| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.997014034040012, | |
| "global_step": 418, | |
| "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": 2.6719, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.404763088546395e-06, | |
| "loss": 2.6631, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 8.566346820627895e-06, | |
| "loss": 2.8389, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.080952617709279e-05, | |
| "loss": 2.6509, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.2549471261506067e-05, | |
| "loss": 2.6902, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.397110990917429e-05, | |
| "loss": 2.6899, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.5173088259185983e-05, | |
| "loss": 2.7358, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.6214289265639183e-05, | |
| "loss": 2.751, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.713269364125579e-05, | |
| "loss": 2.7358, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.7954234350052467e-05, | |
| "loss": 2.7861, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.8697408319761175e-05, | |
| "loss": 2.7234, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.9375872997720687e-05, | |
| "loss": 2.6865, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 2e-05, | |
| "loss": 2.6277, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 2e-05, | |
| "loss": 2.7432, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.995061728395062e-05, | |
| "loss": 2.6519, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.9901234567901237e-05, | |
| "loss": 2.729, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.9851851851851855e-05, | |
| "loss": 2.6909, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.9802469135802472e-05, | |
| "loss": 2.7485, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.9753086419753087e-05, | |
| "loss": 2.667, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.9703703703703704e-05, | |
| "loss": 2.6123, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.9654320987654322e-05, | |
| "loss": 2.647, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.960493827160494e-05, | |
| "loss": 2.6318, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.9555555555555557e-05, | |
| "loss": 2.6511, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.9506172839506175e-05, | |
| "loss": 2.6763, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.9456790123456793e-05, | |
| "loss": 2.6714, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.9407407407407407e-05, | |
| "loss": 2.6836, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.9358024691358025e-05, | |
| "loss": 2.6025, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.9308641975308646e-05, | |
| "loss": 2.6606, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.925925925925926e-05, | |
| "loss": 2.6948, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.9209876543209878e-05, | |
| "loss": 2.6279, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.9160493827160495e-05, | |
| "loss": 2.6401, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.9111111111111113e-05, | |
| "loss": 2.6643, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.906172839506173e-05, | |
| "loss": 2.6191, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.901234567901235e-05, | |
| "loss": 2.6152, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.8962962962962966e-05, | |
| "loss": 2.5776, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.891358024691358e-05, | |
| "loss": 2.5532, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.8864197530864198e-05, | |
| "loss": 2.5884, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.8814814814814816e-05, | |
| "loss": 2.5967, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.8765432098765433e-05, | |
| "loss": 2.4646, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.871604938271605e-05, | |
| "loss": 2.5669, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.866666666666667e-05, | |
| "loss": 2.6025, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.8617283950617286e-05, | |
| "loss": 2.5342, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.85679012345679e-05, | |
| "loss": 2.603, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.851851851851852e-05, | |
| "loss": 2.5439, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.846913580246914e-05, | |
| "loss": 2.6084, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.8419753086419754e-05, | |
| "loss": 2.5496, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.837037037037037e-05, | |
| "loss": 2.5586, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.832098765432099e-05, | |
| "loss": 2.5771, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.8271604938271607e-05, | |
| "loss": 2.5742, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.8222222222222224e-05, | |
| "loss": 2.5305, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.8172839506172842e-05, | |
| "loss": 2.5933, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.812345679012346e-05, | |
| "loss": 2.4597, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.8074074074074074e-05, | |
| "loss": 2.5312, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.802469135802469e-05, | |
| "loss": 2.4907, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.797530864197531e-05, | |
| "loss": 2.396, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.7925925925925927e-05, | |
| "loss": 2.4534, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.7876543209876545e-05, | |
| "loss": 2.4907, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.7827160493827162e-05, | |
| "loss": 2.4932, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.7777777777777777e-05, | |
| "loss": 2.6094, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.7728395061728398e-05, | |
| "loss": 2.449, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.7679012345679015e-05, | |
