2456 lines
49 KiB
JSON
2456 lines
49 KiB
JSON
|
|
{
|
||
|
|
"best_metric": null,
|
||
|
|
"best_model_checkpoint": null,
|
||
|
|
"epoch": 1.0,
|
||
|
|
"eval_steps": 75,
|
||
|
|
"global_step": 391,
|
||
|
|
"is_hyper_param_search": false,
|
||
|
|
"is_local_process_zero": true,
|
||
|
|
"is_world_process_zero": true,
|
||
|
|
"log_history": [
|
||
|
|
{
|
||
|
|
"epoch": 0.0,
|
||
|
|
"learning_rate": 7.5e-07,
|
||
|
|
"loss": 1.0384,
|
||
|
|
"step": 1
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.01,
|
||
|
|
"learning_rate": 1.5e-06,
|
||
|
|
"loss": 1.0344,
|
||
|
|
"step": 2
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.01,
|
||
|
|
"learning_rate": 2.25e-06,
|
||
|
|
"loss": 0.8877,
|
||
|
|
"step": 3
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.01,
|
||
|
|
"learning_rate": 3e-06,
|
||
|
|
"loss": 0.9107,
|
||
|
|
"step": 4
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.01,
|
||
|
|
"learning_rate": 3.75e-06,
|
||
|
|
"loss": 0.8865,
|
||
|
|
"step": 5
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.02,
|
||
|
|
"learning_rate": 4.5e-06,
|
||
|
|
"loss": 1.0755,
|
||
|
|
"step": 6
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.02,
|
||
|
|
"learning_rate": 5.2500000000000006e-06,
|
||
|
|
"loss": 0.9334,
|
||
|
|
"step": 7
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.02,
|
||
|
|
"learning_rate": 6e-06,
|
||
|
|
"loss": 0.8215,
|
||
|
|
"step": 8
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.02,
|
||
|
|
"learning_rate": 6.75e-06,
|
||
|
|
"loss": 0.9601,
|
||
|
|
"step": 9
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.03,
|
||
|
|
"learning_rate": 7.5e-06,
|
||
|
|
"loss": 0.7402,
|
||
|
|
"step": 10
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.03,
|
||
|
|
"learning_rate": 8.25e-06,
|
||
|
|
"loss": 0.8968,
|
||
|
|
"step": 11
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.03,
|
||
|
|
"learning_rate": 9e-06,
|
||
|
|
"loss": 0.7669,
|
||
|
|
"step": 12
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.03,
|
||
|
|
"learning_rate": 8.999845402895058e-06,
|
||
|
|
"loss": 0.7714,
|
||
|
|
"step": 13
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.04,
|
||
|
|
"learning_rate": 8.999381622202572e-06,
|
||
|
|
"loss": 0.6609,
|
||
|
|
"step": 14
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.04,
|
||
|
|
"learning_rate": 8.998608689788832e-06,
|
||
|
|
"loss": 0.8098,
|
||
|
|
"step": 15
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.04,
|
||
|
|
"learning_rate": 8.997526658761886e-06,
|
||
|
|
"loss": 0.8355,
|
||
|
|
"step": 16
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.04,
|
||
|
|
"learning_rate": 8.996135603467899e-06,
|
||
|
|
"loss": 0.8318,
|
||
|
|
"step": 17
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.05,
|
||
|
|
"learning_rate": 8.994435619486036e-06,
|
||
|
|
"loss": 0.9067,
|
||
|
|
"step": 18
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.05,
|
||
|
|
"learning_rate": 8.992426823621897e-06,
|
||
|
|
"loss": 0.7113,
|
||
|
|
"step": 19
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.05,
|
||
|
|
"learning_rate": 8.99010935389949e-06,
|
||
|
|
"loss": 0.804,
|
||
|
|
"step": 20
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.05,
|
||
|
|
"learning_rate": 8.987483369551757e-06,
|
||
|
|
"loss": 0.776,
|
||
|
|
"step": 21
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.06,
|
||
|
|
"learning_rate": 8.984549051009623e-06,
|
||
|
|
"loss": 0.8339,
|
||
|
|
"step": 22
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.06,
|
||
|
|
"learning_rate": 8.981306599889595e-06,
|
||
|
|
"loss": 0.833,
|
||
|
|
"step": 23
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.06,
|
||
|
|
"learning_rate": 8.977756238979921e-06,
|
||
|
|
"loss": 0.7614,
|
||
|
|
"step": 24
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.06,
|
||
|
|
"learning_rate": 8.973898212225277e-06,
|
||
|
|
"loss": 0.7553,
|
||
|
|
"step": 25
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.07,
|
||
|
|
"learning_rate": 8.969732784710005e-06,
|
||
|
|
"loss": 0.6437,
|
||
|
|
"step": 26
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.07,
|
||
|
|
"learning_rate": 8.965260242639897e-06,
|
||
|
|
"loss": 0.7507,
|
||
|
|
"step": 27
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.07,
|
||
|
|
"learning_rate": 8.960480893322532e-06,
|
||
|
|
"loss": 0.722,
|
||
|
|
"step": 28
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.07,
|
||
|
|
"learning_rate": 8.955395065146165e-06,
|
||
|
|
"loss": 0.775,
|
||
|
|
"step": 29
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.08,
|
||
|
|
"learning_rate": 8.950003107557156e-06,
|
||
|
|
"loss": 0.6426,
|
||
|
|
"step": 30
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.08,
|
||
|
|
"learning_rate": 8.944305391035962e-06,
|
||
|
|
"loss": 0.5657,
|
||
|
|
"step": 31
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.08,
|
||
|
|
"learning_rate": 8.938302307071689e-06,
|
||
|
|
"loss": 0.7168,
|
||
|
|
"step": 32
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.08,
|
||
|
|
"learning_rate": 8.93199426813518e-06,
|
||
|
|
"loss": 0.6611,
|
||
|
|
"step": 33
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.09,
|
||
|
|
"learning_rate": 8.92538170765068e-06,
|
||
|
|
"loss": 0.5408,
|
||
|
|
"step": 34
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.09,
|
||
|
|
"learning_rate": 8.918465079966063e-06,
|
||
|
|
"loss": 0.7239,
|
||
|
|
"step": 35
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.09,
|
||
|
|
"learning_rate": 8.9112448603216e-06,
|
||
|
|
"loss": 0.7005,
|
||
|
|
"step": 36
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.09,
|
||
|
|
"learning_rate": 8.90372154481732e-06,
|
||
|
|
"loss": 0.5097,
|
||
|
|
"step": 37
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1,
|
||
|
|
"learning_rate": 8.895895650378904e-06,
|
||
|
|
"loss": 0.584,
|
||
|
|
"step": 38
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1,
|
||
|
|
"learning_rate": 8.887767714722187e-06,
|
||
|
|
"loss": 0.7294,
|
||
|
|
"step": 39
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1,
|
||
|
|
"learning_rate": 8.879338296316203e-06,
|
||
|
|
"loss": 0.7048,
|
||
|
|
"step": 40
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1,
|
||
|
|
"learning_rate": 8.870607974344808e-06,
|
||
|
|
"loss": 0.7611,
|
||
|
|
"step": 41
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.11,
|
||
|
|
"learning_rate": 8.861577348666893e-06,
|
||
|
|
"loss": 0.5931,
|
||
|
|
"step": 42
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.11,
|
||
|
|
"learning_rate": 8.852247039775163e-06,
|
||
|
|
"loss": 0.7365,
|
||
|
|
"step": 43
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.11,
|
||
|
|
"learning_rate": 8.842617688753502e-06,
|
||
|
|
"loss": 0.6636,
|
||
|
|
"step": 44
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.12,
|
||
|
|
"learning_rate": 8.832689957232929e-06,
|
||
|
|
"loss": 0.6412,
|
||
|
|
"step": 45
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.12,
|
||
|
|
"learning_rate": 8.822464527346134e-06,
|
||
|
|
"loss": 0.6989,
|
||
|
|
"step": 46
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.12,
|
||
|
|
"learning_rate": 8.81194210168061e-06,
|
||
|
|
"loss": 0.7198,
|
||
|
|
"step": 47
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.12,
|
||
|
|
"learning_rate": 8.801123403230374e-06,
|
||
|
|
"loss": 0.6605,
|
||
|
|
"step": 48
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.13,
|
||
|
|
"learning_rate": 8.7900091753463e-06,
|
||
|
|
"loss": 0.6546,
|
||
|
|
"step": 49
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.13,
|
||
|
|
"learning_rate": 8.778600181685031e-06,
|
||
|
|
"loss": 0.7965,
|
||
|
|
"step": 50
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.13,
|
||
|
|
"learning_rate": 8.766897206156524e-06,
|
||
|
|
"loss": 0.6213,
|
||
|
|
"step": 51
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.13,
|
||
|
|
