1891 lines
37 KiB
JSON
1891 lines
37 KiB
JSON
{
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 0.9977871655602495,
|
|
"eval_steps": 500,
|
|
"global_step": 310,
|
|
"is_hyper_param_search": false,
|
|
"is_local_process_zero": true,
|
|
"is_world_process_zero": true,
|
|
"log_history": [
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 0.0,
|
|
"loss": 2.0536,
|
|
"step": 1
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 6.309297535714574e-06,
|
|
"loss": 1.8043,
|
|
"step": 2
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.681,
|
|
"step": 3
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.8204,
|
|
"step": 4
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 9.96742671009772e-06,
|
|
"loss": 1.7262,
|
|
"step": 5
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 9.93485342019544e-06,
|
|
"loss": 1.8368,
|
|
"step": 6
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 9.90228013029316e-06,
|
|
"loss": 1.8377,
|
|
"step": 7
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"learning_rate": 9.86970684039088e-06,
|
|
"loss": 1.6889,
|
|
"step": 8
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"learning_rate": 9.8371335504886e-06,
|
|
"loss": 1.57,
|
|
"step": 9
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"learning_rate": 9.80456026058632e-06,
|
|
"loss": 1.9013,
|
|
"step": 10
|
|
},
|
|
{
|
|
"epoch": 0.04,
|
|
"learning_rate": 9.771986970684039e-06,
|
|
"loss": 2.0751,
|
|
"step": 11
|
|
},
|
|
{
|
|
"epoch": 0.04,
|
|
"learning_rate": 9.73941368078176e-06,
|
|
"loss": 1.8623,
|
|
"step": 12
|
|
},
|
|
{
|
|
"epoch": 0.04,
|
|
"learning_rate": 9.706840390879479e-06,
|
|
"loss": 1.8295,
|
|
"step": 13
|
|
},
|
|
{
|
|
"epoch": 0.05,
|
|
"learning_rate": 9.6742671009772e-06,
|
|
"loss": 1.7735,
|
|
"step": 14
|
|
},
|
|
{
|
|
"epoch": 0.05,
|
|
"learning_rate": 9.641693811074918e-06,
|
|
"loss": 1.6631,
|
|
"step": 15
|
|
},
|
|
{
|
|
"epoch": 0.05,
|
|
"learning_rate": 9.609120521172639e-06,
|
|
"loss": 1.6938,
|
|
"step": 16
|
|
},
|
|
{
|
|
"epoch": 0.05,
|
|
"learning_rate": 9.576547231270358e-06,
|
|
"loss": 1.6516,
|
|
"step": 17
|
|
},
|
|
{
|
|
"epoch": 0.06,
|
|
"learning_rate": 9.543973941368079e-06,
|
|
"loss": 1.7867,
|
|
"step": 18
|
|
},
|
|
{
|
|
"epoch": 0.06,
|
|
"learning_rate": 9.511400651465798e-06,
|
|
"loss": 1.5865,
|
|
"step": 19
|
|
},
|
|
{
|
|
"epoch": 0.06,
|
|
"learning_rate": 9.478827361563519e-06,
|
|
"loss": 1.8717,
|
|
"step": 20
|
|
},
|
|
{
|
|
"epoch": 0.07,
|
|
"learning_rate": 9.446254071661238e-06,
|
|
"loss": 1.669,
|
|
"step": 21
|
|
},
|
|
{
|
|
"epoch": 0.07,
|
|
"learning_rate": 9.413680781758958e-06,
|
|
"loss": 1.6665,
|
|
"step": 22
|
|
},
|
|
{
|
|
"epoch": 0.07,
|
|
"learning_rate": 9.381107491856677e-06,
|
|
"loss": 1.8175,
|
|
"step": 23
|
|
},
|
|
{
|
|
"epoch": 0.08,
|
|
"learning_rate": 9.348534201954398e-06,
|
|
"loss": 2.0126,
|
|
"step": 24
|
|
},
|
|
{
|
|
"epoch": 0.08,
|
|
"learning_rate": 9.315960912052117e-06,
|
|
"loss": 1.7382,
|
|
"step": 25
|
|
},
|
|
{
|
|
"epoch": 0.08,
|
|
"learning_rate": 9.283387622149838e-06,
|
|
"loss": 1.522,
|
|
"step": 26
|
|
},
|
|
{
|
|
"epoch": 0.09,
|
|
"learning_rate": 9.250814332247557e-06,
|
|
"loss": 1.8094,
|
|
"step": 27
|
|
},
|
|
{
|
|
"epoch": 0.09,
|
|
"learning_rate": 9.218241042345277e-06,
|
|
"loss": 1.5582,
|
|
"step": 28
|
|
},
|
|
{
|
|
"epoch": 0.09,
|
|
"learning_rate": 9.185667752442996e-06,
|
|
"loss": 1.9294,
|
|
"step": 29
|
|
},
|
|
{
|
|
"epoch": 0.1,
|
|
"learning_rate": 9.153094462540717e-06,
|
|
"loss": 1.583,
|
|
"step": 30
|
|
},
|
|
{
|
|
"epoch": 0.1,
|
|
"learning_rate": 9.120521172638436e-06,
|
|
"loss": 1.7552,
|
|
"step": 31
|
|
},
|
|
{
|
|
"epoch": 0.1,
|
|
"learning_rate": 9.087947882736157e-06,
|
|
"loss": 1.9126,
|
|
"step": 32
|
|
},
|
|
{
|
|
"epoch": 0.11,
|
|
"learning_rate": 9.055374592833876e-06,
|
|
"loss": 1.671,
|
|
"step": 33
|
|
},
|
|
{
|
|
"epoch": 0.11,
|
|
"learning_rate": 9.022801302931597e-06,
|
|
"loss": 1.8853,
|
|
"step": 34
|
|
},
|
|
{
|
|
"epoch": 0.11,
|
|
"learning_rate": 8.990228013029316e-06,
|
|
"loss": 1.6868,
|
|
"step": 35
|
|
},
|
|
{
|
|
"epoch": 0.12,
|
|
"learning_rate": 8.957654723127036e-06,
|
|
"loss": 1.5296,
|
|
"step": 36
|
|
},
|
|
{
|
|
"epoch": 0.12,
|
|
"learning_rate": 8.925081433224755e-06,
|
|
"loss": 1.5385,
|
|
"step": 37
|
|
},
|
|
{
|
|
"epoch": 0.12,
|
|
"learning_rate": 8.892508143322476e-06,
|
|