| "loss": 2.4902, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.7629629629629633e-05, | |
| "loss": 2.5059, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.7580246913580247e-05, | |
| "loss": 2.4456, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.7530864197530865e-05, | |
| "loss": 2.4866, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.7481481481481483e-05, | |
| "loss": 2.5183, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.74320987654321e-05, | |
| "loss": 2.4863, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.7382716049382718e-05, | |
| "loss": 2.5132, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.7333333333333336e-05, | |
| "loss": 2.4458, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.728395061728395e-05, | |
| "loss": 2.5205, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.7234567901234568e-05, | |
| "loss": 2.4124, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.7185185185185185e-05, | |
| "loss": 2.5034, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.7135802469135806e-05, | |
| "loss": 2.4858, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.708641975308642e-05, | |
| "loss": 2.4346, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.7037037037037038e-05, | |
| "loss": 2.4629, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.6987654320987656e-05, | |
| "loss": 2.4077, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.6938271604938274e-05, | |
| "loss": 2.449, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.688888888888889e-05, | |
| "loss": 2.5076, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.683950617283951e-05, | |
| "loss": 2.4656, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.6790123456790123e-05, | |
| "loss": 2.4553, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.674074074074074e-05, | |
| "loss": 2.4868, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.669135802469136e-05, | |
| "loss": 2.4116, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.6641975308641976e-05, | |
| "loss": 2.4417, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.6592592592592594e-05, | |
| "loss": 2.4167, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.654320987654321e-05, | |
| "loss": 2.3931, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.649382716049383e-05, | |
| "loss": 2.4436, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.6444444444444444e-05, | |
| "loss": 2.3916, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.6395061728395065e-05, | |
| "loss": 2.4133, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.6345679012345682e-05, | |
| "loss": 2.4873, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.6296296296296297e-05, | |
| "loss": 2.4497, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.6246913580246914e-05, | |
| "loss": 2.5217, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.6197530864197532e-05, | |
| "loss": 2.4292, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.614814814814815e-05, | |
| "loss": 2.4214, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.6098765432098767e-05, | |
| "loss": 2.4609, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.6049382716049385e-05, | |
| "loss": 2.3953, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.6000000000000003e-05, | |
| "loss": 2.5513, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.5950617283950617e-05, | |
| "loss": 2.4299, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.5901234567901235e-05, | |
| "loss": 2.3875, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.5851851851851852e-05, | |
| "loss": 2.4277, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.580246913580247e-05, | |
| "loss": 2.4368, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.5753086419753088e-05, | |
| "loss": 2.4065, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.5703703703703705e-05, | |
| "loss": 2.4888, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.565432098765432e-05, | |
| "loss": 2.332, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.560493827160494e-05, | |
| "loss": 2.3977, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.555555555555556e-05, | |
| "loss": 2.3906, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.5506172839506176e-05, | |
| "loss": 2.4995, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.545679012345679e-05, | |
| "loss": 2.4897, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.5407407407407408e-05, | |
| "loss": 2.4177, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 1.5358024691358026e-05, | |
| "loss": 2.4058, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 1.5308641975308643e-05, | |
| "loss": 2.4717, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 1.525925925925926e-05, | |
| "loss": 2.4084, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 1.5209876543209879e-05, | |
| "loss": 2.4502, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 1.5160493827160495e-05, | |
| "loss": 2.4185, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 1.5111111111111112e-05, | |
| "loss": 2.45, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 1.506172839506173e-05, | |
| "loss": 2.3867, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 1.5012345679012348e-05, | |
| "loss": 2.5342, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 1.4962962962962964e-05, | |
| "loss": 2.3738, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 1.4913580246913581e-05, | |
| "loss": 2.4482, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 1.4864197530864199e-05, | |
| "loss": 2.4639, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 1.4814814814814815e-05, | |
| "loss": 2.4148, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 1.4765432098765433e-05, | |
| "loss": 2.4602, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 1.471604938271605e-05, | |