"learning_rate": 8.754901052870166e-06,
|
||
|
|
"loss": 0.6297,
|
||
|
|
"step": 52
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.14,
|
||
|
|
"learning_rate": 8.742612546079549e-06,
|
||
|
|
"loss": 0.78,
|
||
|
|
"step": 53
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.14,
|
||
|
|
"learning_rate": 8.730032530125813e-06,
|
||
|
|
"loss": 0.6817,
|
||
|
|
"step": 54
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.14,
|
||
|
|
"learning_rate": 8.717161869379647e-06,
|
||
|
|
"loss": 0.6632,
|
||
|
|
"step": 55
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.14,
|
||
|
|
"learning_rate": 8.70400144818189e-06,
|
||
|
|
"loss": 0.5498,
|
||
|
|
"step": 56
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.15,
|
||
|
|
"learning_rate": 8.690552170782772e-06,
|
||
|
|
"loss": 0.6081,
|
||
|
|
"step": 57
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.15,
|
||
|
|
"learning_rate": 8.676814961279782e-06,
|
||
|
|
"loss": 0.4716,
|
||
|
|
"step": 58
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.15,
|
||
|
|
"learning_rate": 8.662790763554175e-06,
|
||
|
|
"loss": 0.5989,
|
||
|
|
"step": 59
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.15,
|
||
|
|
"learning_rate": 8.64848054120611e-06,
|
||
|
|
"loss": 0.535,
|
||
|
|
"step": 60
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.16,
|
||
|
|
"learning_rate": 8.633885277488455e-06,
|
||
|
|
"loss": 0.5284,
|
||
|
|
"step": 61
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.16,
|
||
|
|
"learning_rate": 8.619005975239218e-06,
|
||
|
|
"loss": 0.7458,
|
||
|
|
"step": 62
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.16,
|
||
|
|
"learning_rate": 8.603843656812642e-06,
|
||
|
|
"loss": 0.6758,
|
||
|
|
"step": 63
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.16,
|
||
|
|
"learning_rate": 8.588399364008963e-06,
|
||
|
|
"loss": 0.6736,
|
||
|
|
"step": 64
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.17,
|
||
|
|
"learning_rate": 8.57267415800283e-06,
|
||
|
|
"loss": 0.4975,
|
||
|
|
"step": 65
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.17,
|
||
|
|
"learning_rate": 8.556669119270387e-06,
|
||
|
|
"loss": 0.5586,
|
||
|
|
"step": 66
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.17,
|
||
|
|
"learning_rate": 8.540385347515032e-06,
|
||
|
|
"loss": 0.6974,
|
||
|
|
"step": 67
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.17,
|
||
|
|
"learning_rate": 8.523823961591867e-06,
|
||
|
|
"loss": 0.6778,
|
||
|
|
"step": 68
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.18,
|
||
|
|
"learning_rate": 8.506986099430807e-06,
|
||
|
|
"loss": 0.6791,
|
||
|
|
"step": 69
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.18,
|
||
|
|
"learning_rate": 8.48987291795841e-06,
|
||
|
|
"loss": 0.5207,
|
||
|
|
"step": 70
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.18,
|
||
|
|
"learning_rate": 8.472485593018366e-06,
|
||
|
|
"loss": 0.5651,
|
||
|
|
"step": 71
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.18,
|
||
|
|
"learning_rate": 8.45482531929072e-06,
|
||
|
|
"loss": 0.6487,
|
||
|
|
"step": 72
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.19,
|
||
|
|
"learning_rate": 8.436893310209779e-06,
|
||
|
|
"loss": 0.6202,
|
||
|
|
"step": 73
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.19,
|
||
|
|
"learning_rate": 8.418690797880737e-06,
|
||
|
|
"loss": 0.5263,
|
||
|
|
"step": 74
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.19,
|
||
|
|
"learning_rate": 8.400219032995022e-06,
|
||
|
|
"loss": 0.7257,
|
||
|
|
"step": 75
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.19,
|
||
|
|
"eval_accuracy": 0.814565802103578,
|
||
|
|
"eval_accuracy_<|content|>": 0.9681089145841104,
|
||
|
|
"eval_accuracy_<|from|>": 0.9962073324905183,
|
||
|
|
"eval_accuracy_<|recipient|>": 0.9987357774968394,
|
||
|
|
"eval_accuracy_<|stop|>": 0.8882521489971347,
|
||
|
|
"eval_accuracy_total_num_<|content|>": 5362,
|
||
|
|
"eval_accuracy_total_num_<|from|>": 791,
|
||
|
|
"eval_accuracy_total_num_<|recipient|>": 791,
|
||
|
|
"eval_accuracy_total_num_<|stop|>": 4537,
|
||
|
|
"eval_loss": 0.618374228477478,
|
||
|
|
"eval_perplexity": 1.0682635492781798,
|
||
|
|
"eval_runtime": 290.3169,
|
||
|
|
"eval_samples_per_second": 4.736,
|
||
|
|
"eval_steps_per_second": 0.148,
|
||
|
|
"step": 75
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.19,
|
||
|
|
"learning_rate": 8.381479284744354e-06,
|
||
|
|
"loss": 0.6557,
|
||
|
|
"step": 76
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2,
|
||
|
|
"learning_rate": 8.362472840733548e-06,
|
||
|
|
"loss": 0.6288,
|
||
|
|
"step": 77
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2,
|
||
|
|
"learning_rate": 8.343201006892032e-06,
|
||
|
|
"loss": 0.5379,
|
||
|
|
"step": 78
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2,
|
||
|
|
"learning_rate": 8.323665107384127e-06,
|
||
|
|
"loss": 0.6197,
|
||
|
|
"step": 79
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2,
|
||
|
|
"learning_rate": 8.303866484518059e-06,
|
||
|
|
"loss": 0.5107,
|
||
|
|
"step": 80
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.21,
|
||
|
|
"learning_rate": 8.283806498653725e-06,
|
||
|
|
"loss": 0.67,
|
||
|
|
"step": 81
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.21,
|
||
|
|
"learning_rate": 8.263486528109237e-06,
|
||
|
|
"loss": 0.6076,
|
||
|
|
"step": 82
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.21,
|
||
|
|
"learning_rate": 8.242907969066198e-06,
|
||
|
|
"loss": 0.6506,
|
||
|
|
"step": 83
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.21,
|
||
|
|
"learning_rate": 8.22207223547379e-06,
|
||
|
|
"loss": 0.6114,
|
||
|
|
"step": 84
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.22,
|
||
|
|
"learning_rate": 8.200980758951609e-06,
|
||
|
|
"loss": 0.4775,
|
||
|
|
"step": 85
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.22,
|
||
|
|
"learning_rate": 8.179634988691303e-06,
|
||
|
|
"loss": 0.6625,
|
||
|
|
"step": 86
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.22,
|
||
|
|
"learning_rate": 8.158036391357e-06,
|
||
|
|
"loss": 0.5161,
|
||
|
|
"step": 87
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.23,
|
||
|
|
"learning_rate": 8.136186450984527e-06,
|
||
|
|
"loss": 0.4956,
|
||
|
|
"step": 88
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.23,
|
||
|
|
"learning_rate": 8.114086668879454e-06,
|
||
|
|
"loss": 0.6379,
|
||
|
|
"step": 89
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.23,
|
||
|
|
"learning_rate": 8.09173856351393e-06,
|
||
|
|
"loss": 0.6708,
|
||
|
|
"step": 90
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.23,
|
||
|
|
"learning_rate": 8.069143670422347e-06,
|
||
|
|
"loss": 0.5111,
|
||
|
|
"step": 91
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.24,
|
||
|
|
"learning_rate": 8.046303542095846e-06,
|
||
|
|
"loss": 0.5079,
|
||
|
|
"step": 92
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.24,
|
||
|
|
"learning_rate": 8.02321974787563e-06,
|
||
|
|
"loss": 0.5029,
|
||
|
|
"step": 93
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.24,
|
||
|
|
"learning_rate": 7.999893873845152e-06,
|
||
|
|
"loss": 0.4098,
|
||
|
|
"step": 94
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.24,
|
||
|
|
"learning_rate": 7.976327522721114e-06,
|
||
|
|
"loss": 0.4683,
|
||
|
|
"step": 95
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.25,
|
||
|
|
"learning_rate": 7.952522313743371e-06,
|
||
|
|
"loss": 0.5922,
|
||
|
|
"step": 96
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.25,
|
||
|
|
"learning_rate": 7.928479882563648e-06,
|
||
|
|
"loss": 0.4946,
|
||
|
|
"step": 97
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.25,
|
||
|
|
"learning_rate": 7.904201881133171e-06,
|
||
|
|
"loss": 0.4977,
|
||
|
|
"step": 98
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.25,
|
||
|
|
"learning_rate": 7.879689977589154e-06,
|
||
|
|
"loss": 0.6153,
|
||
|
|
"step": 99