"loss": 1.6824,
|
|
"step": 38
|
|
},
|
|
{
|
|
"epoch": 0.13,
|
|
"learning_rate": 8.859934853420195e-06,
|
|
"loss": 1.6062,
|
|
"step": 39
|
|
},
|
|
{
|
|
"epoch": 0.13,
|
|
"learning_rate": 8.827361563517916e-06,
|
|
"loss": 1.7446,
|
|
"step": 40
|
|
},
|
|
{
|
|
"epoch": 0.13,
|
|
"learning_rate": 8.794788273615635e-06,
|
|
"loss": 1.6543,
|
|
"step": 41
|
|
},
|
|
{
|
|
"epoch": 0.14,
|
|
"learning_rate": 8.762214983713355e-06,
|
|
"loss": 1.6146,
|
|
"step": 42
|
|
},
|
|
{
|
|
"epoch": 0.14,
|
|
"learning_rate": 8.729641693811076e-06,
|
|
"loss": 1.8047,
|
|
"step": 43
|
|
},
|
|
{
|
|
"epoch": 0.14,
|
|
"learning_rate": 8.697068403908797e-06,
|
|
"loss": 1.6367,
|
|
"step": 44
|
|
},
|
|
{
|
|
"epoch": 0.14,
|
|
"learning_rate": 8.664495114006516e-06,
|
|
"loss": 1.6845,
|
|
"step": 45
|
|
},
|
|
{
|
|
"epoch": 0.15,
|
|
"learning_rate": 8.631921824104237e-06,
|
|
"loss": 1.6436,
|
|
"step": 46
|
|
},
|
|
{
|
|
"epoch": 0.15,
|
|
"learning_rate": 8.599348534201956e-06,
|
|
"loss": 1.8192,
|
|
"step": 47
|
|
},
|
|
{
|
|
"epoch": 0.15,
|
|
"learning_rate": 8.566775244299675e-06,
|
|
"loss": 1.6621,
|
|
"step": 48
|
|
},
|
|
{
|
|
"epoch": 0.16,
|
|
"learning_rate": 8.534201954397395e-06,
|
|
"loss": 1.6001,
|
|
"step": 49
|
|
},
|
|
{
|
|
"epoch": 0.16,
|
|
"learning_rate": 8.501628664495114e-06,
|
|
"loss": 1.5685,
|
|
"step": 50
|
|
},
|
|
{
|
|
"epoch": 0.16,
|
|
"learning_rate": 8.469055374592835e-06,
|
|
"loss": 1.9874,
|
|
"step": 51
|
|
},
|
|
{
|
|
"epoch": 0.17,
|
|
"learning_rate": 8.436482084690554e-06,
|
|
"loss": 1.6139,
|
|
"step": 52
|
|
},
|
|
{
|
|
"epoch": 0.17,
|
|
"learning_rate": 8.403908794788275e-06,
|
|
"loss": 1.6178,
|
|
"step": 53
|
|
},
|
|
{
|
|
"epoch": 0.17,
|
|
"learning_rate": 8.371335504885994e-06,
|
|
"loss": 1.4743,
|
|
"step": 54
|
|
},
|
|
{
|
|
"epoch": 0.18,
|
|
"learning_rate": 8.338762214983714e-06,
|
|
"loss": 1.3837,
|
|
"step": 55
|
|
},
|
|
{
|
|
"epoch": 0.18,
|
|
"learning_rate": 8.306188925081433e-06,
|
|
"loss": 1.5702,
|
|
"step": 56
|
|
},
|
|
{
|
|
"epoch": 0.18,
|
|
"learning_rate": 8.273615635179154e-06,
|
|
"loss": 1.6048,
|
|
"step": 57
|
|
},
|
|
{
|
|
"epoch": 0.19,
|
|
"learning_rate": 8.241042345276873e-06,
|
|
"loss": 1.6952,
|
|
"step": 58
|
|
},
|
|
{
|
|
"epoch": 0.19,
|
|
"learning_rate": 8.208469055374594e-06,
|
|
"loss": 1.8444,
|
|
"step": 59
|
|
},
|
|
{
|
|
"epoch": 0.19,
|
|
"learning_rate": 8.175895765472313e-06,
|
|
"loss": 1.7213,
|
|
"step": 60
|
|
},
|
|
{
|
|
"epoch": 0.2,
|
|
"learning_rate": 8.143322475570034e-06,
|
|
"loss": 1.8617,
|
|
"step": 61
|
|
},
|
|
{
|
|
"epoch": 0.2,
|
|
"learning_rate": 8.110749185667753e-06,
|
|
"loss": 1.8121,
|
|
"step": 62
|
|
},
|
|
{
|
|
"epoch": 0.2,
|
|
"learning_rate": 8.078175895765473e-06,
|
|
"loss": 1.7389,
|
|
"step": 63
|
|
},
|
|
{
|
|
"epoch": 0.21,
|
|
"learning_rate": 8.045602605863192e-06,
|
|
"loss": 1.9176,
|
|
"step": 64
|
|
},
|
|
{
|
|
"epoch": 0.21,
|
|
"learning_rate": 8.013029315960913e-06,
|
|
"loss": 1.5571,
|
|
"step": 65
|
|
},
|
|
{
|
|
"epoch": 0.21,
|
|
"learning_rate": 7.980456026058632e-06,
|
|
"loss": 1.6006,
|
|
"step": 66
|
|
},
|
|
{
|
|
"epoch": 0.22,
|
|
"learning_rate": 7.947882736156353e-06,
|
|
"loss": 1.6972,
|
|
"step": 67
|
|
},
|
|
{
|
|
"epoch": 0.22,
|
|
"learning_rate": 7.915309446254072e-06,
|
|
"loss": 1.7133,
|
|
"step": 68
|
|
},
|
|
{
|
|
"epoch": 0.22,
|
|
"learning_rate": 7.882736156351792e-06,
|
|
"loss": 1.8061,
|
|
"step": 69
|
|
},
|
|
{
|
|
"epoch": 0.23,
|
|
"learning_rate": 7.850162866449511e-06,
|
|
"loss": 1.7225,
|
|
"step": 70
|
|
},
|
|
{
|
|
"epoch": 0.23,
|
|
"learning_rate": 7.817589576547232e-06,
|
|
"loss": 1.473,
|
|
"step": 71
|
|
},
|
|
{
|
|
"epoch": 0.23,
|
|
"learning_rate": 7.785016286644951e-06,
|
|
"loss": 1.492,
|
|
"step": 72
|
|
},
|
|
{
|
|
"epoch": 0.23,
|
|
"learning_rate": 7.752442996742672e-06,
|
|
"loss": 1.8478,
|
|
"step": 73
|
|
},
|
|
{
|
|
"epoch": 0.24,
|
|
"learning_rate": 7.719869706840391e-06,
|
|
"loss": 1.5614,
|
|
"step": 74
|
|
},
|
|
{
|
|
"epoch": 0.24,
|
|
"learning_rate": 7.687296416938112e-06,
|
|
"loss": 1.7853,
|
|
"step": 75
|
|
},
|
|
{
|
|
"epoch": 0.24,
|
|
"learning_rate": 7.65472312703583e-06,
|
|
"loss": 1.723,
|
|
"step": 76
|
|
},
|
|
{
|
|
"epoch": 0.25,
|
|
"learning_rate": 7.622149837133551e-06,
|
|
"loss": 1.4973,
|
|
"step": 77
|
|
},
|
|
{
|
|
"epoch": 0.25,
|
|
"learning_rate": 7.589576547231271e-06,