| "loss": 2.5012, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 1.4666666666666666e-05, | |
| "loss": 2.3435, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 1.4617283950617284e-05, | |
| "loss": 2.4468, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 1.4567901234567903e-05, | |
| "loss": 2.3916, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 1.4518518518518521e-05, | |
| "loss": 2.4968, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 1.4469135802469137e-05, | |
| "loss": 2.47, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.4419753086419755e-05, | |
| "loss": 2.491, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.4370370370370372e-05, | |
| "loss": 2.479, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.4320987654320988e-05, | |
| "loss": 2.3792, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.4271604938271606e-05, | |
| "loss": 2.3987, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.4222222222222224e-05, | |
| "loss": 2.3359, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.417283950617284e-05, | |
| "loss": 2.3521, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.4123456790123457e-05, | |
| "loss": 2.4829, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.4074074074074075e-05, | |
| "loss": 2.4707, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.4024691358024694e-05, | |
| "loss": 2.4478, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.3975308641975309e-05, | |
| "loss": 2.5508, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.3925925925925928e-05, | |
| "loss": 2.3901, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.3876543209876546e-05, | |
| "loss": 2.5083, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.3827160493827162e-05, | |
| "loss": 2.5024, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.377777777777778e-05, | |
| "loss": 2.4165, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.3728395061728397e-05, | |
| "loss": 2.4336, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.3679012345679013e-05, | |
| "loss": 2.4294, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.362962962962963e-05, | |
| "loss": 2.4407, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.3580246913580248e-05, | |
| "loss": 2.4199, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.3530864197530866e-05, | |
| "loss": 2.4011, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.3481481481481482e-05, | |
| "loss": 2.3599, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.34320987654321e-05, | |
| "loss": 2.3601, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.3382716049382717e-05, | |
| "loss": 2.4021, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.3333333333333333e-05, | |
| "loss": 2.5396, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.3283950617283951e-05, | |
| "loss": 2.353, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.323456790123457e-05, | |
| "loss": 2.4529, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.3185185185185185e-05, | |
| "loss": 2.4265, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.3135802469135804e-05, | |
| "loss": 2.46, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.3086419753086422e-05, | |
| "loss": 2.5127, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.303703703703704e-05, | |
| "loss": 2.4165, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.2987654320987655e-05, | |
| "loss": 2.5366, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.2938271604938273e-05, | |
| "loss": 2.4021, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.288888888888889e-05, | |
| "loss": 2.3945, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.2839506172839507e-05, | |
| "loss": 2.5088, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.2790123456790124e-05, | |
| "loss": 2.4368, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.2740740740740742e-05, | |
| "loss": 2.448, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.2691358024691358e-05, | |
| "loss": 2.4092, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.2641975308641976e-05, | |
| "loss": 2.3987, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.2592592592592593e-05, | |
| "loss": 2.4045, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.2543209876543213e-05, | |
| "loss": 2.4639, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.2493827160493827e-05, | |
| "loss": 2.3909, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.2444444444444446e-05, | |
| "loss": 2.439, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.2395061728395064e-05, | |
| "loss": 2.3328, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.234567901234568e-05, | |
| "loss": 2.4443, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.2296296296296298e-05, | |
| "loss": 2.4656, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.2246913580246915e-05, | |
| "loss": 2.4329, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.2197530864197531e-05, | |
| "loss": 2.4866, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.2148148148148149e-05, | |
| "loss": 2.3904, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.2098765432098767e-05, | |
| "loss": 2.4604, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.2049382716049384e-05, | |
| "loss": 2.48, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.2e-05, | |
| "loss": 2.416, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.1950617283950618e-05, | |
| "loss": 2.4468, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.1901234567901236e-05, | |
| "loss": 2.4004, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.1851851851851852e-05, | |
| "loss": 2.4128, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.180246913580247e-05, | |
| "loss": 2.3652, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.1753086419753089e-05, | |
| "loss": 2.4504, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.1703703703703703e-05, | |