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.26,
|
||
|
|
"learning_rate": 7.85494585614019e-06,
|
||
|
|
"loss": 0.5741,
|
||
|
|
"step": 100
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.26,
|
||
|
|
"learning_rate": 7.829971216950513e-06,
|
||
|
|
"loss": 0.4949,
|
||
|
|
"step": 101
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.26,
|
||
|
|
"learning_rate": 7.804767776023202e-06,
|
||
|
|
"loss": 0.5977,
|
||
|
|
"step": 102
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.26,
|
||
|
|
"learning_rate": 7.779337265082256e-06,
|
||
|
|
"loss": 0.6465,
|
||
|
|
"step": 103
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.27,
|
||
|
|
"learning_rate": 7.753681431453614e-06,
|
||
|
|
"loss": 0.6386,
|
||
|
|
"step": 104
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.27,
|
||
|
|
"learning_rate": 7.727802037945105e-06,
|
||
|
|
"loss": 0.6208,
|
||
|
|
"step": 105
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.27,
|
||
|
|
"learning_rate": 7.70170086272531e-06,
|
||
|
|
"loss": 0.6551,
|
||
|
|
"step": 106
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.27,
|
||
|
|
"learning_rate": 7.675379699201395e-06,
|
||
|
|
"loss": 0.662,
|
||
|
|
"step": 107
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.28,
|
||
|
|
"learning_rate": 7.648840355895885e-06,
|
||
|
|
"loss": 0.5964,
|
||
|
|
"step": 108
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.28,
|
||
|
|
"learning_rate": 7.6220846563224e-06,
|
||
|
|
"loss": 0.6025,
|
||
|
|
"step": 109
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.28,
|
||
|
|
"learning_rate": 7.595114438860358e-06,
|
||
|
|
"loss": 0.5055,
|
||
|
|
"step": 110
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.28,
|
||
|
|
"learning_rate": 7.567931556628665e-06,
|
||
|
|
"loss": 0.5483,
|
||
|
|
"step": 111
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.29,
|
||
|
|
"learning_rate": 7.540537877358389e-06,
|
||
|
|
"loss": 0.6338,
|
||
|
|
"step": 112
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.29,
|
||
|
|
"learning_rate": 7.51293528326442e-06,
|
||
|
|
"loss": 0.6748,
|
||
|
|
"step": 113
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.29,
|
||
|
|
"learning_rate": 7.485125670916155e-06,
|
||
|
|
"loss": 0.5733,
|
||
|
|
"step": 114
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.29,
|
||
|
|
"learning_rate": 7.4571109511071714e-06,
|
||
|
|
"loss": 0.6164,
|
||
|
|
"step": 115
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3,
|
||
|
|
"learning_rate": 7.428893048723952e-06,
|
||
|
|
"loss": 0.6496,
|
||
|
|
"step": 116
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3,
|
||
|
|
"learning_rate": 7.400473902613611e-06,
|
||
|
|
"loss": 0.6056,
|
||
|
|
"step": 117
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3,
|
||
|
|
"learning_rate": 7.371855465450694e-06,
|
||
|
|
"loss": 0.4862,
|
||
|
|
"step": 118
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3,
|
||
|
|
"learning_rate": 7.343039703602988e-06,
|
||
|
|
"loss": 0.5984,
|
||
|
|
"step": 119
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.31,
|
||
|
|
"learning_rate": 7.314028596996431e-06,
|
||
|
|
"loss": 0.5891,
|
||
|
|
"step": 120
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.31,
|
||
|
|
"learning_rate": 7.284824138979066e-06,
|
||
|
|
"loss": 0.5345,
|
||
|
|
"step": 121
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.31,
|
||
|
|
"learning_rate": 7.255428336184075e-06,
|
||
|
|
"loss": 0.7941,
|
||
|
|
"step": 122
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.31,
|
||
|
|
"learning_rate": 7.2258432083919064e-06,
|
||
|
|
"loss": 0.552,
|
||
|
|
"step": 123
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.32,
|
||
|
|
"learning_rate": 7.196070788391497e-06,
|
||
|
|
"loss": 0.599,
|
||
|
|
"step": 124
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.32,
|
||
|
|
"learning_rate": 7.166113121840595e-06,
|
||
|
|
"loss": 0.4727,
|
||
|
|
"step": 125
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.32,
|
||
|
|
"learning_rate": 7.135972267125212e-06,
|
||
|
|
"loss": 0.545,
|
||
|
|
"step": 126
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.32,
|
||
|
|
"learning_rate": 7.1056502952181815e-06,
|
||
|
|
"loss": 0.6106,
|
||
|
|
"step": 127
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.33,
|
||
|
|
"learning_rate": 7.075149289536871e-06,
|
||
|
|
"loss": 0.4726,
|
||
|
|
"step": 128
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.33,
|
||
|
|
"learning_rate": 7.044471345800024e-06,
|
||
|
|
"loss": 0.5416,
|
||
|
|
"step": 129
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.33,
|
||
|
|
"learning_rate": 7.0136185718837685e-06,
|
||
|
|
"loss": 0.6889,
|
||
|
|
"step": 130
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.34,
|
||
|
|
"learning_rate": 6.9825930876767834e-06,
|
||
|
|
"loss": 0.594,
|
||
|
|
"step": 131
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.34,
|
||
|
|
"learning_rate": 6.951397024934641e-06,
|
||
|
|
"loss": 0.5456,
|
||
|
|
"step": 132
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.34,
|
||
|
|
"learning_rate": 6.920032527133334e-06,
|
||
|
|
"loss": 0.4282,
|
||
|
|
"step": 133
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.34,
|
||
|
|
"learning_rate": 6.888501749322002e-06,
|
||
|
|
"loss": 0.6089,
|
||
|
|
"step": 134
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.35,
|
||
|
|
"learning_rate": 6.856806857974848e-06,
|
||
|
|
"loss": 0.6507,
|
||
|
|
"step": 135
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.35,
|
||
|
|
"learning_rate": 6.824950030842293e-06,
|
||
|
|
"loss": 0.6411,
|
||
|
|
"step": 136
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.35,
|
||
|
|
"learning_rate": 6.792933456801339e-06,
|
||
|
|
"loss": 0.5493,
|
||
|
|
"step": 137
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.35,
|
||
|
|
"learning_rate": 6.760759335705163e-06,
|
||
|
|
"loss": 0.4705,
|
||
|
|
"step": 138
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.36,
|
||
|
|
"learning_rate": 6.728429878231978e-06,
|
||
|
|
"loss": 0.4708,
|
||
|
|
"step": 139
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.36,
|
||
|
|
"learning_rate": 6.695947305733131e-06,
|
||
|
|
"loss": 0.5714,
|
||
|
|
"step": 140
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.36,
|
||
|
|
"learning_rate": 6.6633138500804735e-06,
|
||
|
|
"loss": 0.5466,
|
||
|
|
"step": 141
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.36,
|
||
|
|
"learning_rate": 6.630531753513014e-06,
|
||
|
|
"loss": 0.4242,
|
||
|
|
"step": 142
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.37,
|
||
|
|
"learning_rate": 6.597603268482853e-06,
|
||
|
|
"loss": 0.6997,
|
||
|
|
"step": 143
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.37,
|
||
|
|
"learning_rate": 6.5645306575004145e-06,
|
||
|
|
"loss": 0.6938,
|
||
|
|
"step": 144
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.37,
|
||
|
|
"learning_rate": 6.531316192978991e-06,
|
||
|
|
"loss": 0.7704,
|
||
|
|
"step": 145
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.37,
|
||
|
|
"learning_rate": 6.497962157078611e-06,
|
||
|
|
"loss": 0.5127,
|
||
|
|
"step": 146
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.38,
|
||
|
|
"learning_rate": 6.4644708415492205e-06,
|
||
|
|
"loss": 0.6113,
|
||
|
|
"step": 147
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.38,
|
||
|
|
"learning_rate": 6.4308445475732315e-06,
|
||
|
|
"loss": 0.563,
|
||
|
|
"step": 148
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.38,
|
||
|
|
"learning_rate": 6.397085585607401e-06,
|
||
|
|
"loss": 0.6154,
|
||
|
|
"step": 149
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.38,
|
||
|
|
"learning_rate": 6.3631962752240746e-06,
|
||
|
|
"loss": 0.5561,
|
||
|
|
"step": 150
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.38,
|
||
|
|
"eval_accuracy": 0.819768782794211,
|
||
|
|
"eval_accuracy_<|content|>": 1.0,
|
||
|
|
"eval_accuracy_<|from|>": 0.9393173198482933,