|
|
"loss": 1.8141,
|
|
"step": 78
|
|
},
|
|
{
|
|
"epoch": 0.25,
|
|
"learning_rate": 7.557003257328991e-06,
|
|
"loss": 1.843,
|
|
"step": 79
|
|
},
|
|
{
|
|
"epoch": 0.26,
|
|
"learning_rate": 7.524429967426711e-06,
|
|
"loss": 1.5906,
|
|
"step": 80
|
|
},
|
|
{
|
|
"epoch": 0.26,
|
|
"learning_rate": 7.491856677524431e-06,
|
|
"loss": 1.8201,
|
|
"step": 81
|
|
},
|
|
{
|
|
"epoch": 0.26,
|
|
"learning_rate": 7.459283387622151e-06,
|
|
"loss": 1.8767,
|
|
"step": 82
|
|
},
|
|
{
|
|
"epoch": 0.27,
|
|
"learning_rate": 7.4267100977198705e-06,
|
|
"loss": 1.6151,
|
|
"step": 83
|
|
},
|
|
{
|
|
"epoch": 0.27,
|
|
"learning_rate": 7.39413680781759e-06,
|
|
"loss": 1.5091,
|
|
"step": 84
|
|
},
|
|
{
|
|
"epoch": 0.27,
|
|
"learning_rate": 7.36156351791531e-06,
|
|
"loss": 1.7801,
|
|
"step": 85
|
|
},
|
|
{
|
|
"epoch": 0.28,
|
|
"learning_rate": 7.32899022801303e-06,
|
|
"loss": 1.7658,
|
|
"step": 86
|
|
},
|
|
{
|
|
"epoch": 0.28,
|
|
"learning_rate": 7.29641693811075e-06,
|
|
"loss": 1.7482,
|
|
"step": 87
|
|
},
|
|
{
|
|
"epoch": 0.28,
|
|
"learning_rate": 7.26384364820847e-06,
|
|
"loss": 1.8217,
|
|
"step": 88
|
|
},
|
|
{
|
|
"epoch": 0.29,
|
|
"learning_rate": 7.23127035830619e-06,
|
|
"loss": 1.7081,
|
|
"step": 89
|
|
},
|
|
{
|
|
"epoch": 0.29,
|
|
"learning_rate": 7.1986970684039095e-06,
|
|
"loss": 1.7887,
|
|
"step": 90
|
|
},
|
|
{
|
|
"epoch": 0.29,
|
|
"learning_rate": 7.166123778501629e-06,
|
|
"loss": 1.6873,
|
|
"step": 91
|
|
},
|
|
{
|
|
"epoch": 0.3,
|
|
"learning_rate": 7.133550488599349e-06,
|
|
"loss": 1.5507,
|
|
"step": 92
|
|
},
|
|
{
|
|
"epoch": 0.3,
|
|
"learning_rate": 7.100977198697069e-06,
|
|
"loss": 1.5357,
|
|
"step": 93
|
|
},
|
|
{
|
|
"epoch": 0.3,
|
|
"learning_rate": 7.068403908794789e-06,
|
|
"loss": 1.5712,
|
|
"step": 94
|
|
},
|
|
{
|
|
"epoch": 0.31,
|
|
"learning_rate": 7.035830618892509e-06,
|
|
"loss": 1.7986,
|
|
"step": 95
|
|
},
|
|
{
|
|
"epoch": 0.31,
|
|
"learning_rate": 7.003257328990229e-06,
|
|
"loss": 1.5605,
|
|
"step": 96
|
|
},
|
|
{
|
|
"epoch": 0.31,
|
|
"learning_rate": 6.9706840390879485e-06,
|
|
"loss": 1.7968,
|
|
"step": 97
|
|
},
|
|
{
|
|
"epoch": 0.32,
|
|
"learning_rate": 6.938110749185668e-06,
|
|
"loss": 1.6536,
|
|
"step": 98
|
|
},
|
|
{
|
|
"epoch": 0.32,
|
|
"learning_rate": 6.905537459283388e-06,
|
|
"loss": 1.6511,
|
|
"step": 99
|
|
},
|
|
{
|
|
"epoch": 0.32,
|
|
"learning_rate": 6.872964169381108e-06,
|
|
"loss": 1.874,
|
|
"step": 100
|
|
},
|
|
{
|
|
"epoch": 0.33,
|
|
"learning_rate": 6.840390879478828e-06,
|
|
"loss": 1.8521,
|
|
"step": 101
|
|
},
|
|
{
|
|
"epoch": 0.33,
|
|
"learning_rate": 6.807817589576548e-06,
|
|
"loss": 1.8295,
|
|
"step": 102
|
|
},
|
|
{
|
|
"epoch": 0.33,
|
|
"learning_rate": 6.775244299674268e-06,
|
|
"loss": 1.5801,
|
|
"step": 103
|
|
},
|
|
{
|
|
"epoch": 0.33,
|
|
"learning_rate": 6.7426710097719875e-06,
|
|
"loss": 1.6233,
|
|
"step": 104
|
|
},
|
|
{
|
|
"epoch": 0.34,
|
|
"learning_rate": 6.710097719869707e-06,
|
|
"loss": 1.609,
|
|
"step": 105
|
|
},
|
|
{
|
|
"epoch": 0.34,
|
|
"learning_rate": 6.677524429967427e-06,
|
|
"loss": 1.7378,
|
|
"step": 106
|
|
},
|
|
{
|
|
"epoch": 0.34,
|
|
"learning_rate": 6.644951140065147e-06,
|
|
"loss": 1.6588,
|
|
"step": 107
|
|
},
|
|
{
|
|
"epoch": 0.35,
|
|
"learning_rate": 6.612377850162867e-06,
|
|
"loss": 1.5261,
|
|
"step": 108
|
|
},
|
|
{
|
|
"epoch": 0.35,
|
|
"learning_rate": 6.579804560260587e-06,
|
|
"loss": 1.991,
|
|
"step": 109
|
|
},
|
|
{
|
|
"epoch": 0.35,
|
|
"learning_rate": 6.547231270358307e-06,
|
|
"loss": 1.4179,
|
|
"step": 110
|
|
},
|
|
{
|
|
"epoch": 0.36,
|
|
"learning_rate": 6.5146579804560266e-06,
|
|
"loss": 1.6307,
|
|
"step": 111
|
|
},
|
|
{
|
|
"epoch": 0.36,
|
|
"learning_rate": 6.482084690553746e-06,
|
|
"loss": 1.4607,
|
|
"step": 112
|
|
},
|
|
{
|
|
"epoch": 0.36,
|
|
"learning_rate": 6.449511400651466e-06,
|
|
"loss": 1.6902,
|
|
"step": 113
|
|
},
|
|
{
|
|
"epoch": 0.37,
|
|
"learning_rate": 6.416938110749186e-06,
|
|
"loss": 1.7303,
|
|
"step": 114
|
|
},
|
|
{
|
|
"epoch": 0.37,
|
|
"learning_rate": 6.384364820846905e-06,
|
|
"loss": 1.7285,
|
|
"step": 115
|
|
},
|
|
{
|
|
"epoch": 0.37,
|
|
"learning_rate": 6.351791530944625e-06,
|
|
"loss": 1.673,
|
|
"step": 116
|
|
},
|
|
{
|
|
"epoch": 0.38,
|
|
"learning_rate": 6.319218241042345e-06,
|
|
"loss": 1.7658,
|
|
"step": 117
|
|
},
|
|
{
|
|
"epoch": 0.38,
|
|