| "loss": 2.5515, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.1654320987654322e-05, | |
| "loss": 2.4644, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1.160493827160494e-05, | |
| "loss": 2.3645, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1.1555555555555556e-05, | |
| "loss": 2.3525, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.1506172839506174e-05, | |
| "loss": 2.4739, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.1456790123456791e-05, | |
| "loss": 2.4653, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.1407407407407409e-05, | |
| "loss": 2.4653, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.1358024691358025e-05, | |
| "loss": 2.4128, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.1308641975308643e-05, | |
| "loss": 2.384, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.125925925925926e-05, | |
| "loss": 2.5105, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.1209876543209876e-05, | |
| "loss": 2.3933, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.1160493827160494e-05, | |
| "loss": 2.332, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.1111111111111113e-05, | |
| "loss": 2.3579, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.1061728395061728e-05, | |
| "loss": 2.4592, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 1.1012345679012347e-05, | |
| "loss": 2.4473, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 1.0962962962962965e-05, | |
| "loss": 2.4998, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 1.0913580246913582e-05, | |
| "loss": 2.386, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 1.0864197530864198e-05, | |
| "loss": 2.3928, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 1.0814814814814816e-05, | |
| "loss": 2.3994, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 1.0765432098765434e-05, | |
| "loss": 2.4502, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.071604938271605e-05, | |
| "loss": 2.4927, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.0666666666666667e-05, | |
| "loss": 2.4275, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.0617283950617285e-05, | |
| "loss": 2.4023, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.0567901234567901e-05, | |
| "loss": 2.415, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.0518518518518519e-05, | |
| "loss": 2.4023, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.0469135802469136e-05, | |
| "loss": 2.4685, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.0419753086419756e-05, | |
| "loss": 2.4756, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 1.037037037037037e-05, | |
| "loss": 2.3962, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 1.032098765432099e-05, | |
| "loss": 2.478, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 1.0271604938271607e-05, | |
| "loss": 2.406, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 1.0222222222222223e-05, | |
| "loss": 2.4189, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 1.017283950617284e-05, | |
| "loss": 2.3921, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 1.0123456790123458e-05, | |
| "loss": 2.4441, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 1.0074074074074074e-05, | |
| "loss": 2.4219, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 1.0024691358024692e-05, | |
| "loss": 2.4368, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 9.97530864197531e-06, | |
| "loss": 2.3352, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 9.925925925925927e-06, | |
| "loss": 2.353, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 9.876543209876543e-06, | |
| "loss": 2.3994, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 9.827160493827161e-06, | |
| "loss": 2.4207, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 9.777777777777779e-06, | |
| "loss": 2.3557, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 9.728395061728396e-06, | |
| "loss": 2.333, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 9.679012345679012e-06, | |
| "loss": 2.5195, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 9.62962962962963e-06, | |
| "loss": 2.4937, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 9.580246913580248e-06, | |
| "loss": 2.5098, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 9.530864197530865e-06, | |
| "loss": 2.4614, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 9.481481481481483e-06, | |
| "loss": 2.3752, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 9.432098765432099e-06, | |
| "loss": 2.4026, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 9.382716049382717e-06, | |
| "loss": 2.3867, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 9.333333333333334e-06, | |
| "loss": 2.2954, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 9.28395061728395e-06, | |
| "loss": 2.4646, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 9.23456790123457e-06, | |
| "loss": 2.3757, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 9.185185185185186e-06, | |
| "loss": 2.3906, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 9.135802469135803e-06, | |
| "loss": 2.4526, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 9.086419753086421e-06, | |
| "loss": 2.4751, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 9.037037037037037e-06, | |
| "loss": 2.3799, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 8.987654320987655e-06, | |
| "loss": 2.4114, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 8.938271604938272e-06, | |
| "loss": 2.4612, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 8.888888888888888e-06, | |
| "loss": 2.4302, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 8.839506172839508e-06, | |
| "loss": 2.4646, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 8.790123456790124e-06, | |
| "loss": 2.4419, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 8.740740740740741e-06, | |
