"loss": 1.0576,
"step": 1799
},
{
"epoch": 0.9,
"learning_rate": 1.3545689574841342e-06,
"loss": 1.0213,
"step": 1800
},
{
"epoch": 0.9,
"eval_acceptance_rate": 0.8056168556213379,
"eval_acceptance_rate_1": 0.8049073219299316,
"eval_acceptance_rate_2": 0.8058131337165833,
"eval_acceptance_rate_3": 0.8049455285072327,
"eval_acceptance_rate_4": 0.8055549263954163,
"eval_avg_generation_length": 215.4989013671875,
"eval_avg_input_length": 365.6648864746094,
"eval_expected_speedup": 2.7189765345835317,
"eval_expected_tokens": 3.3987206682294144,
"eval_loss": 0.9362132549285889,
"eval_max_generation_length": 424.7227783203125,
"eval_max_input_length": 630.4989624023438,
"eval_min_generation_length": 59.60786437988281,
"eval_min_input_length": 160.72378540039062,
"eval_pulsar_acceptance_rate": 0.5996801670573536,
"eval_runtime": 117.7186,
"eval_samples_per_second": 4.205,
"eval_steps_per_second": 0.136,
"step": 1800
},
{
"epoch": 0.9,
"learning_rate": 1.3411792671491424e-06,
"loss": 1.0487,
"step": 1801
},
{
"epoch": 0.9,
"learning_rate": 1.3278542591266291e-06,
"loss": 1.0164,
"step": 1802
},
{
"epoch": 0.9,
"learning_rate": 1.3145939698466502e-06,
"loss": 1.0293,
"step": 1803
},
{
"epoch": 0.9,
"learning_rate": 1.3013984355623315e-06,
"loss": 1.1083,
"step": 1804
},
{
"epoch": 0.9,
"learning_rate": 1.288267692349765e-06,
"loss": 1.0126,
"step": 1805
},
{
"epoch": 0.9,
"learning_rate": 1.2752017761078927e-06,
"loss": 1.0457,
"step": 1806
},
{
"epoch": 0.9,
"learning_rate": 1.2622007225584431e-06,
"loss": 1.0452,
"step": 1807
},
{
"epoch": 0.9,
"learning_rate": 1.2492645672457837e-06,
"loss": 1.0583,
"step": 1808
},
{
"epoch": 0.9,
"learning_rate": 1.2363933455368792e-06,
"loss": 1.073,
"step": 1809
},
{
"epoch": 0.91,
"learning_rate": 1.2235870926211619e-06,
"loss": 1.0061,
"step": 1810
},
{
"epoch": 0.91,
"learning_rate": 1.2108458435104274e-06,
"loss": 1.04,
"step": 1811
},
{
"epoch": 0.91,
"learning_rate": 1.1981696330387787e-06,
"loss": 1.0073,
"step": 1812
},
{
"epoch": 0.91,
"learning_rate": 1.1855584958624765e-06,
"loss": 1.0498,
"step": 1813
},
{
"epoch": 0.91,
"learning_rate": 1.1730124664598912e-06,
"loss": 1.0291,
"step": 1814
},
{
"epoch": 0.91,
"learning_rate": 1.1605315791313964e-06,
"loss": 1.0121,
"step": 1815
},
{
"epoch": 0.91,
"learning_rate": 1.1481158679992555e-06,
"loss": 1.0092,
"step": 1816
},
{
"epoch": 0.91,
"learning_rate": 1.135765367007552e-06,
"loss": 1.0793,
"step": 1817
},
{
"epoch": 0.91,
"learning_rate": 1.1234801099220787e-06,
"loss": 1.014,
"step": 1818
},
{
"epoch": 0.91,
"learning_rate": 1.1112601303302605e-06,
"loss": 1.0659,
"step": 1819
},
{
"epoch": 0.91,
"learning_rate": 1.0991054616410589e-06,
"loss": 1.0831,
"step": 1820
},
{
"epoch": 0.91,
"learning_rate": 1.087016137084873e-06,
"loss": 0.9983,
"step": 1821
},
{
"epoch": 0.91,
"learning_rate": 1.074992189713453e-06,
"loss": 1.0953,
"step": 1822
},
{
"epoch": 0.91,
"learning_rate": 1.0630336523998113e-06,
"loss": 0.975,
"step": 1823
},
{
"epoch": 0.91,
"learning_rate": 1.051140557838129e-06,
"loss": 1.0235,
"step": 1824
},
{
"epoch": 0.91,
"learning_rate": 1.0393129385436824e-06,
"loss": 1.0468,
"step": 1825
},
{
"epoch": 0.91,
"learning_rate": 1.0275508268527246e-06,
"loss": 1.0269,
"step": 1826
},
{
"epoch": 0.91,
"learning_rate": 1.0158542549224186e-06,
"loss": 1.0665,
"step": 1827
},
{
"epoch": 0.91,
"learning_rate": 1.004223254730749e-06,
"loss": 1.0247,
"step": 1828
},
{
"epoch": 0.91,
"learning_rate": 9.926578580764234e-07,
"loss": 1.0367,
"step": 1829
},
{
"epoch": 0.92,
"learning_rate": 9.811580965787965e-07,
"loss": 1.0595,
"step": 1830
},
{
"epoch": 0.92,
"learning_rate": 9.69724001677777e-07,
"loss": 1.0648,
"step": 1831
},
{
"epoch": 0.92,
"learning_rate": 9.5835560463374e-07,
"loss": 1.0148,
"step": 1832
},
{
"epoch": 0.92,
"learning_rate": 9.470529365274566e-07,
"loss": 1.0162,
"step": 1833
},
{
"epoch": 0.92,
"learning_rate": 9.358160282599809e-07,
"loss": 1.0398,
"step": 1834
},
{
"epoch": 0.92,
"learning_rate": 9.246449105525995e-07,
"loss": 1.0733,
"step": 1835
},
{
"epoch": 0.92,
"learning_rate": 9.135396139467151e-07,
"loss": 1.0501,
"step": 1836
},
{
"epoch": 0.92,
"learning_rate": 9.025001688037854e-07,
"loss": 1.043,
"step": 1837
},
{
"epoch": 0.92,
"learning_rate": 8.915266053052373e-07,
"loss": 1.0533,
"step": 1838
},
{
"epoch": 0.92,
"learning_rate": 8.806189534523668e-07,
"loss": 1.019,
"step": 1839
},
{
"epoch": 0.92,
"learning_rate": 8.697772430662859e-07,
"loss": 1.0182,
"step": 1840
},
{
"epoch": 0.92,
"learning_rate": 8.590015037878068e-07,
"loss": 0.9945,
"step": 1841
},
{
"epoch": 0.92,
"learning_rate": 8.48291765077397e-07,
"loss": 1.0192,
"step": 1842
},
{
"epoch": 0.92,
"learning_rate": 8.376480562150707e-07,
"loss": 1.0746,
"step": 1843
},
{
"epoch": 0.92,
"learning_rate": 8.270704063003232e-07,
"loss": 1.0299,
"step": 1844
},
{
"epoch": 0.92,
"learning_rate": 8.165588442520439e-07,
"loss": 1.0445,
"step": 1845
},
{
"epoch": 0.92,
"learning_rate": 8.061133988084391e-07,
"loss": 1.0117,
"step": 1846
},
{
"epoch": 0.92,
"learning_rate": 7.957340985269596e-07,
"loss": 1.0694,
"step": 1847
},
{
"epoch": 0.92,
"learning_rate": 7.854209717842231e-07,
"loss": 1.054,
"step": 1848
},
{
"epoch": 0.92,
"learning_rate": 7.751740467759145e-07,
"loss": 1.0164,
"step": 1849
},
{
"epoch": 0.93,
"learning_rate": 7.649933515167407e-07,
"loss": 1.043,
"step": 1850
},
{
"epoch": 0.93,
"learning_rate": 7.548789138403317e-07,
"loss": 1.0131,
"step": 1851
},
{
"epoch": 0.93,
"learning_rate": 7.448307613991734e-07,
"loss": 1.0369,
"step": 1852
},
{
"epoch": 0.93,
"learning_rate": 7.348489216645327e-07,
"loss": 1.0765,
"step": 1853
},
{
"epoch": 0.93,
"learning_rate": 7.249334219263799e-07,
"loss": 1.0446,
"step": 1854
},
{
"epoch": 0.93,
"learning_rate": 7.150842892933107e-07,
"loss": 1.0724,
"step": 1855
},
{
"epoch": 0.93,
"learning_rate": 7.053015506924748e-07,