"learning_rate": 6.286644951140065e-06,
|
|
"loss": 1.842,
|
|
"step": 118
|
|
},
|
|
{
|
|
"epoch": 0.38,
|
|
"learning_rate": 6.2540716612377846e-06,
|
|
"loss": 1.4013,
|
|
"step": 119
|
|
},
|
|
{
|
|
"epoch": 0.39,
|
|
"learning_rate": 6.221498371335506e-06,
|
|
"loss": 1.5498,
|
|
"step": 120
|
|
},
|
|
{
|
|
"epoch": 0.39,
|
|
"learning_rate": 6.188925081433226e-06,
|
|
"loss": 1.6219,
|
|
"step": 121
|
|
},
|
|
{
|
|
"epoch": 0.39,
|
|
"learning_rate": 6.156351791530946e-06,
|
|
"loss": 1.6188,
|
|
"step": 122
|
|
},
|
|
{
|
|
"epoch": 0.4,
|
|
"learning_rate": 6.123778501628666e-06,
|
|
"loss": 1.7275,
|
|
"step": 123
|
|
},
|
|
{
|
|
"epoch": 0.4,
|
|
"learning_rate": 6.0912052117263856e-06,
|
|
"loss": 1.677,
|
|
"step": 124
|
|
},
|
|
{
|
|
"epoch": 0.4,
|
|
"learning_rate": 6.058631921824105e-06,
|
|
"loss": 1.8475,
|
|
"step": 125
|
|
},
|
|
{
|
|
"epoch": 0.41,
|
|
"learning_rate": 6.026058631921825e-06,
|
|
"loss": 1.6474,
|
|
"step": 126
|
|
},
|
|
{
|
|
"epoch": 0.41,
|
|
"learning_rate": 5.993485342019545e-06,
|
|
"loss": 1.6625,
|
|
"step": 127
|
|
},
|
|
{
|
|
"epoch": 0.41,
|
|
"learning_rate": 5.960912052117265e-06,
|
|
"loss": 1.4661,
|
|
"step": 128
|
|
},
|
|
{
|
|
"epoch": 0.42,
|
|
"learning_rate": 5.928338762214985e-06,
|
|
"loss": 1.3887,
|
|
"step": 129
|
|
},
|
|
{
|
|
"epoch": 0.42,
|
|
"learning_rate": 5.895765472312705e-06,
|
|
"loss": 1.7437,
|
|
"step": 130
|
|
},
|
|
{
|
|
"epoch": 0.42,
|
|
"learning_rate": 5.8631921824104246e-06,
|
|
"loss": 1.688,
|
|
"step": 131
|
|
},
|
|
{
|
|
"epoch": 0.42,
|
|
"learning_rate": 5.8306188925081444e-06,
|
|
"loss": 1.7416,
|
|
"step": 132
|
|
},
|
|
{
|
|
"epoch": 0.43,
|
|
"learning_rate": 5.798045602605864e-06,
|
|
"loss": 1.6838,
|
|
"step": 133
|
|
},
|
|
{
|
|
"epoch": 0.43,
|
|
"learning_rate": 5.765472312703584e-06,
|
|
"loss": 1.737,
|
|
"step": 134
|
|
},
|
|
{
|
|
"epoch": 0.43,
|
|
"learning_rate": 5.732899022801304e-06,
|
|
"loss": 1.5068,
|
|
"step": 135
|
|
},
|
|
{
|
|
"epoch": 0.44,
|
|
"learning_rate": 5.700325732899024e-06,
|
|
"loss": 1.6162,
|
|
"step": 136
|
|
},
|
|
{
|
|
"epoch": 0.44,
|
|
"learning_rate": 5.667752442996744e-06,
|
|
"loss": 1.6951,
|
|
"step": 137
|
|
},
|
|
{
|
|
"epoch": 0.44,
|
|
"learning_rate": 5.635179153094464e-06,
|
|
"loss": 1.6945,
|
|
"step": 138
|
|
},
|
|
{
|
|
"epoch": 0.45,
|
|
"learning_rate": 5.6026058631921834e-06,
|
|
"loss": 1.6939,
|
|
"step": 139
|
|
},
|
|
{
|
|
"epoch": 0.45,
|
|
"learning_rate": 5.570032573289903e-06,
|
|
"loss": 1.6487,
|
|
"step": 140
|
|
},
|
|
{
|
|
"epoch": 0.45,
|
|
"learning_rate": 5.537459283387623e-06,
|
|
"loss": 1.6105,
|
|
"step": 141
|
|
},
|
|
{
|
|
"epoch": 0.46,
|
|
"learning_rate": 5.504885993485343e-06,
|
|
"loss": 1.7248,
|
|
"step": 142
|
|
},
|
|
{
|
|
"epoch": 0.46,
|
|
"learning_rate": 5.472312703583063e-06,
|
|
"loss": 1.6465,
|
|
"step": 143
|
|
},
|
|
{
|
|
"epoch": 0.46,
|
|
"learning_rate": 5.439739413680783e-06,
|
|
"loss": 1.7724,
|
|
"step": 144
|
|
},
|
|
{
|
|
"epoch": 0.47,
|
|
"learning_rate": 5.407166123778503e-06,
|
|
"loss": 1.7574,
|
|
"step": 145
|
|
},
|
|
{
|
|
"epoch": 0.47,
|
|
"learning_rate": 5.3745928338762225e-06,
|
|
"loss": 1.4914,
|
|
"step": 146
|
|
},
|
|
{
|
|
"epoch": 0.47,
|
|
"learning_rate": 5.342019543973942e-06,
|
|
"loss": 1.7353,
|
|
"step": 147
|
|
},
|
|
{
|
|
"epoch": 0.48,
|
|
"learning_rate": 5.309446254071662e-06,
|
|
"loss": 1.4765,
|
|
"step": 148
|
|
},
|
|
{
|
|
"epoch": 0.48,
|
|
"learning_rate": 5.276872964169382e-06,
|
|
"loss": 1.9143,
|
|
"step": 149
|
|
},
|
|
{
|
|
"epoch": 0.48,
|
|
"learning_rate": 5.244299674267102e-06,
|
|
"loss": 1.4074,
|
|
"step": 150
|
|
},
|
|
{
|
|
"epoch": 0.49,
|
|
"learning_rate": 5.211726384364822e-06,
|
|
"loss": 1.4897,
|
|
"step": 151
|
|
},
|
|
{
|
|
"epoch": 0.49,
|
|
"learning_rate": 5.179153094462542e-06,
|
|
"loss": 1.4912,
|
|
"step": 152
|
|
},
|
|
{
|
|
"epoch": 0.49,
|
|
"learning_rate": 5.1465798045602615e-06,
|
|
"loss": 1.6008,
|
|
"step": 153
|
|
},
|
|
{
|
|
"epoch": 0.5,
|
|
"learning_rate": 5.114006514657981e-06,
|
|
"loss": 1.6002,
|
|
"step": 154
|
|
},
|
|
{
|
|
"epoch": 0.5,
|
|
"learning_rate": 5.081433224755701e-06,
|
|
"loss": 1.2512,
|
|
"step": 155
|
|
},
|
|
{
|
|
"epoch": 0.5,
|
|
"learning_rate": 5.048859934853421e-06,
|
|
"loss": 1.5504,
|
|
"step": 156
|
|
},
|
|
{
|
|
"epoch": 0.51,
|
|
"learning_rate": 5.016286644951141e-06,
|
|
"loss": 1.4701,
|
|
"step": 157