| "loss": 2.481, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 8.691358024691359e-06, | |
| "loss": 2.5029, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 8.641975308641975e-06, | |
| "loss": 2.4531, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 8.592592592592593e-06, | |
| "loss": 2.3938, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 8.54320987654321e-06, | |
| "loss": 2.3582, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 8.493827160493828e-06, | |
| "loss": 2.3772, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 8.444444444444446e-06, | |
| "loss": 2.3962, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 8.395061728395062e-06, | |
| "loss": 2.4185, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 8.34567901234568e-06, | |
| "loss": 2.4741, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 8.296296296296297e-06, | |
| "loss": 2.4082, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 8.246913580246915e-06, | |
| "loss": 2.3188, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 8.197530864197532e-06, | |
| "loss": 2.4124, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 8.148148148148148e-06, | |
| "loss": 2.4519, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 8.098765432098766e-06, | |
| "loss": 2.3682, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 8.049382716049384e-06, | |
| "loss": 2.408, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 2.3875, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 7.950617283950617e-06, | |
| "loss": 2.4319, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 7.901234567901235e-06, | |
| "loss": 2.3491, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 7.851851851851853e-06, | |
| "loss": 2.3796, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 7.80246913580247e-06, | |
| "loss": 2.4121, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 7.753086419753088e-06, | |
| "loss": 2.4019, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 7.703703703703704e-06, | |
| "loss": 2.4526, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 7.654320987654322e-06, | |
| "loss": 2.4099, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 7.604938271604939e-06, | |
| "loss": 2.4873, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 7.555555555555556e-06, | |
| "loss": 2.3733, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 7.506172839506174e-06, | |
| "loss": 2.2979, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 7.456790123456791e-06, | |
| "loss": 2.4429, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 7.4074074074074075e-06, | |
| "loss": 2.4038, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 7.358024691358025e-06, | |
| "loss": 2.4146, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 7.308641975308642e-06, | |
| "loss": 2.4619, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 7.2592592592592605e-06, | |
| "loss": 2.3833, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 7.209876543209877e-06, | |
| "loss": 2.3669, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 7.160493827160494e-06, | |
| "loss": 2.4238, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 7.111111111111112e-06, | |
| "loss": 2.3777, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 7.061728395061729e-06, | |
| "loss": 2.3972, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 7.012345679012347e-06, | |
| "loss": 2.4312, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 6.962962962962964e-06, | |
| "loss": 2.4358, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 6.913580246913581e-06, | |
| "loss": 2.3928, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 6.8641975308641985e-06, | |
| "loss": 2.4553, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 6.814814814814815e-06, | |
| "loss": 2.4233, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 6.765432098765433e-06, | |
| "loss": 2.3767, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 6.71604938271605e-06, | |
| "loss": 2.4814, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 2.3896, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 6.617283950617285e-06, | |
| "loss": 2.3113, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 6.567901234567902e-06, | |
| "loss": 2.3293, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 6.51851851851852e-06, | |
| "loss": 2.4878, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 6.4691358024691365e-06, | |
| "loss": 2.3333, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 6.419753086419753e-06, | |
| "loss": 2.491, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 6.370370370370371e-06, | |
| "loss": 2.439, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 6.320987654320988e-06, | |
| "loss": 2.4424, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 6.271604938271606e-06, | |
| "loss": 2.4631, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 6.222222222222223e-06, | |
| "loss": 2.4316, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 6.17283950617284e-06, | |
| "loss": 2.4048, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 6.123456790123458e-06, | |
| "loss": 2.3936, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 6.0740740740740745e-06, | |
| "loss": 2.3599, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 6.024691358024692e-06, | |
| "loss": 2.437, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 5.975308641975309e-06, | |
| "loss": 2.321, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 5.925925925925926e-06, | |
| "loss": 2.376, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 5.876543209876544e-06, | |
| "loss": 2.3794, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 5.827160493827161e-06, | |
| "loss": 2.3853, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 5.777777777777778e-06, | |
| "loss": 2.2947, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 5.728395061728396e-06, | |