|
||
|
|
"eval_accuracy_<|recipient|>": 1.0,
|
||
|
|
"eval_accuracy_<|stop|>": 0.8756887811329072,
|
||
|
|
"eval_accuracy_total_num_<|content|>": 5362,
|
||
|
|
"eval_accuracy_total_num_<|from|>": 791,
|
||
|
|
"eval_accuracy_total_num_<|recipient|>": 791,
|
||
|
|
"eval_accuracy_total_num_<|stop|>": 4537,
|
||
|
|
"eval_loss": 0.586061418056488,
|
||
|
|
"eval_perplexity": 1.0650794931769392,
|
||
|
|
"eval_runtime": 223.6772,
|
||
|
|
"eval_samples_per_second": 6.147,
|
||
|
|
"eval_steps_per_second": 0.192,
|
||
|
|
"step": 150
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.39,
|
||
|
|
"learning_rate": 6.32917894495182e-06,
|
||
|
|
"loss": 0.4405,
|
||
|
|
"step": 151
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.39,
|
||
|
|
"learning_rate": 6.295035932115428e-06,
|
||
|
|
"loss": 0.3631,
|
||
|
|
"step": 152
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.39,
|
||
|
|
"learning_rate": 6.260769582675315e-06,
|
||
|
|
"loss": 0.6282,
|
||
|
|
"step": 153
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.39,
|
||
|
|
"learning_rate": 6.226382251066333e-06,
|
||
|
|
"loss": 0.7472,
|
||
|
|
"step": 154
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4,
|
||
|
|
"learning_rate": 6.191876300036003e-06,
|
||
|
|
"loss": 0.5803,
|
||
|
|
"step": 155
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4,
|
||
|
|
"learning_rate": 6.1572541004821585e-06,
|
||
|
|
"loss": 0.5861,
|
||
|
|
"step": 156
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4,
|
||
|
|
"learning_rate": 6.122518031290052e-06,
|
||
|
|
"loss": 0.6237,
|
||
|
|
"step": 157
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4,
|
||
|
|
"learning_rate": 6.0876704791689e-06,
|
||
|
|
"loss": 0.5705,
|
||
|
|
"step": 158
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.41,
|
||
|
|
"learning_rate": 6.0527138384878885e-06,
|
||
|
|
"loss": 0.6131,
|
||
|
|
"step": 159
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.41,
|
||
|
|
"learning_rate": 6.017650511111662e-06,
|
||
|
|
"loss": 0.6446,
|
||
|
|
"step": 160
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.41,
|
||
|
|
"learning_rate": 5.98248290623529e-06,
|
||
|
|
"loss": 0.5882,
|
||
|
|
"step": 161
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.41,
|
||
|
|
"learning_rate": 5.947213440218726e-06,
|
||
|
|
"loss": 0.5134,
|
||
|
|
"step": 162
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.42,
|
||
|
|
"learning_rate": 5.911844536420788e-06,
|
||
|
|
"loss": 0.6261,
|
||
|
|
"step": 163
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.42,
|
||
|
|
"learning_rate": 5.8763786250326455e-06,
|
||
|
|
"loss": 0.5609,
|
||
|
|
"step": 164
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.42,
|
||
|
|
"learning_rate": 5.8408181429108425e-06,
|
||
|
|
"loss": 0.5729,
|
||
|
|
"step": 165
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.42,
|
||
|
|
"learning_rate": 5.805165533409863e-06,
|
||
|
|
"loss": 0.5378,
|
||
|
|
"step": 166
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.43,
|
||
|
|
"learning_rate": 5.769423246214248e-06,
|
||
|
|
"loss": 0.5914,
|
||
|
|
"step": 167
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.43,
|
||
|
|
"learning_rate": 5.733593737170271e-06,
|
||
|
|
"loss": 0.5126,
|
||
|
|
"step": 168
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.43,
|
||
|
|
"learning_rate": 5.6976794681172075e-06,
|
||
|
|
"loss": 0.5309,
|
||
|
|
"step": 169
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.43,
|
||
|
|
"learning_rate": 5.661682906718181e-06,
|
||
|
|
"loss": 0.5544,
|
||
|
|
"step": 170
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.44,
|
||
|
|
"learning_rate": 5.625606526290604e-06,
|
||
|
|
"loss": 0.552,
|
||
|
|
"step": 171
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.44,
|
||
|
|
"learning_rate": 5.5894528056362406e-06,
|
||
|
|
"loss": 0.5787,
|
||
|
|
"step": 172
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.44,
|
||
|
|
"learning_rate": 5.55322422887089e-06,
|
||
|
|
"loss": 0.5153,
|
||
|
|
"step": 173
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.45,
|
||
|
|
"learning_rate": 5.516923285253701e-06,
|
||
|
|
"loss": 0.4472,
|
||
|
|
"step": 174
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.45,
|
||
|
|
"learning_rate": 5.4805524690161325e-06,
|
||
|
|
"loss": 0.6333,
|
||
|
|
"step": 175
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.45,
|
||
|
|
"learning_rate": 5.444114279190586e-06,
|
||
|
|
"loss": 0.4386,
|
||
|
|
"step": 176
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.45,
|
||
|
|
"learning_rate": 5.407611219438685e-06,
|
||
|
|
"loss": 0.613,
|
||
|
|
"step": 177
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.46,
|
||
|
|
"learning_rate": 5.371045797879255e-06,
|
||
|
|
"loss": 0.5707,
|
||
|
|
"step": 178
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.46,
|
||
|
|
"learning_rate": 5.334420526915993e-06,
|
||
|
|
"loss": 0.5866,
|
||
|
|
"step": 179
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.46,
|
||
|
|
"learning_rate": 5.297737923064836e-06,
|
||
|
|
"loss": 0.4922,
|
||
|
|
"step": 180
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.46,
|
||
|
|
"learning_rate": 5.261000506781051e-06,
|
||
|
|
"loss": 0.5826,
|
||
|
|
"step": 181
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.47,
|
||
|
|
"learning_rate": 5.224210802286064e-06,
|
||
|
|
"loss": 0.6741,
|
||
|
|
"step": 182
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.47,
|
||
|
|
"learning_rate": 5.187371337394009e-06,
|
||
|
|
"loss": 0.3813,
|
||
|
|
"step": 183
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.47,
|
||
|
|
"learning_rate": 5.150484643338051e-06,
|
||
|
|
"loss": 0.6125,
|
||
|
|
"step": 184
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.47,
|
||
|
|
"learning_rate": 5.113553254596463e-06,
|
||
|
|
"loss": 0.66,
|
||
|
|
"step": 185
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.48,
|
||
|
|
"learning_rate": 5.076579708718481e-06,
|
||
|
|
"loss": 0.5914,
|
||
|
|
"step": 186
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.48,
|
||
|
|
"learning_rate": 5.039566546149946e-06,
|
||
|
|
"loss": 0.4527,
|
||
|
|
"step": 187
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.48,
|
||
|
|
"learning_rate": 5.002516310058766e-06,
|
||
|
|
"loss": 0.5896,
|
||
|
|
"step": 188
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.48,
|
||
|
|
"learning_rate": 4.965431546160153e-06,
|
||
|
|
"loss": 0.4518,
|
||
|
|
"step": 189
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.49,
|
||
|
|
"learning_rate": 4.928314802541726e-06,
|
||
|
|
"loss": 0.5798,
|
||
|
|
"step": 190
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.49,
|
||
|
|
"learning_rate": 4.891168629488419e-06,
|
||
|
|
"loss": 0.5669,
|
||
|
|
"step": 191
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.49,
|
||
|
|
"learning_rate": 4.853995579307262e-06,
|
||
|
|
"loss": 0.6075,
|
||
|
|
"step": 192
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.49,
|
||
|
|
"learning_rate": 4.816798206152006e-06,
|
||
|
|
"loss": 0.6057,
|
||
|
|
"step": 193
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5,
|
||
|
|
"learning_rate": 4.77957906584763e-06,
|
||
|
|
"loss": 0.5076,
|
||
|
|
"step": 194
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5,
|
||
|
|
"learning_rate": 4.742340715714727e-06,
|
||
|
|
"loss": 0.5997,
|
||
|
|
"step": 195
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5,
|
||
|
|
"learning_rate": 4.705085714393797e-06,
|
||
|
|
"loss": 0.5709,
|
||
|
|
"step": 196
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5,
|
||
|
|
"learning_rate": 4.667816621669442e-06,
|
||
|
|
"loss": 0.5029,
|
||
|
|
"step": 197
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.51,
|
||
|
|
"learning_rate": 4.630535998294477e-06,
|
||
|
|
"loss": 0.5885,
|
||
|
|
"step": 198
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.51,
|
||
|
|
"learning_rate": 4.5932464058139885e-06,
|
||