|
|
},
|
|
{
|
|
"epoch": 0.51,
|
|
"learning_rate": 4.98371335504886e-06,
|
|
"loss": 1.3369,
|
|
"step": 158
|
|
},
|
|
{
|
|
"epoch": 0.51,
|
|
"learning_rate": 4.95114006514658e-06,
|
|
"loss": 1.7382,
|
|
"step": 159
|
|
},
|
|
{
|
|
"epoch": 0.51,
|
|
"learning_rate": 4.9185667752443e-06,
|
|
"loss": 1.6354,
|
|
"step": 160
|
|
},
|
|
{
|
|
"epoch": 0.52,
|
|
"learning_rate": 4.8859934853420195e-06,
|
|
"loss": 1.5768,
|
|
"step": 161
|
|
},
|
|
{
|
|
"epoch": 0.52,
|
|
"learning_rate": 4.853420195439739e-06,
|
|
"loss": 1.7945,
|
|
"step": 162
|
|
},
|
|
{
|
|
"epoch": 0.52,
|
|
"learning_rate": 4.820846905537459e-06,
|
|
"loss": 1.4955,
|
|
"step": 163
|
|
},
|
|
{
|
|
"epoch": 0.53,
|
|
"learning_rate": 4.788273615635179e-06,
|
|
"loss": 1.6107,
|
|
"step": 164
|
|
},
|
|
{
|
|
"epoch": 0.53,
|
|
"learning_rate": 4.755700325732899e-06,
|
|
"loss": 1.4584,
|
|
"step": 165
|
|
},
|
|
{
|
|
"epoch": 0.53,
|
|
"learning_rate": 4.723127035830619e-06,
|
|
"loss": 1.7635,
|
|
"step": 166
|
|
},
|
|
{
|
|
"epoch": 0.54,
|
|
"learning_rate": 4.690553745928339e-06,
|
|
"loss": 1.8326,
|
|
"step": 167
|
|
},
|
|
{
|
|
"epoch": 0.54,
|
|
"learning_rate": 4.6579804560260585e-06,
|
|
"loss": 1.5407,
|
|
"step": 168
|
|
},
|
|
{
|
|
"epoch": 0.54,
|
|
"learning_rate": 4.625407166123778e-06,
|
|
"loss": 1.3624,
|
|
"step": 169
|
|
},
|
|
{
|
|
"epoch": 0.55,
|
|
"learning_rate": 4.592833876221498e-06,
|
|
"loss": 1.4966,
|
|
"step": 170
|
|
},
|
|
{
|
|
"epoch": 0.55,
|
|
"learning_rate": 4.560260586319218e-06,
|
|
"loss": 1.3342,
|
|
"step": 171
|
|
},
|
|
{
|
|
"epoch": 0.55,
|
|
"learning_rate": 4.527687296416938e-06,
|
|
"loss": 1.7541,
|
|
"step": 172
|
|
},
|
|
{
|
|
"epoch": 0.56,
|
|
"learning_rate": 4.495114006514658e-06,
|
|
"loss": 1.6425,
|
|
"step": 173
|
|
},
|
|
{
|
|
"epoch": 0.56,
|
|
"learning_rate": 4.462540716612378e-06,
|
|
"loss": 1.525,
|
|
"step": 174
|
|
},
|
|
{
|
|
"epoch": 0.56,
|
|
"learning_rate": 4.4299674267100975e-06,
|
|
"loss": 1.7863,
|
|
"step": 175
|
|
},
|
|
{
|
|
"epoch": 0.57,
|
|
"learning_rate": 4.397394136807817e-06,
|
|
"loss": 1.7503,
|
|
"step": 176
|
|
},
|
|
{
|
|
"epoch": 0.57,
|
|
"learning_rate": 4.364820846905538e-06,
|
|
"loss": 1.6258,
|
|
"step": 177
|
|
},
|
|
{
|
|
"epoch": 0.57,
|
|
"learning_rate": 4.332247557003258e-06,
|
|
"loss": 1.6659,
|
|
"step": 178
|
|
},
|
|
{
|
|
"epoch": 0.58,
|
|
"learning_rate": 4.299674267100978e-06,
|
|
"loss": 1.6783,
|
|
"step": 179
|
|
},
|
|
{
|
|
"epoch": 0.58,
|
|
"learning_rate": 4.267100977198698e-06,
|
|
"loss": 1.5565,
|
|
"step": 180
|
|
},
|
|
{
|
|
"epoch": 0.58,
|
|
"learning_rate": 4.2345276872964175e-06,
|
|
"loss": 1.5389,
|
|
"step": 181
|
|
},
|
|
{
|
|
"epoch": 0.59,
|
|
"learning_rate": 4.201954397394137e-06,
|
|
"loss": 1.6566,
|
|
"step": 182
|
|
},
|
|
{
|
|
"epoch": 0.59,
|
|
"learning_rate": 4.169381107491857e-06,
|
|
"loss": 1.5555,
|
|
"step": 183
|
|
},
|
|
{
|
|
"epoch": 0.59,
|
|
"learning_rate": 4.136807817589577e-06,
|
|
"loss": 1.5426,
|
|
"step": 184
|
|
},
|
|
{
|
|
"epoch": 0.6,
|
|
"learning_rate": 4.104234527687297e-06,
|
|
"loss": 1.6025,
|
|
"step": 185
|
|
},
|
|
{
|
|
"epoch": 0.6,
|
|
"learning_rate": 4.071661237785017e-06,
|
|
"loss": 1.5559,
|
|
"step": 186
|
|
},
|
|
{
|
|
"epoch": 0.6,
|
|
"learning_rate": 4.039087947882737e-06,
|
|
"loss": 1.5398,
|
|
"step": 187
|
|
},
|
|
{
|
|
"epoch": 0.61,
|
|
"learning_rate": 4.0065146579804565e-06,
|
|
"loss": 1.7874,
|
|
"step": 188
|
|
},
|
|
{
|
|
"epoch": 0.61,
|
|
"learning_rate": 3.973941368078176e-06,
|
|
"loss": 1.6019,
|
|
"step": 189
|
|
},
|
|
{
|
|
"epoch": 0.61,
|
|
"learning_rate": 3.941368078175896e-06,
|
|
"loss": 1.5946,
|
|
"step": 190
|
|
},
|
|
{
|
|
"epoch": 0.61,
|
|
"learning_rate": 3.908794788273616e-06,
|
|
"loss": 1.4845,
|
|
"step": 191
|
|
},
|
|
{
|
|
"epoch": 0.62,
|
|
"learning_rate": 3.876221498371336e-06,
|
|
"loss": 1.6955,
|
|
"step": 192
|
|
},
|
|
{
|
|
"epoch": 0.62,
|
|
"learning_rate": 3.843648208469056e-06,
|
|
"loss": 1.661,
|
|
"step": 193
|
|
},
|
|
{
|
|
"epoch": 0.62,
|
|
"learning_rate": 3.8110749185667757e-06,
|
|
"loss": 1.5976,
|
|
"step": 194
|
|
},
|
|
{
|
|
"epoch": 0.63,
|
|
"learning_rate": 3.7785016286644955e-06,
|
|
"loss": 1.2598,
|
|
"step": 195
|
|
},
|
|
{
|
|
"epoch": 0.63,
|
|
"learning_rate": 3.7459283387622154e-06,
|
|
"loss": 1.785,
|
|
"step": 196
|
|
},
|
|
{
|
|
"epoch": 0.63,
|
|