| "loss": 2.46, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 5.6790123456790125e-06, | |
| "loss": 2.4282, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 5.62962962962963e-06, | |
| "loss": 2.4805, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 5.580246913580247e-06, | |
| "loss": 2.4863, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 5.530864197530864e-06, | |
| "loss": 2.4629, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 5.481481481481482e-06, | |
| "loss": 2.395, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 5.432098765432099e-06, | |
| "loss": 2.3828, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 5.382716049382717e-06, | |
| "loss": 2.4131, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 5.333333333333334e-06, | |
| "loss": 2.4619, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 5.2839506172839505e-06, | |
| "loss": 2.4124, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 5.234567901234568e-06, | |
| "loss": 2.4143, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 5.185185185185185e-06, | |
| "loss": 2.4607, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 5.1358024691358035e-06, | |
| "loss": 2.4197, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 5.08641975308642e-06, | |
| "loss": 2.4399, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 5.037037037037037e-06, | |
| "loss": 2.4785, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 4.987654320987655e-06, | |
| "loss": 2.4285, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 4.938271604938272e-06, | |
| "loss": 2.396, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 4.888888888888889e-06, | |
| "loss": 2.3718, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 4.839506172839506e-06, | |
| "loss": 2.3928, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 4.790123456790124e-06, | |
| "loss": 2.3792, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 4.7407407407407415e-06, | |
| "loss": 2.3079, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 4.691358024691358e-06, | |
| "loss": 2.4497, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 4.641975308641975e-06, | |
| "loss": 2.4268, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 4.592592592592593e-06, | |
| "loss": 2.3848, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 4.5432098765432105e-06, | |
| "loss": 2.3687, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 4.493827160493827e-06, | |
| "loss": 2.3835, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 4.444444444444444e-06, | |
| "loss": 2.3953, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 4.395061728395062e-06, | |
| "loss": 2.4038, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 4.3456790123456795e-06, | |
| "loss": 2.4141, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 4.296296296296296e-06, | |
| "loss": 2.4561, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 4.246913580246914e-06, | |
| "loss": 2.4038, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 4.197530864197531e-06, | |
| "loss": 2.3938, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 4.1481481481481485e-06, | |
| "loss": 2.3979, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 4.098765432098766e-06, | |
| "loss": 2.3779, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 4.049382716049383e-06, | |
| "loss": 2.3391, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 2.3933, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 3.9506172839506175e-06, | |
| "loss": 2.3989, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 3.901234567901235e-06, | |
| "loss": 2.4363, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 3.851851851851852e-06, | |
| "loss": 2.3862, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 3.8024691358024697e-06, | |
| "loss": 2.3481, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 3.753086419753087e-06, | |
| "loss": 2.4031, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 3.7037037037037037e-06, | |
| "loss": 2.3621, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 3.654320987654321e-06, | |
| "loss": 2.458, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 3.6049382716049387e-06, | |
| "loss": 2.3096, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 3.555555555555556e-06, | |
| "loss": 2.4158, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 3.5061728395061736e-06, | |
| "loss": 2.3813, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 3.4567901234567904e-06, | |
| "loss": 2.4363, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 3.4074074074074077e-06, | |
| "loss": 2.4714, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 3.358024691358025e-06, | |
| "loss": 2.4241, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 3.3086419753086426e-06, | |
| "loss": 2.3762, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 3.25925925925926e-06, | |
| "loss": 2.3765, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 3.2098765432098767e-06, | |
| "loss": 2.377, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 3.160493827160494e-06, | |
| "loss": 2.4639, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 3.1111111111111116e-06, | |
| "loss": 2.3855, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 3.061728395061729e-06, | |
| "loss": 2.3376, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 3.012345679012346e-06, | |
| "loss": 2.4277, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 2.962962962962963e-06, | |
| "loss": 2.4199, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 2.9135802469135806e-06, | |
| "loss": 2.4023, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 2.864197530864198e-06, | |
| "loss": 2.4697, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 2.814814814814815e-06, | |
| "loss": 2.4285, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 2.765432098765432e-06, | |