|
|
"loss": 0.5776,
|
||
|
|
"step": 199
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.51,
|
||
|
|
"learning_rate": 4.55595040638933e-06,
|
||
|
|
"loss": 0.4949,
|
||
|
|
"step": 200
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.51,
|
||
|
|
"learning_rate": 4.5186505626220725e-06,
|
||
|
|
"loss": 0.4202,
|
||
|
|
"step": 201
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.52,
|
||
|
|
"learning_rate": 4.48134943737793e-06,
|
||
|
|
"loss": 0.5012,
|
||
|
|
"step": 202
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.52,
|
||
|
|
"learning_rate": 4.444049593610671e-06,
|
||
|
|
"loss": 0.613,
|
||
|
|
"step": 203
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.52,
|
||
|
|
"learning_rate": 4.406753594186011e-06,
|
||
|
|
"loss": 0.4806,
|
||
|
|
"step": 204
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.52,
|
||
|
|
"learning_rate": 4.369464001705524e-06,
|
||
|
|
"loss": 0.6567,
|
||
|
|
"step": 205
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.53,
|
||
|
|
"learning_rate": 4.332183378330558e-06,
|
||
|
|
"loss": 0.3533,
|
||
|
|
"step": 206
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.53,
|
||
|
|
"learning_rate": 4.294914285606203e-06,
|
||
|
|
"loss": 0.7252,
|
||
|
|
"step": 207
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.53,
|
||
|
|
"learning_rate": 4.257659284285274e-06,
|
||
|
|
"loss": 0.4848,
|
||
|
|
"step": 208
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.53,
|
||
|
|
"learning_rate": 4.220420934152371e-06,
|
||
|
|
"loss": 0.3598,
|
||
|
|
"step": 209
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.54,
|
||
|
|
"learning_rate": 4.1832017938479936e-06,
|
||
|
|
"loss": 0.5295,
|
||
|
|
"step": 210
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.54,
|
||
|
|
"learning_rate": 4.146004420692739e-06,
|
||
|
|
"loss": 0.5091,
|
||
|
|
"step": 211
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.54,
|
||
|
|
"learning_rate": 4.108831370511581e-06,
|
||
|
|
"loss": 0.5788,
|
||
|
|
"step": 212
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.54,
|
||
|
|
"learning_rate": 4.071685197458274e-06,
|
||
|
|
"loss": 0.462,
|
||
|
|
"step": 213
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.55,
|
||
|
|
"learning_rate": 4.034568453839847e-06,
|
||
|
|
"loss": 0.5832,
|
||
|
|
"step": 214
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.55,
|
||
|
|
"learning_rate": 3.997483689941234e-06,
|
||
|
|
"loss": 0.4365,
|
||
|
|
"step": 215
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.55,
|
||
|
|
"learning_rate": 3.960433453850053e-06,
|
||
|
|
"loss": 0.6247,
|
||
|
|
"step": 216
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.55,
|
||
|
|
"learning_rate": 3.923420291281522e-06,
|
||
|
|
"loss": 0.538,
|
||
|
|
"step": 217
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.56,
|
||
|
|
"learning_rate": 3.886446745403538e-06,
|
||
|
|
"loss": 0.5798,
|
||
|
|
"step": 218
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.56,
|
||
|
|
"learning_rate": 3.849515356661949e-06,
|
||
|
|
"loss": 0.568,
|
||
|
|
"step": 219
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.56,
|
||
|
|
"learning_rate": 3.8126286626059916e-06,
|
||
|
|
"loss": 0.4764,
|
||
|
|
"step": 220
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.57,
|
||
|
|
"learning_rate": 3.7757891977139374e-06,
|
||
|
|
"loss": 0.5134,
|
||
|
|
"step": 221
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.57,
|
||
|
|
"learning_rate": 3.738999493218949e-06,
|
||
|
|
"loss": 0.5493,
|
||
|
|
"step": 222
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.57,
|
||
|
|
"learning_rate": 3.7022620769351665e-06,
|
||
|
|
"loss": 0.5529,
|
||
|
|
"step": 223
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.57,
|
||
|
|
"learning_rate": 3.665579473084008e-06,
|
||
|
|
"loss": 0.4234,
|
||
|
|
"step": 224
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.58,
|
||
|
|
"learning_rate": 3.6289542021207454e-06,
|
||
|
|
"loss": 0.6216,
|
||
|
|
"step": 225
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.58,
|
||
|
|
"eval_accuracy": 0.8257179536751563,
|
||
|
|
"eval_accuracy_<|content|>": 1.0,
|
||
|
|
"eval_accuracy_<|from|>": 0.9924146649810367,
|
||
|
|
"eval_accuracy_<|recipient|>": 1.0,
|
||
|
|
"eval_accuracy_<|stop|>": 0.9440158695173022,
|
||
|
|
"eval_accuracy_total_num_<|content|>": 5362,
|
||
|
|
"eval_accuracy_total_num_<|from|>": 791,
|
||
|
|
"eval_accuracy_total_num_<|recipient|>": 791,
|
||
|
|
"eval_accuracy_total_num_<|stop|>": 4537,
|
||
|
|
"eval_loss": 0.5591428279876709,
|
||
|
|
"eval_perplexity": 1.0623002333585039,
|
||
|
|
"eval_runtime": 223.7635,
|
||
|
|
"eval_samples_per_second": 6.145,
|
||
|
|
"eval_steps_per_second": 0.192,
|
||
|
|
"step": 225
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.58,
|
||
|
|
"learning_rate": 3.5923887805613165e-06,
|
||
|
|
"loss": 0.6102,
|
||
|
|
"step": 226
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.58,
|
||
|
|
"learning_rate": 3.5558857208094145e-06,
|
||
|
|
"loss": 0.5378,
|
||
|
|
"step": 227
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.58,
|
||
|
|
"learning_rate": 3.5194475309838677e-06,
|
||
|
|
"loss": 0.6289,
|
||
|
|
"step": 228
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.59,
|
||
|
|
"learning_rate": 3.4830767147463014e-06,
|
||
|
|
"loss": 0.3814,
|
||
|
|
"step": 229
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.59,
|
||
|
|
"learning_rate": 3.4467757711291108e-06,
|
||
|
|
"loss": 0.5506,
|
||
|
|
"step": 230
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.59,
|
||
|
|
"learning_rate": 3.4105471943637592e-06,
|
||
|
|
"loss": 0.5942,
|
||
|
|
"step": 231
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.59,
|
||
|
|
"learning_rate": 3.374393473709396e-06,
|
||
|
|
"loss": 0.5092,
|
||
|
|
"step": 232
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6,
|
||
|
|
"learning_rate": 3.3383170932818184e-06,
|
||
|
|
"loss": 0.5533,
|
||
|
|
"step": 233
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6,
|
||
|
|
"learning_rate": 3.302320531882792e-06,
|
||
|
|
"loss": 0.454,
|
||
|
|
"step": 234
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6,
|
||
|
|
"learning_rate": 3.266406262829731e-06,
|
||
|
|
"loss": 0.5039,
|
||
|
|
"step": 235
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6,
|
||
|
|
"learning_rate": 3.2305767537857534e-06,
|
||
|
|
"loss": 0.5447,
|
||
|
|
"step": 236
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.61,
|
||
|
|
"learning_rate": 3.194834466590136e-06,
|
||
|
|
"loss": 0.4985,
|
||
|
|
"step": 237
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.61,
|
||
|
|
"learning_rate": 3.1591818570891573e-06,
|
||
|
|
"loss": 0.6181,
|
||
|
|
"step": 238
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.61,
|
||
|
|
"learning_rate": 3.1236213749673555e-06,
|
||
|
|
"loss": 0.5648,
|
||
|
|
"step": 239
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.61,
|
||
|
|
"learning_rate": 3.0881554635792124e-06,
|
||
|
|
"loss": 0.6169,
|
||
|
|
"step": 240
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.62,
|
||
|
|
"learning_rate": 3.0527865597812755e-06,
|
||
|
|
"loss": 0.6264,
|
||
|
|
"step": 241
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.62,
|
||
|
|
"learning_rate": 3.017517093764711e-06,
|
||
|
|
"loss": 0.5292,
|
||
|
|
"step": 242
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.62,
|
||
|
|
"learning_rate": 2.982349488888337e-06,
|
||
|
|
"loss": 0.5578,
|
||
|
|
"step": 243
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.62,
|
||
|
|
"learning_rate": 2.9472861615121117e-06,
|
||
|
|
"loss": 0.4992,
|
||
|
|
"step": 244
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.63,
|
||
|
|
"learning_rate": 2.9123295208311007e-06,
|
||
|
|
"loss": 0.5403,
|
||
|
|
"step": 245
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.63,
|
||
|
|
"learning_rate": 2.877481968709948e-06,
|
||
|
|
"loss": 0.5385,
|
||
|
|
"step": 246
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.63,
|
||
|
|
"learning_rate": 2.842745899517843e-06,
|
||
|
|
"loss": 0.4517,
|
||
|
|
"step": 247
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.63,
|
||
|
|
"learning_rate": 2.8081236999639976e-06,
|
||
|
|
"loss": 0.5326,
|
||
|
|
"step": 248
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.64,
|
||
|
|
"learning_rate": 2.7736177489336662e-06,
|
||
|
|
"loss": 0.4631,
|
||
|
|
"step": 249
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.64,
|
||
|
|
"learning_rate": 2.739230417324686e-06,
|
||
|
|
"loss": 0.5639,
|
||
|
|
"step": 250
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.64,
|
||
|
|
"learning_rate": 2.7049640678845723e-06,
|
||
|
|
"loss": 0.4956,
|
||
|
|
"step": 251
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.64,
|
||
|
|
"learning_rate": 2.6708210550481794e-06,
|
||
|
|
"loss": 0.5281,
|
||
|
|
"step": 252
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.65,
|
||
|
|
"learning_rate": 2.636803724775927e-06,
|
||
|
|
"loss": 0.2976,
|
||
|
|
"step": 253
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.65,
|
||
|
|
"learning_rate": 2.6029144143926003e-06,
|
||
|
|
"loss": 0.5906,
|
||
|
|
"step": 254
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.65,
|
||
|
|
"learning_rate": 2.569155452426767e-06,
|
||
|
|
"loss": 0.5625,
|
||
|
|
"step": 255
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.65,
|
||
|
|
"learning_rate": 2.5355291584507814e-06,
|
||
|
|
"loss": 0.5878,
|
||
|
|
"step": 256
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.66,
|
||
|
|
"learning_rate": 2.5020378429213904e-06,
|
||
|
|
"loss": 0.595,
|
||
|
|
"step": 257
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.66,
|
||
|
|
"learning_rate": 2.4686838070210094e-06,
|
||
|
|
"loss": 0.3659,
|
||
|
|
"step": 258
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.66,
|
||
|
|
"learning_rate": 2.435469342499587e-06,
|
||
|
|
"loss": 0.5602,
|
||
|
|
"step": 259
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.66,
|
||
|
|
"learning_rate": 2.402396731517147e-06,
|
||
|
|
"loss": 0.4733,
|
||
|
|
"step": 260
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.67,
|
||
|
|
"learning_rate": 2.3694682464869856e-06,
|
||
|
|
"loss": 0.5969,
|
||
|
|
"step": 261
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.67,
|
||
|
|
"learning_rate": 2.336686149919527e-06,
|
||
|
|
"loss": 0.4799,
|
||
|
|
"step": 262
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.67,
|
||
|
|
"learning_rate": 2.3040526942668707e-06,
|
||
|
|
"loss": 0.3591,
|
||
|
|
"step": 263
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.68,
|
||
|
|
"learning_rate": 2.271570121768021e-06,
|
||
|
|
"loss": 0.4472,
|
||
|
|
"step": 264
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.68,
|
||
|
|
"learning_rate": 2.239240664294837e-06,
|
||
|
|
"loss": 0.4683,
|
||
|
|
"step": 265
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.68,
|
||
|
|
"learning_rate": 2.2070665431986617e-06,
|
||
|
|
"loss": 0.4283,
|
||
|
|
"step": 266
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.68,
|
||
|
|
"learning_rate": 2.1750499691577055e-06,
|
||
|
|
"loss": 0.6666,
|
||
|
|
"step": 267
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.69,
|
||
|
|
"learning_rate": 2.1431931420251543e-06,
|
||
|
|
"loss": 0.5802,
|
||
|
|
"step": 268
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.69,
|
||
|
|
"learning_rate": 2.1114982506779997e-06,
|
||
|
|
"loss": 0.5939,
|
||
|
|
"step": 269
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.69,
|
||
|
|
"learning_rate": 2.0799674728666665e-06,
|
||
|
|
"loss": 0.5326,
|
||
|
|
"step": 270
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.69,
|
||
|
|
"learning_rate": 2.0486029750653605e-06,
|
||
|
|
"loss": 0.6165,
|
||
|
|
"step": 271
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7,
|
||
|
|
"learning_rate": 2.017406912323217e-06,
|
||
|
|
"loss": 0.557,
|
||
|
|
"step": 272
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7,
|
||
|
|
"learning_rate": 1.986381428116232e-06,
|
||
|
|
"loss": 0.5167,
|
||
|
|
"step": 273
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7,
|
||
|
|
"learning_rate": 1.9555286541999766e-06,
|
||
|
|
"loss": 0.6048,
|
||
|
|
"step": 274
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7,
|
||
|
|
"learning_rate": 1.92485071046313e-06,
|
||
|
|
"loss": 0.4407,
|
||
|
|
"step": 275
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.71,
|
||
|
|
"learning_rate": 1.8943497047818179e-06,
|
||
|
|
"loss": 0.6113,
|
||
|
|
"step": 276
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.71,
|
||
|
|
"learning_rate": 1.864027732874788e-06,
|
||
|
|
"loss": 0.7606,
|
||
|
|
"step": 277
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.71,
|
||
|
|
"learning_rate": 1.8338868781594052e-06,
|
||
|
|
"loss": 0.6256,
|
||
|
|
"step": 278
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.71,
|
||
|
|
"learning_rate": 1.8039292116085027e-06,
|
||
|
|
"loss": 0.6035,
|
||
|
|
"step": 279
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.72,
|
||
|
|
"learning_rate": 1.7741567916080946e-06,
|
||
|
|
"loss": 0.5717,
|
||
|
|
"step": 280
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.72,
|
||
|
|
"learning_rate": 1.7445716638159248e-06,
|
||
|
|
"loss": 0.475,
|
||
|
|
"step": 281
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.72,
|
||
|
|
"learning_rate": 1.715175861020934e-06,
|
||
|
|
"loss": 0.4888,
|
||
|
|
"step": 282
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.72,
|
||
|
|
"learning_rate": 1.6859714030035694e-06,
|
||
|
|
"loss": 0.5913,
|
||
|
|
"step": 283
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.73,
|
||
|
|
"learning_rate": 1.6569602963970123e-06,
|
||
|
|
"loss": 0.3934,
|
||
|
|
"step": 284
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.73,
|
||
|
|
"learning_rate": 1.628144534549307e-06,
|
||
|
|
"loss": 0.5487,
|
||
|
|
"step": 285
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.73,
|
||
|
|
"learning_rate": 1.5995260973863885e-06,
|
||
|
|
"loss": 0.494,
|
||
|
|
"step": 286
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.73,
|
||
|
|
"learning_rate": 1.5711069512760497e-06,
|
||
|
|
"loss": 0.4315,
|
||
|
|
"step": 287
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.74,
|
||
|
|
"learning_rate": 1.5428890488928284e-06,
|
||
|
|
"loss": 0.4848,
|
||
|
|
"step": 288
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.74,
|
||
|
|
"learning_rate": 1.5148743290838455e-06,
|
||
|
|
"loss": 0.4264,
|
||
|
|
"step": 289
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.74,
|
||
|
|
"learning_rate": 1.4870647167355795e-06,
|
||
|
|
"loss": 0.6395,
|
||
|
|
"step": 290
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.74,
|
||
|
|
"learning_rate": 1.4594621226416103e-06,
|
||
|
|
"loss": 0.5726,
|
||
|
|
"step": 291
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.75,
|
||
|
|
"learning_rate": 1.4320684433713354e-06,
|
||
|
|
"loss": 0.5543,
|
||
|
|
"step": 292
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.75,
|
||
|
|
"learning_rate": 1.4048855611396424e-06,
|
||
|
|
"loss": 0.6457,
|
||
|
|
"step": 293
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.75,
|
||
|
|
"learning_rate": 1.3779153436776006e-06,
|
||
|
|
"loss": 0.3915,
|
||
|
|
"step": 294
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.75,
|
||
|
|
"learning_rate": 1.351159644104115e-06,
|
||
|
|
"loss": 0.5187,
|
||
|
|
"step": 295
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.76,
|
||
|
|
"learning_rate": 1.3246203007986048e-06,
|
||
|
|
"loss": 0.6015,
|
||
|
|
"step": 296
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.76,
|
||
|
|
"learning_rate": 1.298299137274692e-06,
|
||
|
|
"loss": 0.6067,
|
||
|
|
"step": 297
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.76,
|
||
|
|
"learning_rate": 1.2721979620548955e-06,
|
||
|
|
"loss": 0.4431,
|
||
|
|
"step": 298
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.76,
|
||
|
|
"learning_rate": 1.2463185685463859e-06,
|
||
|
|
"loss": 0.3615,
|
||
|
|
"step": 299
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.77,
|
||
|
|
"learning_rate": 1.220662734917746e-06,
|
||
|
|
"loss": 0.5003,
|
||
|
|
"step": 300
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.77,
|
||
|
|
"eval_accuracy": 0.8336343532396525,
|
||
|
|
"eval_accuracy_<|content|>": 1.0,
|
||
|
|
"eval_accuracy_<|from|>": 0.9873577749683944,
|
||
|
|
"eval_accuracy_<|recipient|>": 1.0,
|
||
|
|
"eval_accuracy_<|stop|>": 0.91359929468812,
|
||
|
|
"eval_accuracy_total_num_<|content|>": 5362,
|
||
|
|
"eval_accuracy_total_num_<|from|>": 791,
|
||
|
|
"eval_accuracy_total_num_<|recipient|>": 791,
|
||
|
|
"eval_accuracy_total_num_<|stop|>": 4537,
|
||
|
|
"eval_loss": 0.5285482406616211,
|
||
|
|
"eval_perplexity": 1.0587839466417142,
|
||
|
|
"eval_runtime": 223.59,
|
||
|
|
"eval_samples_per_second": 6.15,
|
||
|
|
"eval_steps_per_second": 0.192,
|
||
|
|
"step": 300
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.77,
|
||
|
|
"learning_rate": 1.1952322239767983e-06,
|
||
|
|
"loss": 0.3402,
|
||
|
|
"step": 301
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.77,
|
||
|
|
"learning_rate": 1.170028783049487e-06,
|
||
|
|
"loss": 0.6256,
|
||
|
|
"step": 302
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.77,
|
||
|
|
"learning_rate": 1.1450541438598118e-06,
|
||
|
|
"loss": 0.5378,
|
||
|
|
"step": 303
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.78,
|
||
|
|
"learning_rate": 1.1203100224108464e-06,
|
||
|
|
"loss": 0.5334,
|
||
|
|
"step": 304
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.78,
|
||
|
|
"learning_rate": 1.09579811886683e-06,
|
||
|
|
"loss": 0.4257,
|
||
|
|
"step": 305
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.78,
|
||
|
|
"learning_rate": 1.0715201174363525e-06,
|
||
|
|
"loss": 0.4915,
|
||
|
|
"step": 306
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.79,
|
||
|
|
"learning_rate": 1.0474776862566299e-06,
|
||
|
|
"loss": 0.4102,
|
||
|
|
"step": 307
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.79,
|
||
|
|
"learning_rate": 1.0236724772788846e-06,
|
||
|
|
"loss": 0.3463,
|
||
|
|
"step": 308
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.79,
|
||
|
|
"learning_rate": 1.00010612615485e-06,
|
||
|
|
"loss": 0.5686,
|
||
|
|
"step": 309
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.79,
|
||
|
|
"learning_rate": 9.76780252124369e-07,
|
||
|
|
"loss": 0.5914,
|
||
|
|
"step": 310
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8,
|
||
|
|
"learning_rate": 9.536964579041548e-07,
|
||
|
|
"loss": 0.6401,
|
||
|
|
"step": 311
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8,
|
||
|
|
"learning_rate": 9.308563295776531e-07,
|
||
|
|
"loss": 0.5474,
|
||
|
|
"step": 312
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8,
|
||
|
|
"learning_rate": 9.082614364860701e-07,
|
||
|
|
"loss": 0.4519,
|
||
|
|
"step": 313
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8,
|
||
|
|
"learning_rate": 8.859133311205453e-07,
|
||
|
|
"loss": 0.5629,
|
||
|
|
"step": 314
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.81,
|
||
|
|
"learning_rate": 8.638135490154735e-07,
|
||
|
|
"loss": 0.4641,
|
||
|
|
"step": 315
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.81,
|
||
|
|
"learning_rate": 8.419636086430022e-07,
|
||
|
|
"loss": 0.4696,
|
||
|
|
"step": 316
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.81,
|
||
|
|
"learning_rate": 8.203650113086972e-07,
|
||
|
|
"loss": 0.538,
|
||
|
|
"step": 317
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.81,
|
||
|
|
"learning_rate": 7.990192410483916e-07,
|
||
|
|
"loss": 0.6092,
|
||
|
|
"step": 318
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.82,
|
||
|
|
"learning_rate": 7.779277645262103e-07,
|
||
|
|
"loss": 0.407,
|
||
|
|
"step": 319
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.82,
|
||
|
|
"learning_rate": 7.570920309338014e-07,
|
||
|
|
"loss": 0.5779,
|
||
|
|
"step": 320
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.82,
|
||
|
|
"learning_rate": 7.365134718907647e-07,
|
||
|
|
"loss": 0.5572,
|
||
|
|
"step": 321
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.82,
|
||
|
|
"learning_rate": 7.161935013462746e-07,
|
||
|
|
"loss": 0.6165,
|
||
|
|
"step": 322
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.83,
|
||
|
|
"learning_rate": 6.961335154819422e-07,
|
||
|
|
"loss": 0.5908,
|
||
|
|
"step": 323
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.83,
|
||
|
|
"learning_rate": 6.763348926158732e-07,
|
||
|
|
"loss": 0.5739,
|
||
|
|
"step": 324
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.83,
|
||
|
|
"learning_rate": 6.567989931079675e-07,
|
||
|
|
"loss": 0.5375,
|
||
|
|
"step": 325
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.83,
|
||
|
|
"learning_rate": 6.375271592664525e-07,
|
||
|
|
"loss": 0.5726,
|
||
|
|
"step": 326
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.84,
|
||
|
|
"learning_rate": 6.18520715255646e-07,
|
||
|
|
"loss": 0.5559,
|
||
|
|
"step": 327
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.84,
|
||
|
|
"learning_rate": 5.997809670049795e-07,
|
||
|
|
"loss": 0.5309,
|
||
|
|
"step": 328
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.84,
|
||
|
|
"learning_rate": 5.813092021192639e-07,
|
||
|
|
"loss": 0.5068,
|
||
|
|
"step": 329
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.84,
|
||
|
|
"learning_rate": 5.631066897902227e-07,
|
||
|
|
"loss": 0.4953,
|
||
|
|
"step": 330
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.85,
|
||
|
|
"learning_rate": 5.451746807092811e-07,
|
||
|
|
"loss": 0.5307,
|
||
|
|
"step": 331
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.85,
|
||
|
|
"learning_rate": 5.275144069816338e-07,
|
||
|
|
"loss": 0.5182,
|
||
|
|
"step": 332
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.85,
|
||
|
|
"learning_rate": 5.101270820415908e-07,
|
||
|
|
"loss": 0.5599,
|
||
|
|
"step": 333
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.85,
|
||
|
|
"learning_rate": 4.930139005691915e-07,
|
||
|
|
"loss": 0.4452,
|
||
|
|
"step": 334
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.86,
|
||
|
|
"learning_rate": 4.761760384081335e-07,
|
||
|
|
"loss": 0.5605,
|
||
|
|
"step": 335
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.86,
|
||
|
|
"learning_rate": 4.596146524849682e-07,
|
||
|
|
"loss": 0.3468,
|
||
|
|
"step": 336
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.86,
|
||
|
|
"learning_rate": 4.433308807296139e-07,
|
||
|
|
"loss": 0.5684,
|
||
|
|
"step": 337
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.86,
|
||
|
|
"learning_rate": 4.2732584199717023e-07,
|
||
|
|
"loss": 0.3672,
|
||
|
|
"step": 338
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.87,
|
||
|
|
"learning_rate": 4.116006359910375e-07,
|
||
|
|
"loss": 0.5228,
|
||
|
|
"step": 339
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.87,
|
||
|
|
"learning_rate": 3.961563431873597e-07,
|
||
|
|
"loss": 0.4777,
|
||
|
|
"step": 340
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.87,
|
||
|
|
"learning_rate": 3.809940247607826e-07,
|
||
|
|
"loss": 0.4455,
|
||
|
|
"step": 341
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.87,
|
||
|
|
"learning_rate": 3.661147225115447e-07,
|
||
|
|
"loss": 0.5601,
|
||
|
|
"step": 342
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.88,
|
||
|
|
"learning_rate": 3.515194587938898e-07,
|
||
|
|
"loss": 0.5627,
|
||
|
|
"step": 343
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.88,
|
||
|
|
"learning_rate": 3.3720923644582554e-07,
|
||
|
|
"loss": 0.4795,
|
||
|
|
"step": 344
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.88,
|
||
|
|
"learning_rate": 3.23185038720218e-07,
|
||
|
|
"loss": 0.481,
|
||
|
|
"step": 345
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.88,
|
||
|
|
"learning_rate": 3.0944782921722787e-07,
|
||
|
|
"loss": 0.4925,
|
||
|
|
"step": 346
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.89,
|
||
|
|
"learning_rate": 2.959985518181108e-07,
|
||
|
|
"loss": 0.481,
|
||
|
|
"step": 347
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.89,
|
||
|
|
"learning_rate": 2.8283813062035433e-07,
|
||
|
|
"loss": 0.462,
|
||
|
|
"step": 348
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.89,
|
||
|
|
"learning_rate": 2.699674698741872e-07,
|
||
|
|
"loss": 0.5985,
|
||
|
|
"step": 349
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9,
|
||
|
|
"learning_rate": 2.5738745392045136e-07,
|
||
|
|
"loss": 0.4534,
|
||
|
|
"step": 350
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9,
|
||
|
|
"learning_rate": 2.4509894712983346e-07,
|
||
|
|
"loss": 0.4966,
|
||
|
|
"step": 351
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9,
|
||
|
|
"learning_rate": 2.3310279384347738e-07,
|
||
|
|
"loss": 0.4829,
|
||
|
|
"step": 352
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9,
|
||
|
|
"learning_rate": 2.2139981831496808e-07,
|
||
|
|
"loss": 0.5734,
|
||
|
|
"step": 353
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.91,
|
||
|
|
"learning_rate": 2.099908246537011e-07,
|
||
|
|
"loss": 0.3882,
|
||
|
|
"step": 354
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.91,
|
||
|
|
"learning_rate": 1.9887659676962695e-07,
|
||
|
|
"loss": 0.508,
|
||
|
|
"step": 355
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.91,
|
||
|
|
"learning_rate": 1.8805789831939146e-07,
|
||
|
|
"loss": 0.3879,
|
||
|
|
"step": 356
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.91,
|
||
|
|
"learning_rate": 1.7753547265386706e-07,
|
||
|
|
"loss": 0.4153,
|
||
|
|
"step": 357
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.92,
|
||
|
|
"learning_rate": 1.6731004276707185e-07,
|
||
|
|
"loss": 0.5335,
|
||
|
|
"step": 358
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.92,
|
||
|
|
"learning_rate": 1.5738231124649955e-07,
|
||
|
|
"loss": 0.5547,
|
||
|
|
"step": 359
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.92,
|
||
|
|
"learning_rate": 1.4775296022483897e-07,
|
||
|
|
"loss": 0.4444,
|
||
|
|
"step": 360
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.92,
|
||
|
|
"learning_rate": 1.3842265133310762e-07,
|
||
|
|
"loss": 0.536,
|
||
|
|
"step": 361
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.93,
|
||
|
|
"learning_rate": 1.2939202565519243e-07,
|
||
|
|
"loss": 0.4738,
|
||
|
|
"step": 362
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.93,
|
||
|
|
"learning_rate": 1.2066170368379763e-07,
|
||
|
|
"loss": 0.4359,
|
||
|
|
"step": 363
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.93,
|
||
|
|
"learning_rate": 1.1223228527781281e-07,
|
||
|
|
"loss": 0.4132,
|
||
|
|
"step": 364
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.93,
|
||
|
|
"learning_rate": 1.0410434962109633e-07,
|
||
|
|
"loss": 0.514,
|
||
|
|
"step": 365
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.94,
|
||
|
|
"learning_rate": 9.627845518268069e-08,
|
||
|
|
"loss": 0.4843,
|
||
|
|
"step": 366
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.94,
|
||
|
|
"learning_rate": 8.875513967839826e-08,
|
||
|
|
"loss": 0.4589,
|
||
|
|
"step": 367
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.94,
|
||
|
|
"learning_rate": 8.15349200339363e-08,
|
||
|
|
"loss": 0.5233,
|
||
|
|
"step": 368
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.94,
|
||
|
|
"learning_rate": 7.461829234931988e-08,
|
||
|
|
"loss": 0.3706,
|
||
|
|
"step": 369
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.95,
|
||
|
|
"learning_rate": 6.800573186482111e-08,
|
||
|
|
"loss": 0.4884,
|
||
|
|
"step": 370
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.95,
|
||
|
|
"learning_rate": 6.169769292831134e-08,
|
||
|
|
"loss": 0.4554,
|
||
|
|
"step": 371
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.95,
|
||
|
|
"learning_rate": 5.569460896403755e-08,
|
||
|
|
"loss": 0.4819,
|
||
|
|
"step": 372
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.95,
|
||
|
|
"learning_rate": 4.999689244284472e-08,
|
||
|
|
"loss": 0.49,
|
||
|
|
"step": 373
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.96,
|
||
|
|
"learning_rate": 4.4604934853834986e-08,
|
||
|
|
"loss": 0.6322,
|
||
|
|
"step": 374
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.96,
|
||
|
|
"learning_rate": 3.9519106677467664e-08,
|
||
|
|
"loss": 0.4662,
|
||
|
|
"step": 375
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.96,
|
||
|
|
"eval_accuracy": 0.835875467640784,
|
||
|
|
"eval_accuracy_<|content|>": 1.0,
|
||
|
|
"eval_accuracy_<|from|>": 0.9949431099873578,
|
||
|
|
"eval_accuracy_<|recipient|>": 1.0,
|
||
|
|
"eval_accuracy_<|stop|>": 0.9279259422525898,
|
||
|
|
"eval_accuracy_total_num_<|content|>": 5362,
|
||
|
|
"eval_accuracy_total_num_<|from|>": 791,
|
||
|
|
"eval_accuracy_total_num_<|recipient|>": 791,
|
||
|
|
"eval_accuracy_total_num_<|stop|>": 4537,
|
||
|
|
"eval_loss": 0.5191700458526611,
|
||
|
|
"eval_perplexity": 1.057695465829475,
|
||
|
|
"eval_runtime": 223.7003,
|
||
|
|
"eval_samples_per_second": 6.147,
|
||
|
|
"eval_steps_per_second": 0.192,
|
||
|
|
"step": 375
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.96,
|
||
|
|
"learning_rate": 3.4739757360103594e-08,
|
||
|
|
"loss": 0.4331,
|
||
|
|
"step": 376
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.96,
|
||
|
|
"learning_rate": 3.026721528999487e-08,
|
||
|
|
"loss": 0.4685,
|
||
|
|
"step": 377
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.97,
|
||
|
|
"learning_rate": 2.6101787774722885e-08,
|
||
|
|
"loss": 0.479,
|
||
|
|
"step": 378
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.97,
|
||
|
|
"learning_rate": 2.224376102007919e-08,
|
||
|
|
"loss": 0.5054,
|
||
|
|
"step": 379
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.97,
|
||
|
|
"learning_rate": 1.8693400110406267e-08,
|
||
|
|
"loss": 0.5825,
|
||
|
|
"step": 380
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.97,
|
||
|
|
"learning_rate": 1.5450948990378077e-08,
|
||
|
|
"loss": 0.4259,
|
||
|
|
"step": 381
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.98,
|
||
|
|
"learning_rate": 1.2516630448241973e-08,
|
||
|
|
"loss": 0.4667,
|
||
|
|
"step": 382
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.98,
|
||
|
|
"learning_rate": 9.890646100509937e-09,
|
||
|
|
"loss": 0.5553,
|
||
|
|
"step": 383
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.98,
|
||
|
|
"learning_rate": 7.573176378104674e-09,
|
||
|
|
"loss": 0.3993,
|
||
|
|
"step": 384
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.98,
|
||
|
|
"learning_rate": 5.564380513964518e-09,
|
||
|
|
"loss": 0.4619,
|
||
|
|
"step": 385
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.99,
|
||
|
|
"learning_rate": 3.864396532100689e-09,
|
||
|
|
"loss": 0.4039,
|
||
|
|
"step": 386
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.99,
|
||
|
|
"learning_rate": 2.4733412381143796e-09,
|
||
|
|
"loss": 0.4518,
|
||
|
|
"step": 387
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.99,
|
||
|
|
"learning_rate": 1.3913102111691722e-09,
|
||
|
|
"loss": 0.4452,
|
||
|
|
"step": 388
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.99,
|
||
|
|
"learning_rate": 6.183777974287929e-10,
|
||
|
|
"loss": 0.3681,
|
||
|
|
"step": 389
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0,
|
||
|
|
"learning_rate": 1.5459710494220015e-10,
|
||
|
|
"loss": 0.3691,
|
||
|
|
"step": 390
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0,
|
||
|
|
"learning_rate": 0.0,
|
||
|
|
"loss": 0.3283,
|
||
|
|
"step": 391
|
||
|
|
}
|
||
|
|
],
|
||
|
|
"logging_steps": 1.0,
|
||
|
|
"max_steps": 391,
|
||
|
|
"num_train_epochs": 1,
|
||
|
|
"save_steps": 200.0,
|
||
|
|
"total_flos": 163734890741760.0,
|
||
|
|
"trial_name": null,
|
||
|
|
"trial_params": null
|
||
|
|
}
|