"learning_rate": 3.7133550488599353e-06,
|
|
"loss": 1.5645,
|
|
"step": 197
|
|
},
|
|
{
|
|
"epoch": 0.64,
|
|
"learning_rate": 3.680781758957655e-06,
|
|
"loss": 1.455,
|
|
"step": 198
|
|
},
|
|
{
|
|
"epoch": 0.64,
|
|
"learning_rate": 3.648208469055375e-06,
|
|
"loss": 1.7124,
|
|
"step": 199
|
|
},
|
|
{
|
|
"epoch": 0.64,
|
|
"learning_rate": 3.615635179153095e-06,
|
|
"loss": 1.8653,
|
|
"step": 200
|
|
},
|
|
{
|
|
"epoch": 0.65,
|
|
"learning_rate": 3.5830618892508147e-06,
|
|
"loss": 1.7524,
|
|
"step": 201
|
|
},
|
|
{
|
|
"epoch": 0.65,
|
|
"learning_rate": 3.5504885993485345e-06,
|
|
"loss": 1.5227,
|
|
"step": 202
|
|
},
|
|
{
|
|
"epoch": 0.65,
|
|
"learning_rate": 3.5179153094462544e-06,
|
|
"loss": 1.6606,
|
|
"step": 203
|
|
},
|
|
{
|
|
"epoch": 0.66,
|
|
"learning_rate": 3.4853420195439743e-06,
|
|
"loss": 1.6551,
|
|
"step": 204
|
|
},
|
|
{
|
|
"epoch": 0.66,
|
|
"learning_rate": 3.452768729641694e-06,
|
|
"loss": 1.8254,
|
|
"step": 205
|
|
},
|
|
{
|
|
"epoch": 0.66,
|
|
"learning_rate": 3.420195439739414e-06,
|
|
"loss": 1.6391,
|
|
"step": 206
|
|
},
|
|
{
|
|
"epoch": 0.67,
|
|
"learning_rate": 3.387622149837134e-06,
|
|
"loss": 1.7302,
|
|
"step": 207
|
|
},
|
|
{
|
|
"epoch": 0.67,
|
|
"learning_rate": 3.3550488599348537e-06,
|
|
"loss": 1.6899,
|
|
"step": 208
|
|
},
|
|
{
|
|
"epoch": 0.67,
|
|
"learning_rate": 3.3224755700325736e-06,
|
|
"loss": 1.6084,
|
|
"step": 209
|
|
},
|
|
{
|
|
"epoch": 0.68,
|
|
"learning_rate": 3.2899022801302934e-06,
|
|
"loss": 1.7026,
|
|
"step": 210
|
|
},
|
|
{
|
|
"epoch": 0.68,
|
|
"learning_rate": 3.2573289902280133e-06,
|
|
"loss": 1.7044,
|
|
"step": 211
|
|
},
|
|
{
|
|
"epoch": 0.68,
|
|
"learning_rate": 3.224755700325733e-06,
|
|
"loss": 1.2729,
|
|
"step": 212
|
|
},
|
|
{
|
|
"epoch": 0.69,
|
|
"learning_rate": 3.1921824104234526e-06,
|
|
"loss": 1.8227,
|
|
"step": 213
|
|
},
|
|
{
|
|
"epoch": 0.69,
|
|
"learning_rate": 3.1596091205211724e-06,
|
|
"loss": 1.6096,
|
|
"step": 214
|
|
},
|
|
{
|
|
"epoch": 0.69,
|
|
"learning_rate": 3.1270358306188923e-06,
|
|
"loss": 1.8309,
|
|
"step": 215
|
|
},
|
|
{
|
|
"epoch": 0.7,
|
|
"learning_rate": 3.094462540716613e-06,
|
|
"loss": 1.4867,
|
|
"step": 216
|
|
},
|
|
{
|
|
"epoch": 0.7,
|
|
"learning_rate": 3.061889250814333e-06,
|
|
"loss": 1.243,
|
|
"step": 217
|
|
},
|
|
{
|
|
"epoch": 0.7,
|
|
"learning_rate": 3.0293159609120527e-06,
|
|
"loss": 1.723,
|
|
"step": 218
|
|
},
|
|
{
|
|
"epoch": 0.7,
|
|
"learning_rate": 2.9967426710097726e-06,
|
|
"loss": 1.7381,
|
|
"step": 219
|
|
},
|
|
{
|
|
"epoch": 0.71,
|
|
"learning_rate": 2.9641693811074924e-06,
|
|
"loss": 1.7258,
|
|
"step": 220
|
|
},
|
|
{
|
|
"epoch": 0.71,
|
|
"learning_rate": 2.9315960912052123e-06,
|
|
"loss": 1.7599,
|
|
"step": 221
|
|
},
|
|
{
|
|
"epoch": 0.71,
|
|
"learning_rate": 2.899022801302932e-06,
|
|
"loss": 1.5384,
|
|
"step": 222
|
|
},
|
|
{
|
|
"epoch": 0.72,
|
|
"learning_rate": 2.866449511400652e-06,
|
|
"loss": 1.5919,
|
|
"step": 223
|
|
},
|
|
{
|
|
"epoch": 0.72,
|
|
"learning_rate": 2.833876221498372e-06,
|
|
"loss": 1.6868,
|
|
"step": 224
|
|
},
|
|
{
|
|
"epoch": 0.72,
|
|
"learning_rate": 2.8013029315960917e-06,
|
|
"loss": 1.7313,
|
|
"step": 225
|
|
},
|
|
{
|
|
"epoch": 0.73,
|
|
"learning_rate": 2.7687296416938116e-06,
|
|
"loss": 1.6369,
|
|
"step": 226
|
|
},
|
|
{
|
|
"epoch": 0.73,
|
|
"learning_rate": 2.7361563517915314e-06,
|
|
"loss": 1.7747,
|
|
"step": 227
|
|
},
|
|
{
|
|
"epoch": 0.73,
|
|
"learning_rate": 2.7035830618892513e-06,
|
|
"loss": 1.7925,
|
|
"step": 228
|
|
},
|
|
{
|
|
"epoch": 0.74,
|
|
"learning_rate": 2.671009771986971e-06,
|
|
"loss": 1.5667,
|
|
"step": 229
|
|
},
|
|
{
|
|
"epoch": 0.74,
|
|
"learning_rate": 2.638436482084691e-06,
|
|
"loss": 1.5783,
|
|
"step": 230
|
|
},
|
|
{
|
|
"epoch": 0.74,
|
|
"learning_rate": 2.605863192182411e-06,
|
|
"loss": 1.9011,
|
|
"step": 231
|
|
},
|
|
{
|
|
"epoch": 0.75,
|
|
"learning_rate": 2.5732899022801307e-06,
|
|
"loss": 1.9104,
|
|
"step": 232
|
|
},
|
|
{
|
|
"epoch": 0.75,
|
|
"learning_rate": 2.5407166123778506e-06,
|
|
"loss": 1.6284,
|
|
"step": 233
|
|
},
|
|
{
|
|
"epoch": 0.75,
|
|
"learning_rate": 2.5081433224755705e-06,
|
|
"loss": 1.6858,
|
|
"step": 234
|
|
},
|
|
{
|
|
"epoch": 0.76,
|
|
"learning_rate": 2.47557003257329e-06,
|
|
"loss": 1.4187,
|
|
"step": 235
|
|
},
|
|
{
|
|
"epoch": 0.76,
|
|
"learning_rate": 2.4429967426710097e-06,
|
|
"loss": 1.5768,
|
|
"step": 236
|
|
},
|
|
{
|
|
"epoch": 0.76,
|
|
"learning_rate": 2.4104234527687296e-06,
|
|
"loss": 1.7045,
|
|
"step": 237
|
|
},
|
|
{
|
|
"epoch": 0.77,
|
|
"learning_rate": 2.3778501628664495e-06,
|
|
"loss": 1.5737,
|
|
"step": 238
|
|
},
|
|
{
|
|
"epoch": 0.77,
|
|
"learning_rate": 2.3452768729641693e-06,
|
|
"loss": 1.7599,
|
|
"step": 239
|
|
},
|
|
{
|
|
"epoch": 0.77,
|
|
"learning_rate": 2.312703583061889e-06,
|
|
"loss": 1.7713,
|
|
"step": 240
|
|
},
|
|
{
|
|
"epoch": 0.78,
|
|
"learning_rate": 2.280130293159609e-06,
|
|
"loss": 1.6503,
|
|
"step": 241
|
|
},
|
|
{
|
|
"epoch": 0.78,
|
|
"learning_rate": 2.247557003257329e-06,
|
|
"loss": 1.5395,
|
|
"step": 242
|
|
},
|
|
{
|
|
"epoch": 0.78,
|
|
"learning_rate": 2.2149837133550488e-06,
|
|
"loss": 1.4427,
|
|
"step": 243
|
|
},
|
|
{
|
|
"epoch": 0.79,
|
|
"learning_rate": 2.182410423452769e-06,
|
|
"loss": 1.7187,
|
|
"step": 244
|
|
},
|
|
{
|
|
"epoch": 0.79,
|
|
"learning_rate": 2.149837133550489e-06,
|
|
"loss": 1.4899,
|
|
"step": 245
|
|
},
|
|
{
|
|
"epoch": 0.79,
|
|
"learning_rate": 2.1172638436482088e-06,
|
|
"loss": 1.7651,
|
|
"step": 246
|
|
},
|
|
{
|
|
"epoch": 0.8,
|
|
"learning_rate": 2.0846905537459286e-06,
|
|
"loss": 1.6961,
|
|
"step": 247
|
|
},
|
|
{
|
|
"epoch": 0.8,
|
|
"learning_rate": 2.0521172638436485e-06,
|
|
"loss": 1.4032,
|
|
"step": 248
|
|
},
|
|
{
|
|
"epoch": 0.8,
|
|
"learning_rate": 2.0195439739413683e-06,
|
|
"loss": 1.67,
|
|
"step": 249
|
|
},
|
|
{
|
|
"epoch": 0.8,
|
|
"learning_rate": 1.986970684039088e-06,
|
|
"loss": 1.8905,
|
|
"step": 250
|
|
},
|
|
{
|
|
"epoch": 0.81,
|
|
"learning_rate": 1.954397394136808e-06,
|
|
"loss": 1.5761,
|
|
"step": 251
|
|
},
|
|
{
|
|
"epoch": 0.81,
|
|
"learning_rate": 1.921824104234528e-06,
|
|
"loss": 1.8614,
|
|
"step": 252
|
|
},
|
|
{
|
|
"epoch": 0.81,
|
|
"learning_rate": 1.8892508143322478e-06,
|
|
"loss": 1.4054,
|
|
"step": 253
|
|
},
|
|
{
|
|
"epoch": 0.82,
|
|
"learning_rate": 1.8566775244299676e-06,
|
|
"loss": 1.4088,
|
|
"step": 254
|
|
},
|
|
{
|
|
"epoch": 0.82,
|
|
"learning_rate": 1.8241042345276875e-06,
|
|
"loss": 1.5462,
|
|
"step": 255
|
|
},
|
|
{
|
|
"epoch": 0.82,
|
|
"learning_rate": 1.7915309446254073e-06,
|
|
"loss": 1.896,
|
|
"step": 256
|
|
},
|
|
{
|
|
"epoch": 0.83,
|
|
"learning_rate": 1.7589576547231272e-06,
|
|
"loss": 1.7283,
|
|
"step": 257
|
|
},
|
|
{
|
|
"epoch": 0.83,
|
|
"learning_rate": 1.726384364820847e-06,
|
|
"loss": 1.5757,
|
|
"step": 258
|
|
},
|
|
{
|
|
"epoch": 0.83,
|
|
"learning_rate": 1.693811074918567e-06,
|
|
"loss": 1.5051,
|
|
"step": 259
|
|
},
|
|
{
|
|
"epoch": 0.84,
|
|
"learning_rate": 1.6612377850162868e-06,
|
|
"loss": 1.8005,
|
|
"step": 260
|
|
},
|
|
{
|
|
"epoch": 0.84,
|
|
"learning_rate": 1.6286644951140066e-06,
|
|
"loss": 1.802,
|
|
"step": 261
|
|
},
|
|
{
|
|
"epoch": 0.84,
|
|
"learning_rate": 1.5960912052117263e-06,
|
|
"loss": 1.6952,
|
|
"step": 262
|
|
},
|
|
{
|
|
"epoch": 0.85,
|
|
"learning_rate": 1.5635179153094461e-06,
|
|
"loss": 1.5592,
|
|
"step": 263
|
|
},
|
|
{
|
|
"epoch": 0.85,
|
|
"learning_rate": 1.5309446254071664e-06,
|
|
"loss": 1.6402,
|
|
"step": 264
|
|
},
|
|
{
|
|
"epoch": 0.85,
|
|
"learning_rate": 1.4983713355048863e-06,
|
|
"loss": 1.5711,
|
|
"step": 265
|
|
},
|
|
{
|
|
"epoch": 0.86,
|
|
"learning_rate": 1.4657980456026061e-06,
|
|
"loss": 1.5452,
|
|
"step": 266
|
|
},
|
|
{
|
|
"epoch": 0.86,
|
|
"learning_rate": 1.433224755700326e-06,
|
|
"loss": 1.7925,
|
|
"step": 267
|
|
},
|
|
{
|
|
"epoch": 0.86,
|
|
"learning_rate": 1.4006514657980459e-06,
|
|
"loss": 1.5405,
|
|
"step": 268
|
|
},
|
|
{
|
|
"epoch": 0.87,
|
|
"learning_rate": 1.3680781758957657e-06,
|
|
"loss": 1.6837,
|
|
"step": 269
|
|
},
|
|
{
|
|
"epoch": 0.87,
|
|
"learning_rate": 1.3355048859934856e-06,
|
|
"loss": 1.6495,
|
|
"step": 270
|
|
},
|
|
{
|
|
"epoch": 0.87,
|
|
"learning_rate": 1.3029315960912054e-06,
|
|
"loss": 1.4279,
|
|
"step": 271
|
|
},
|
|
{
|
|
"epoch": 0.88,
|
|
"learning_rate": 1.2703583061889253e-06,
|
|
"loss": 1.5919,
|
|
"step": 272
|
|
},
|
|
{
|
|
"epoch": 0.88,
|
|
"learning_rate": 1.237785016286645e-06,
|
|
"loss": 1.6427,
|
|
"step": 273
|
|
},
|
|
{
|
|
"epoch": 0.88,
|
|
"learning_rate": 1.2052117263843648e-06,
|
|
"loss": 1.7683,
|
|
"step": 274
|
|
},
|
|
{
|
|
"epoch": 0.89,
|
|
"learning_rate": 1.1726384364820847e-06,
|
|
"loss": 1.7068,
|
|
"step": 275
|
|
},
|
|
{
|
|
"epoch": 0.89,
|
|
"learning_rate": 1.1400651465798045e-06,
|
|
"loss": 1.7888,
|
|
"step": 276
|
|
},
|
|
{
|
|
"epoch": 0.89,
|
|
"learning_rate": 1.1074918566775244e-06,
|
|
"loss": 1.8027,
|
|
"step": 277
|
|
},
|
|
{
|
|
"epoch": 0.89,
|
|
"learning_rate": 1.0749185667752444e-06,
|
|
"loss": 1.5216,
|
|
"step": 278
|
|
},
|
|
{
|
|
"epoch": 0.9,
|
|
"learning_rate": 1.0423452768729643e-06,
|
|
"loss": 1.6276,
|
|
"step": 279
|
|
},
|
|
{
|
|
"epoch": 0.9,
|
|
"learning_rate": 1.0097719869706842e-06,
|
|
"loss": 1.3902,
|
|
"step": 280
|
|
},
|
|
{
|
|
"epoch": 0.9,
|
|
"learning_rate": 9.77198697068404e-07,
|
|
"loss": 1.9063,
|
|
"step": 281
|
|
},
|
|
{
|
|
"epoch": 0.91,
|
|
"learning_rate": 9.446254071661239e-07,
|
|
"loss": 1.6886,
|
|
"step": 282
|
|
},
|
|
{
|
|
"epoch": 0.91,
|
|
"learning_rate": 9.120521172638437e-07,
|
|
"loss": 1.6662,
|
|
"step": 283
|
|
},
|
|
{
|
|
"epoch": 0.91,
|
|
"learning_rate": 8.794788273615636e-07,
|
|
"loss": 1.6444,
|
|
"step": 284
|
|
},
|
|
{
|
|
"epoch": 0.92,
|
|
"learning_rate": 8.469055374592835e-07,
|
|
"loss": 1.4257,
|
|
"step": 285
|
|
},
|
|
{
|
|
"epoch": 0.92,
|
|
"learning_rate": 8.143322475570033e-07,
|
|
"loss": 1.5382,
|
|
"step": 286
|
|
},
|
|
{
|
|
"epoch": 0.92,
|
|
"learning_rate": 7.817589576547231e-07,
|
|
"loss": 1.5131,
|
|
"step": 287
|
|
},
|
|
{
|
|
"epoch": 0.93,
|
|
"learning_rate": 7.491856677524431e-07,
|
|
"loss": 1.7067,
|
|
"step": 288
|
|
},
|
|
{
|
|
"epoch": 0.93,
|
|
"learning_rate": 7.16612377850163e-07,
|
|
"loss": 1.7196,
|
|
"step": 289
|
|
},
|
|
{
|
|
"epoch": 0.93,
|
|
"learning_rate": 6.840390879478829e-07,
|
|
"loss": 1.7255,
|
|
"step": 290
|
|
},
|
|
{
|
|
"epoch": 0.94,
|
|
"learning_rate": 6.514657980456027e-07,
|
|
"loss": 1.7869,
|
|
"step": 291
|
|
},
|
|
{
|
|
"epoch": 0.94,
|
|
"learning_rate": 6.188925081433225e-07,
|
|
"loss": 1.5559,
|
|
"step": 292
|
|
},
|
|
{
|
|
"epoch": 0.94,
|
|
"learning_rate": 5.863192182410423e-07,
|
|
"loss": 1.734,
|
|
"step": 293
|
|
},
|
|
{
|
|
"epoch": 0.95,
|
|
"learning_rate": 5.537459283387622e-07,
|
|
"loss": 1.6327,
|
|
"step": 294
|
|
},
|
|
{
|
|
"epoch": 0.95,
|
|
"learning_rate": 5.211726384364822e-07,
|
|
"loss": 1.6533,
|
|
"step": 295
|
|
},
|
|
{
|
|
"epoch": 0.95,
|
|
"learning_rate": 4.88599348534202e-07,
|
|
"loss": 1.8027,
|
|
"step": 296
|
|
},
|
|
{
|
|
"epoch": 0.96,
|
|
"learning_rate": 4.5602605863192187e-07,
|
|
"loss": 1.653,
|
|
"step": 297
|
|
},
|
|
{
|
|
"epoch": 0.96,
|
|
"learning_rate": 4.2345276872964173e-07,
|
|
"loss": 1.8039,
|
|
"step": 298
|
|
},
|
|
{
|
|
"epoch": 0.96,
|
|
"learning_rate": 3.9087947882736154e-07,
|
|
"loss": 1.6957,
|
|
"step": 299
|
|
},
|
|
{
|
|
"epoch": 0.97,
|
|
"learning_rate": 3.583061889250815e-07,
|
|
"loss": 1.6976,
|
|
"step": 300
|
|
},
|
|
{
|
|
"epoch": 0.97,
|
|
"learning_rate": 3.2573289902280136e-07,
|
|
"loss": 1.572,
|
|
"step": 301
|
|
},
|
|
{
|
|
"epoch": 0.97,
|
|
"learning_rate": 2.9315960912052117e-07,
|
|
"loss": 1.6661,
|
|
"step": 302
|
|
},
|
|
{
|
|
"epoch": 0.98,
|
|
"learning_rate": 2.605863192182411e-07,
|
|
"loss": 1.8134,
|
|
"step": 303
|
|
},
|
|
{
|
|
"epoch": 0.98,
|
|
"learning_rate": 2.2801302931596094e-07,
|
|
"loss": 1.5449,
|
|
"step": 304
|
|
},
|
|
{
|
|
"epoch": 0.98,
|
|
"learning_rate": 1.9543973941368077e-07,
|
|
"loss": 1.7742,
|
|
"step": 305
|
|
},
|
|
{
|
|
"epoch": 0.98,
|
|
"learning_rate": 1.6286644951140068e-07,
|
|
"loss": 1.6709,
|
|
"step": 306
|
|
},
|
|
{
|
|
"epoch": 0.99,
|
|
"learning_rate": 1.3029315960912054e-07,
|
|
"loss": 1.6056,
|
|
"step": 307
|
|
},
|
|
{
|
|
"epoch": 0.99,
|
|
"learning_rate": 9.771986970684038e-08,
|
|
"loss": 1.6165,
|
|
"step": 308
|
|
},
|
|
{
|
|
"epoch": 0.99,
|
|
"learning_rate": 6.514657980456027e-08,
|
|
"loss": 1.3148,
|
|
"step": 309
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.2573289902280135e-08,
|
|
"loss": 1.569,
|
|
"step": 310
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"step": 310,
|
|
"total_flos": 6.305195734597632e+16,
|
|
"train_loss": 1.6624480166742879,
|
|
"train_runtime": 1167.1011,
|
|
"train_samples_per_second": 4.259,
|
|
"train_steps_per_second": 0.266
|
|
}
|
|
],
|
|
"logging_steps": 1.0,
|
|
"max_steps": 310,
|
|
"num_input_tokens_seen": 0,
|
|
"num_train_epochs": 1,
|
|
"save_steps": 500,
|
|
"total_flos": 6.305195734597632e+16,
|
|
"train_batch_size": 1,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|