"loss": 1.0252,
"step": 1856
},
{
"epoch": 0.93,
"learning_rate": 6.955852328695056e-07,
"loss": 1.0499,
"step": 1857
},
{
"epoch": 0.93,
"learning_rate": 6.859353623884567e-07,
"loss": 0.9969,
"step": 1858
},
{
"epoch": 0.93,
"learning_rate": 6.763519656316914e-07,
"loss": 1.0718,
"step": 1859
},
{
"epoch": 0.93,
"learning_rate": 6.668350687998565e-07,
"loss": 1.0356,
"step": 1860
},
{
"epoch": 0.93,
"learning_rate": 6.573846979117809e-07,
"loss": 1.0071,
"step": 1861
},
{
"epoch": 0.93,
"learning_rate": 6.480008788044223e-07,
"loss": 1.0012,
"step": 1862
},
{
"epoch": 0.93,
"learning_rate": 6.386836371327892e-07,
"loss": 1.0654,
"step": 1863
},
{
"epoch": 0.93,
"learning_rate": 6.2943299836985e-07,
"loss": 1.0438,
"step": 1864
},
{
"epoch": 0.93,
"learning_rate": 6.202489878065071e-07,
"loss": 1.0114,
"step": 1865
},
{
"epoch": 0.93,
"learning_rate": 6.111316305514925e-07,
"loss": 1.0471,
"step": 1866
},
{
"epoch": 0.93,
"learning_rate": 6.020809515313142e-07,
"loss": 1.0422,
"step": 1867
},
{
"epoch": 0.93,
"learning_rate": 5.930969754901843e-07,
"loss": 1.0628,
"step": 1868
},
{
"epoch": 0.93,
"learning_rate": 5.841797269899447e-07,
"loss": 1.0523,
"step": 1869
},
{
"epoch": 0.94,
"learning_rate": 5.753292304100183e-07,
"loss": 1.0533,
"step": 1870
},
{
"epoch": 0.94,
"learning_rate": 5.665455099473221e-07,
"loss": 1.0418,
"step": 1871
},
{
"epoch": 0.94,
"learning_rate": 5.578285896162106e-07,
"loss": 1.0207,
"step": 1872
},
{
"epoch": 0.94,
"learning_rate": 5.491784932484173e-07,
"loss": 1.006,
"step": 1873
},
{
"epoch": 0.94,
"learning_rate": 5.405952444929668e-07,
"loss": 1.0387,
"step": 1874
},
{
"epoch": 0.94,
"learning_rate": 5.32078866816138e-07,
"loss": 1.0243,
"step": 1875
},
{
"epoch": 0.94,
"learning_rate": 5.236293835013839e-07,
"loss": 1.0285,
"step": 1876
},
{
"epoch": 0.94,
"learning_rate": 5.152468176492675e-07,
"loss": 1.0484,
"step": 1877
},
{
"epoch": 0.94,
"learning_rate": 5.069311921774039e-07,
"loss": 1.0017,
"step": 1878
},
{
"epoch": 0.94,
"learning_rate": 4.986825298203935e-07,
"loss": 1.0356,
"step": 1879
},
{
"epoch": 0.94,
"learning_rate": 4.905008531297661e-07,
"loss": 1.0651,
"step": 1880
},
{
"epoch": 0.94,
"learning_rate": 4.823861844739152e-07,
"loss": 1.0348,
"step": 1881
},
{
"epoch": 0.94,
"learning_rate": 4.743385460380334e-07,
"loss": 1.061,
"step": 1882
},
{
"epoch": 0.94,
"learning_rate": 4.663579598240569e-07,
"loss": 1.0435,
"step": 1883
},
{
"epoch": 0.94,
"learning_rate": 4.5844444765059945e-07,
"loss": 1.0007,
"step": 1884
},
{
"epoch": 0.94,
"learning_rate": 4.505980311529101e-07,
"loss": 1.0314,
"step": 1885
},
{
"epoch": 0.94,
"learning_rate": 4.4281873178278475e-07,
"loss": 1.0571,
"step": 1886
},
{
"epoch": 0.94,
"learning_rate": 4.351065708085328e-07,
"loss": 1.0501,
"step": 1887
},
{
"epoch": 0.94,
"learning_rate": 4.2746156931490754e-07,
"loss": 1.0112,
"step": 1888
},
{
"epoch": 0.94,
"learning_rate": 4.1988374820305377e-07,
"loss": 1.0404,
"step": 1889
},
{
"epoch": 0.94,
"learning_rate": 4.1237312819044085e-07,
"loss": 1.013,
"step": 1890
},
{
"epoch": 0.95,
"learning_rate": 4.049297298108212e-07,
"loss": 1.0486,
"step": 1891
},
{
"epoch": 0.95,
"learning_rate": 3.9755357341415835e-07,
"loss": 0.9871,
"step": 1892
},
{
"epoch": 0.95,
"learning_rate": 3.902446791665848e-07,
"loss": 1.0775,
"step": 1893
},
{
"epoch": 0.95,
"learning_rate": 3.83003067050336e-07,
"loss": 1.0516,
"step": 1894
},
{
"epoch": 0.95,
"learning_rate": 3.758287568637081e-07,
"loss": 1.0603,
"step": 1895
},
{
"epoch": 0.95,
"learning_rate": 3.687217682209837e-07,
"loss": 1.0598,
"step": 1896
},
{
"epoch": 0.95,
"learning_rate": 3.616821205524007e-07,
"loss": 1.0456,
"step": 1897
},
{
"epoch": 0.95,
"learning_rate": 3.5470983310409157e-07,
"loss": 1.0245,
"step": 1898
},
{
"epoch": 0.95,
"learning_rate": 3.478049249380194e-07,
"loss": 1.0201,
"step": 1899
},
{
"epoch": 0.95,
"learning_rate": 3.4096741493194197e-07,
"loss": 1.0096,
"step": 1900
},
{
"epoch": 0.95,
"eval_acceptance_rate": 0.8056114912033081,
"eval_acceptance_rate_1": 0.8048818707466125,
"eval_acceptance_rate_2": 0.8058326840400696,
"eval_acceptance_rate_3": 0.8049170970916748,
"eval_acceptance_rate_4": 0.8055611848831177,
"eval_avg_generation_length": 215.4989013671875,
"eval_avg_input_length": 365.6648864746094,
"eval_expected_speedup": 2.7189479972053645,
"eval_expected_tokens": 3.398684996506706,
"eval_loss": 0.9361222386360168,
"eval_max_generation_length": 424.7227783203125,
"eval_max_input_length": 630.4989624023438,
"eval_min_generation_length": 59.60786437988281,
"eval_min_input_length": 160.72378540039062,
"eval_pulsar_acceptance_rate": 0.5996712491266765,
"eval_runtime": 121.0063,
"eval_samples_per_second": 4.091,
"eval_steps_per_second": 0.132,
"step": 1900
},
{
"epoch": 0.95,
"learning_rate": 3.341973217793476e-07,
"loss": 1.0343,
"step": 1901
},
{
"epoch": 0.95,
"learning_rate": 3.2749466398941384e-07,
"loss": 1.0196,
"step": 1902
},
{
"epoch": 0.95,
"learning_rate": 3.2085945988695464e-07,
"loss": 1.0649,
"step": 1903
},
{
"epoch": 0.95,
"learning_rate": 3.142917276123564e-07,
"loss": 1.0531,
"step": 1904
},
{
"epoch": 0.95,
"learning_rate": 3.077914851215585e-07,
"loss": 1.0683,
"step": 1905
},
{
"epoch": 0.95,
"learning_rate": 3.013587501859677e-07,
"loss": 1.0391,
"step": 1906
},
{
"epoch": 0.95,
"learning_rate": 2.949935403924381e-07,
"loss": 1.0373,
"step": 1907
},
{
"epoch": 0.95,
"learning_rate": 2.886958731432132e-07,
"loss": 0.9893,
"step": 1908
},
{
"epoch": 0.95,
"learning_rate": 2.824657656558705e-07,
"loss": 1.0253,
"step": 1909
},
{
"epoch": 0.95,
"learning_rate": 2.763032349632877e-07,
"loss": 1.0286,
"step": 1910
},
{
"epoch": 0.96,
"learning_rate": 2.7020829791359057e-07,
"loss": 1.0266,
"step": 1911
},
{
"epoch": 0.96,
"learning_rate": 2.641809711700999e-07,
"loss": 1.0269,
"step": 1912
},
{
"epoch": 0.96,
"learning_rate": 2.582212712113008e-07,
"loss": 1.0465,
"step": 1913
},
{
"epoch": 0.96,
"learning_rate": 2.5232921433078483e-07,
"loss": 1.0377,
"step": 1914
},
{
"epoch": 0.96,
"learning_rate": 2.4650481663720525e-07,
"loss": 1.0286,
"step": 1915
},
{
"epoch": 0.96,
"learning_rate": 2.4074809405425225e-07,
"loss": 1.0304,
"step": 1916
},
{
"epoch": 0.96,
"learning_rate": 2.3505906232057505e-07,
"loss": 1.0213,
"step": 1917
},
{
"epoch": 0.96,
"learning_rate": 2.2943773698977932e-07,
"loss": 1.0257,
"step": 1918
},
{
"epoch": 0.96,
"learning_rate": 2.2388413343034652e-07,
"loss": 1.0147,
"step": 1919
},
{
"epoch": 0.96,
"learning_rate": 2.1839826682562015e-07,
"loss": 1.0204,
"step": 1920
},
{
"epoch": 0.96,
"learning_rate": 2.129801521737501e-07,
"loss": 1.081,
"step": 1921
},
{
"epoch": 0.96,
"learning_rate": 2.0762980428765122e-07,
"loss": 1.0299,
"step": 1922
},
{
"epoch": 0.96,
"learning_rate": 2.0234723779497533e-07,
"loss": 1.0389,
"step": 1923
},
{
"epoch": 0.96,
"learning_rate": 1.9713246713805588e-07,
"loss": 1.0487,
"step": 1924
},
{
"epoch": 0.96,
"learning_rate": 1.919855065738746e-07,
"loss": 1.0534,
"step": 1925
},
{
"epoch": 0.96,
"learning_rate": 1.8690637017402534e-07,
"loss": 1.0634,
"step": 1926
},
{
"epoch": 0.96,
"learning_rate": 1.8189507182467535e-07,
"loss": 1.0394,
"step": 1927
},
{
"epoch": 0.96,
"learning_rate": 1.7695162522652353e-07,
"loss": 1.0369,
"step": 1928
},
{
"epoch": 0.96,
"learning_rate": 1.720760438947644e-07,
"loss": 0.9994,
"step": 1929
},
{
"epoch": 0.96,
"learning_rate": 1.6726834115904643e-07,
"loss": 0.9936,
"step": 1930
},
{
"epoch": 0.97,
"learning_rate": 1.6252853016344716e-07,
"loss": 1.0279,
"step": 1931
},
{
"epoch": 0.97,
"learning_rate": 1.578566238664314e-07,
"loss": 1.0518,
"step": 1932
},
{
"epoch": 0.97,
"learning_rate": 1.5325263504081256e-07,
"loss": 0.993,
"step": 1933
},
{
"epoch": 0.97,
"learning_rate": 1.4871657627371916e-07,
"loss": 1.0272,
"step": 1934
},
{
"epoch": 0.97,
"learning_rate": 1.4424845996655888e-07,
"loss": 1.0401,
"step": 1935
},
{
"epoch": 0.97,
"learning_rate": 1.3984829833499636e-07,
"loss": 1.0372,
"step": 1936
},
{
"epoch": 0.97,
"learning_rate": 1.3551610340890307e-07,
"loss": 1.0033,
"step": 1937
},
{
"epoch": 0.97,
"learning_rate": 1.3125188703233816e-07,
"loss": 1.0374,
"step": 1938
},
{
"epoch": 0.97,
"learning_rate": 1.2705566086350097e-07,
"loss": 1.0516,
"step": 1939
},
{
"epoch": 0.97,
"learning_rate": 1.229274363747146e-07,
"loss": 1.0489,
"step": 1940
},
{
"epoch": 0.97,
"learning_rate": 1.1886722485238978e-07,
"loss": 1.0244,
"step": 1941
},
{
"epoch": 0.97,
"learning_rate": 1.1487503739698869e-07,
"loss": 1.0664,
"step": 1942
},
{
"epoch": 0.97,
"learning_rate": 1.109508849230001e-07,
"loss": 1.0236,
"step": 1943
},
{
"epoch": 0.97,
"learning_rate": 1.0709477815890601e-07,
"loss": 1.0221,
"step": 1944
},
{
"epoch": 0.97,
"learning_rate": 1.0330672764715387e-07,
"loss": 1.034,
"step": 1945
},
{
"epoch": 0.97,
"learning_rate": 9.958674374413168e-08,
"loss": 1.0451,
"step": 1946
},
{
"epoch": 0.97,
"learning_rate": 9.59348366201318e-08,
"loss": 1.0319,
"step": 1947
},
{
"epoch": 0.97,
"learning_rate": 9.235101625932885e-08,
"loss": 1.0202,
"step": 1948
},
{
"epoch": 0.97,
"learning_rate": 8.883529245975186e-08,
"loss": 1.0208,
"step": 1949
},
{
"epoch": 0.97,
"learning_rate": 8.538767483325383e-08,
"loss": 1.0155,
"step": 1950
},
{
"epoch": 0.98,
"learning_rate": 8.200817280549222e-08,
"loss": 1.0102,
"step": 1951
},
{
"epoch": 0.98,
"learning_rate": 7.869679561589293e-08,
"loss": 1.0196,
"step": 1952
},
{
"epoch": 0.98,
"learning_rate": 7.545355231763918e-08,
"loss": 1.0505,
"step": 1953
},
{
"epoch": 0.98,
"learning_rate": 7.227845177762982e-08,
"loss": 1.0344,
"step": 1954
},
{
"epoch": 0.98,
"learning_rate": 6.91715026764711e-08,
"loss": 1.012,
"step": 1955
},
{
"epoch": 0.98,
"learning_rate": 6.613271350844608e-08,
"loss": 1.026,
"step": 1956
},
{
"epoch": 0.98,
"learning_rate": 6.316209258148131e-08,
"loss": 1.0403,
"step": 1957
},
{
"epoch": 0.98,
"learning_rate": 6.025964801714412e-08,
"loss": 1.0685,
"step": 1958
},
{
"epoch": 0.98,
"learning_rate": 5.742538775061201e-08,
"loss": 1.0138,
"step": 1959
},
{
"epoch": 0.98,
"learning_rate": 5.4659319530636633e-08,
"loss": 1.0136,
"step": 1960
},
{
"epoch": 0.98,
"learning_rate": 5.196145091954652e-08,
"loss": 1.016,
"step": 1961
},
{
"epoch": 0.98,
"learning_rate": 4.9331789293211026e-08,
"loss": 1.0318,
"step": 1962
},
{
"epoch": 0.98,
"learning_rate": 4.677034184102369e-08,
"loss": 1.0961,
"step": 1963
},
{
"epoch": 0.98,
"learning_rate": 4.427711556588832e-08,
"loss": 1.0609,
"step": 1964
},
{
"epoch": 0.98,
"learning_rate": 4.18521172841857e-08,
"loss": 1.0276,
"step": 1965
},
{
"epoch": 0.98,
"learning_rate": 3.949535362576806e-08,
"loss": 1.0196,
"step": 1966
},
{
"epoch": 0.98,
"learning_rate": 3.72068310339424e-08,
"loss": 1.0558,
"step": 1967
},
{
"epoch": 0.98,
"learning_rate": 3.4986555765434413e-08,
"loss": 0.9643,
"step": 1968
},
{
"epoch": 0.98,
"learning_rate": 3.283453389039959e-08,
"loss": 1.0254,
"step": 1969
},
{
"epoch": 0.98,
"learning_rate": 3.075077129238158e-08,
"loss": 1.0258,
"step": 1970
},
{
"epoch": 0.99,
"learning_rate": 2.8735273668303863e-08,
"loss": 1.0483,
"step": 1971
},
{
"epoch": 0.99,
"learning_rate": 2.6788046528461453e-08,
"loss": 1.0653,
"step": 1972
},
{
"epoch": 0.99,
"learning_rate": 2.490909519650142e-08,
"loss": 1.0227,
"step": 1973
},
{
"epoch": 0.99,
"learning_rate": 2.309842480940627e-08,
"loss": 1.0532,
"step": 1974
},
{
"epoch": 0.99,
"learning_rate": 2.1356040317474512e-08,
"loss": 1.0395,
"step": 1975
},
{
"epoch": 0.99,
"learning_rate": 1.9681946484320644e-08,
"loss": 0.9769,
"step": 1976
},
{
"epoch": 0.99,
"learning_rate": 1.8076147886855744e-08,
"loss": 1.032,
"step": 1977
},
{
"epoch": 0.99,
"learning_rate": 1.6538648915270792e-08,
"loss": 1.034,
"step": 1978
},
{
"epoch": 0.99,
"learning_rate": 1.50694537730256e-08,
"loss": 1.0486,
"step": 1979
},
{
"epoch": 0.99,
"learning_rate": 1.3668566476848777e-08,
"loss": 1.0247,
"step": 1980
},
{
"epoch": 0.99,
"learning_rate": 1.233599085671e-08,
"loss": 1.0239,
"step": 1981
},
{
"epoch": 0.99,
"learning_rate": 1.107173055582278e-08,
"loss": 0.9958,
"step": 1982
},
{
"epoch": 0.99,
"learning_rate": 9.875789030622252e-09,
"loss": 1.0838,
"step": 1983
},
{
"epoch": 0.99,
"learning_rate": 8.74816955076796e-09,
"loss": 1.0383,
"step": 1984
},
{
"epoch": 0.99,
"learning_rate": 7.688875199132751e-09,
"loss": 1.0347,
"step": 1985
},
{
"epoch": 0.99,
"learning_rate": 6.697908871780567e-09,
"loss": 1.026,
"step": 1986
},
{
"epoch": 0.99,
"learning_rate": 5.775273277980331e-09,
"loss": 1.0376,
"step": 1987
},
{
"epoch": 0.99,
"learning_rate": 4.920970940180958e-09,
"loss": 1.0359,
"step": 1988
},
{
"epoch": 0.99,
"learning_rate": 4.1350041940113604e-09,
"loss": 1.0066,
"step": 1989
},
{
"epoch": 0.99,
"learning_rate": 3.417375188274896e-09,
"loss": 1.0479,
"step": 1990
},
{
"epoch": 1.0,
"learning_rate": 2.768085884943816e-09,
"loss": 1.015,
"step": 1991
},
{
"epoch": 1.0,
"learning_rate": 2.1871380591509392e-09,
"loss": 1.0411,
"step": 1992
},
{
"epoch": 1.0,
"learning_rate": 1.6745332991813245e-09,
"loss": 1.0089,
"step": 1993
},
{
"epoch": 1.0,
"learning_rate": 1.2302730064778222e-09,
"loss": 1.0002,
"step": 1994
},
{
"epoch": 1.0,
"learning_rate": 8.543583956355239e-10,
"loss": 0.9952,
"step": 1995
},
{
"epoch": 1.0,
"learning_rate": 5.467904943851077e-10,
"loss": 1.049,
"step": 1996
},
{
"epoch": 1.0,
"learning_rate": 3.0757014360671686e-10,
"loss": 1.0491,
"step": 1997
},
{
"epoch": 1.0,
"learning_rate": 1.3669799732163313e-10,
"loss": 1.0033,
"step": 1998
},
{
"epoch": 1.0,
"learning_rate": 3.417452268950072e-11,
"loss": 1.0398,
"step": 1999
},
{
"epoch": 1.0,
"learning_rate": 0.0,
"loss": 1.0098,
"step": 2000
},
{
"epoch": 1.0,
"eval_acceptance_rate": 0.8056314587593079,
"eval_acceptance_rate_1": 0.8049386739730835,
"eval_acceptance_rate_2": 0.805794894695282,
"eval_acceptance_rate_3": 0.8049510717391968,
"eval_acceptance_rate_4": 0.8055833578109741,
"eval_avg_generation_length": 215.4989013671875,
"eval_avg_input_length": 365.6648864746094,
"eval_expected_speedup": 2.719054221373988,
"eval_expected_tokens": 3.3988177767174848,
"eval_loss": 0.9361461400985718,
"eval_max_generation_length": 424.7227783203125,
"eval_max_input_length": 630.4989624023438,
"eval_min_generation_length": 59.60786437988281,
"eval_min_input_length": 160.72378540039062,
"eval_pulsar_acceptance_rate": 0.5997044441793712,
"eval_runtime": 119.8126,
"eval_samples_per_second": 4.131,
"eval_steps_per_second": 0.134,
"step": 2000
}
],
"logging_steps": 1.0,
"max_steps": 2000,
"num_input_tokens_seen": 0,
"num_train_epochs": 9223372036854775807,
"save_steps": 1000,
"total_flos": 1.3440382893293568e+19,
"train_batch_size": 4,
"trial_name": null,
"trial_params": null
}