| "loss": 2.4294, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 2.7160493827160496e-06, | |
| "loss": 2.3633, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 2.666666666666667e-06, | |
| "loss": 2.3977, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 2.617283950617284e-06, | |
| "loss": 2.376, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 2.5679012345679018e-06, | |
| "loss": 2.4163, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 2.5185185185185186e-06, | |
| "loss": 2.3892, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 2.469135802469136e-06, | |
| "loss": 2.3967, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 2.419753086419753e-06, | |
| "loss": 2.4109, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 2.3703703703703707e-06, | |
| "loss": 2.4153, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 2.3209876543209876e-06, | |
| "loss": 2.3718, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 2.2716049382716052e-06, | |
| "loss": 2.3638, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 2.222222222222222e-06, | |
| "loss": 2.4248, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 2.1728395061728397e-06, | |
| "loss": 2.4958, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 2.123456790123457e-06, | |
| "loss": 2.3921, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 2.0740740740740742e-06, | |
| "loss": 2.439, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 2.0246913580246915e-06, | |
| "loss": 2.3945, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 1.9753086419753087e-06, | |
| "loss": 2.3296, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 1.925925925925926e-06, | |
| "loss": 2.4314, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 1.8765432098765435e-06, | |
| "loss": 2.354, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 1.8271604938271605e-06, | |
| "loss": 2.3184, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 1.777777777777778e-06, | |
| "loss": 2.4026, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 1.7283950617283952e-06, | |
| "loss": 2.3965, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 1.6790123456790125e-06, | |
| "loss": 2.3975, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 1.62962962962963e-06, | |
| "loss": 2.3599, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 1.580246913580247e-06, | |
| "loss": 2.3606, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 1.5308641975308644e-06, | |
| "loss": 2.3994, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 1.4814814814814815e-06, | |
| "loss": 2.3625, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 1.432098765432099e-06, | |
| "loss": 2.3979, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 1.382716049382716e-06, | |
| "loss": 2.4016, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 1.3333333333333334e-06, | |
| "loss": 2.3835, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 1.2839506172839509e-06, | |
| "loss": 2.4111, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 1.234567901234568e-06, | |
| "loss": 2.3447, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 1.1851851851851854e-06, | |
| "loss": 2.4204, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 1.1358024691358026e-06, | |
| "loss": 2.5098, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 1.0864197530864199e-06, | |
| "loss": 2.4341, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 1.0370370370370371e-06, | |
| "loss": 2.4258, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 9.876543209876544e-07, | |
| "loss": 2.3809, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 9.382716049382717e-07, | |
| "loss": 2.3953, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 8.88888888888889e-07, | |
| "loss": 2.4319, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 8.395061728395062e-07, | |
| "loss": 2.4688, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 7.901234567901235e-07, | |
| "loss": 2.5205, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 7.407407407407407e-07, | |
| "loss": 2.4153, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 6.91358024691358e-07, | |
| "loss": 2.3594, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 6.419753086419754e-07, | |
| "loss": 2.3601, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 5.925925925925927e-07, | |
| "loss": 2.4023, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 5.432098765432099e-07, | |
| "loss": 2.4355, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 4.938271604938272e-07, | |
| "loss": 2.353, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 4.444444444444445e-07, | |
| "loss": 2.4204, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 3.9506172839506174e-07, | |
| "loss": 2.4241, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 3.45679012345679e-07, | |
| "loss": 2.3945, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 2.9629629629629634e-07, | |
| "loss": 2.4585, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 2.469135802469136e-07, | |
| "loss": 2.3467, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 1.9753086419753087e-07, | |
| "loss": 2.3896, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 1.4814814814814817e-07, | |
| "loss": 2.4189, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 9.876543209876543e-08, | |
| "loss": 2.4871, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 4.938271604938272e-08, | |
| "loss": 2.4495, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "step": 418, | |
| "total_flos": 3.5393273788471706e+17, | |
| "train_loss": 2.4495703592254783, | |
| "train_runtime": 1378.2566, | |
| "train_samples_per_second": 38.878, | |
| "train_steps_per_second": 0.303 | |
| } | |
| ], | |
| "max_steps": 418, | |
| "num_train_epochs": 2, | |
| "total_flos": 3.5393273788471706e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |