|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 2.963855421686747, |
|
"global_step": 123, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 5e-06, |
|
"loss": 12.1442, |
|
"step": 1 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 1e-05, |
|
"loss": 12.2066, |
|
"step": 2 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 1.5000000000000002e-05, |
|
"loss": 11.4198, |
|
"step": 3 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 2e-05, |
|
"loss": 10.553, |
|
"step": 4 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 1.9996515418688493e-05, |
|
"loss": 11.5989, |
|
"step": 5 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 1.998606410321534e-05, |
|
"loss": 10.1464, |
|
"step": 6 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 1.9968653337272262e-05, |
|
"loss": 9.2548, |
|
"step": 7 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 1.9944295254705187e-05, |
|
"loss": 8.5892, |
|
"step": 8 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 1.9913006831057967e-05, |
|
"loss": 7.8936, |
|
"step": 9 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 1.9874809871741877e-05, |
|
"loss": 8.0968, |
|
"step": 10 |
|
}, |
|
{ |
|
"epoch": 0.27, |
|
"learning_rate": 1.982973099683902e-05, |
|
"loss": 7.8131, |
|
"step": 11 |
|
}, |
|
{ |
|
"epoch": 0.29, |
|
"learning_rate": 1.977780162255041e-05, |
|
"loss": 7.504, |
|
"step": 12 |
|
}, |
|
{ |
|
"epoch": 0.31, |
|
"learning_rate": 1.9719057939301477e-05, |
|
"loss": 7.3231, |
|
"step": 13 |
|
}, |
|
{ |
|
"epoch": 0.34, |
|
"learning_rate": 1.9653540886520387e-05, |
|
"loss": 7.0976, |
|
"step": 14 |
|
}, |
|
{ |
|
"epoch": 0.36, |
|
"learning_rate": 1.9581296124106682e-05, |
|
"loss": 6.9036, |
|
"step": 15 |
|
}, |
|
{ |
|
"epoch": 0.39, |
|
"learning_rate": 1.9502374000610152e-05, |
|
"loss": 6.6765, |
|
"step": 16 |
|
}, |
|
{ |
|
"epoch": 0.41, |
|
"learning_rate": 1.941682951814212e-05, |
|
"loss": 6.4822, |
|
"step": 17 |
|
}, |
|
{ |
|
"epoch": 0.43, |
|
"learning_rate": 1.932472229404356e-05, |
|
"loss": 6.4258, |
|
"step": 18 |
|
}, |
|
{ |
|
"epoch": 0.46, |
|
"learning_rate": 1.922611651933683e-05, |
|
"loss": 6.2448, |
|
"step": 19 |
|
}, |
|
{ |
|
"epoch": 0.48, |
|
"learning_rate": 1.912108091398988e-05, |
|
"loss": 6.0179, |
|
"step": 20 |
|
}, |
|
{ |
|
"epoch": 0.51, |
|
"learning_rate": 1.900968867902419e-05, |
|
"loss": 5.8236, |
|
"step": 21 |
|
}, |
|
{ |
|
"epoch": 0.53, |
|
"learning_rate": 1.8892017445499812e-05, |
|
"loss": 5.8436, |
|
"step": 22 |
|
}, |
|
{ |
|
"epoch": 0.55, |
|
"learning_rate": 1.876814922041299e-05, |
|
"loss": 5.7092, |
|
"step": 23 |
|
}, |
|
{ |
|
"epoch": 0.58, |
|
"learning_rate": 1.8638170329544164e-05, |
|
"loss": 5.5598, |
|
"step": 24 |
|
}, |
|
{ |
|
"epoch": 0.6, |
|
"learning_rate": 1.8502171357296144e-05, |
|
"loss": 5.5397, |
|
"step": 25 |
|
}, |
|
{ |
|
"epoch": 0.63, |
|
"learning_rate": 1.8360247083564343e-05, |
|
"loss": 5.3061, |
|
"step": 26 |
|
}, |
|
{ |
|
"epoch": 0.65, |
|
"learning_rate": 1.8212496417683135e-05, |
|
"loss": 5.2484, |
|
"step": 27 |
|
}, |
|
{ |
|
"epoch": 0.67, |
|
"learning_rate": 1.805902232949435e-05, |
|
"loss": 5.4282, |
|
"step": 28 |
|
}, |
|
{ |
|
"epoch": 0.7, |
|
"learning_rate": 1.789993177758588e-05, |
|
"loss": 5.1144, |
|
"step": 29 |
|
}, |
|
{ |
|
"epoch": 0.72, |
|
"learning_rate": 1.773533563475053e-05, |
|
"loss": 4.9468, |
|
"step": 30 |
|
}, |
|
{ |
|
"epoch": 0.75, |
|
"learning_rate": 1.7565348610716963e-05, |
|
"loss": 5.0151, |
|
"step": 31 |
|
}, |
|
{ |
|
"epoch": 0.77, |
|
"learning_rate": 1.7390089172206594e-05, |
|
"loss": 4.7214, |
|
"step": 32 |
|
}, |
|
{ |
|
"epoch": 0.8, |
|
"learning_rate": 1.720967946037225e-05, |
|
"loss": 4.6076, |
|
"step": 33 |
|
}, |
|
{ |
|
"epoch": 0.82, |
|
"learning_rate": 1.7024245205675986e-05, |
|
"loss": 4.7376, |
|
"step": 34 |
|
}, |
|
{ |
|
"epoch": 0.84, |
|
"learning_rate": 1.6833915640265485e-05, |
|
"loss": 4.4775, |
|
"step": 35 |
|
}, |
|
{ |
|
"epoch": 0.87, |
|
"learning_rate": 1.6638823407910085e-05, |
|
"loss": 4.5881, |
|
"step": 36 |
|
}, |
|
{ |
|
"epoch": 0.89, |
|
"learning_rate": 1.6439104471559157e-05, |
|
"loss": 4.3676, |
|
"step": 37 |
|
}, |
|
{ |
|
"epoch": 0.92, |
|
"learning_rate": 1.6234898018587336e-05, |
|
"loss": 4.6396, |
|
"step": 38 |
|
}, |
|
{ |
|
"epoch": 0.94, |
|
"learning_rate": 1.6026346363792565e-05, |
|
"loss": 4.1056, |
|
"step": 39 |
|
}, |
|
{ |
|
"epoch": 0.96, |
|
"learning_rate": 1.58135948502146e-05, |
|
"loss": 4.0982, |
|
"step": 40 |
|
}, |
|
{ |
|
"epoch": 0.99, |
|
"learning_rate": 1.5596791747843083e-05, |
|
"loss": 4.2723, |
|
"step": 41 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 1.5376088150285777e-05, |
|
"loss": 3.942, |
|
"step": 42 |
|
}, |
|
{ |
|
"epoch": 1.04, |
|
"learning_rate": 1.515163786946896e-05, |
|
"loss": 3.6775, |
|
"step": 43 |
|
}, |
|
{ |
|
"epoch": 1.06, |
|
"learning_rate": 1.4923597328443423e-05, |
|
"loss": 3.3249, |
|
"step": 44 |
|
}, |
|
{ |
|
"epoch": 1.08, |
|
"learning_rate": 1.4692125452370664e-05, |
|
"loss": 3.4726, |
|
"step": 45 |
|
}, |
|
{ |
|
"epoch": 1.11, |
|
"learning_rate": 1.4457383557765385e-05, |
|
"loss": 3.2118, |
|
"step": 46 |
|
}, |
|
{ |
|
"epoch": 1.13, |
|
"learning_rate": 1.4219535240071378e-05, |
|
"loss": 3.2224, |
|
"step": 47 |
|
}, |
|
{ |
|
"epoch": 1.16, |
|
"learning_rate": 1.397874625964921e-05, |
|
"loss": 3.1223, |
|
"step": 48 |
|
}, |
|
{ |
|
"epoch": 1.18, |
|
"learning_rate": 1.3735184426255117e-05, |
|
"loss": 3.1626, |
|
"step": 49 |
|
}, |
|
{ |
|
"epoch": 1.2, |
|
"learning_rate": 1.348901948209167e-05, |
|
"loss": 2.9617, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 1.23, |
|
"learning_rate": 1.324042298351166e-05, |
|
"loss": 2.8352, |
|
"step": 51 |
|
}, |
|
{ |
|
"epoch": 1.25, |
|
"learning_rate": 1.2989568181457704e-05, |
|
"loss": 3.3436, |
|
"step": 52 |
|
}, |
|
{ |
|
"epoch": 1.28, |
|
"learning_rate": 1.2736629900720832e-05, |
|
"loss": 3.2803, |
|
"step": 53 |
|
}, |
|
{ |
|
"epoch": 1.3, |
|
"learning_rate": 1.248178441810224e-05, |
|
"loss": 3.129, |
|
"step": 54 |
|
}, |
|
{ |
|
"epoch": 1.33, |
|
"learning_rate": 1.2225209339563144e-05, |
|
"loss": 3.0431, |
|
"step": 55 |
|
}, |
|
{ |
|
"epoch": 1.35, |
|
"learning_rate": 1.1967083476448282e-05, |
|
"loss": 3.0114, |
|
"step": 56 |
|
}, |
|
{ |
|
"epoch": 1.37, |
|
"learning_rate": 1.1707586720869375e-05, |
|
"loss": 3.0263, |
|
"step": 57 |
|
}, |
|
{ |
|
"epoch": 1.4, |
|
"learning_rate": 1.1446899920335407e-05, |
|
"loss": 2.8614, |
|
"step": 58 |
|
}, |
|
{ |
|
"epoch": 1.42, |
|
"learning_rate": 1.118520475171703e-05, |
|
"loss": 2.8583, |
|
"step": 59 |
|
}, |
|
{ |
|
"epoch": 1.45, |
|
"learning_rate": 1.092268359463302e-05, |
|
"loss": 2.8472, |
|
"step": 60 |
|
}, |
|
{ |
|
"epoch": 1.47, |
|
"learning_rate": 1.0659519404346955e-05, |
|
"loss": 2.7311, |
|
"step": 61 |
|
}, |
|
{ |
|
"epoch": 1.49, |
|
"learning_rate": 1.0395895584262696e-05, |
|
"loss": 2.6286, |
|
"step": 62 |
|
}, |
|
{ |
|
"epoch": 1.52, |
|
"learning_rate": 1.013199585810759e-05, |
|
"loss": 2.7301, |
|
"step": 63 |
|
}, |
|
{ |
|
"epoch": 1.54, |
|
"learning_rate": 9.868004141892412e-06, |
|
"loss": 2.7739, |
|
"step": 64 |
|
}, |
|
{ |
|
"epoch": 1.57, |
|
"learning_rate": 9.604104415737309e-06, |
|
"loss": 2.6254, |
|
"step": 65 |
|
}, |
|
{ |
|
"epoch": 1.59, |
|
"learning_rate": 9.340480595653047e-06, |
|
"loss": 2.6452, |
|
"step": 66 |
|
}, |
|
{ |
|
"epoch": 1.61, |
|
"learning_rate": 9.07731640536698e-06, |
|
"loss": 2.7975, |
|
"step": 67 |
|
}, |
|
{ |
|
"epoch": 1.64, |
|
"learning_rate": 8.814795248282974e-06, |
|
"loss": 2.7089, |
|
"step": 68 |
|
}, |
|
{ |
|
"epoch": 1.66, |
|
"learning_rate": 8.553100079664598e-06, |
|
"loss": 2.7918, |
|
"step": 69 |
|
}, |
|
{ |
|
"epoch": 1.69, |
|
"learning_rate": 8.292413279130625e-06, |
|
"loss": 2.5251, |
|
"step": 70 |
|
}, |
|
{ |
|
"epoch": 1.71, |
|
"learning_rate": 8.03291652355172e-06, |
|
"loss": 2.5613, |
|
"step": 71 |
|
}, |
|
{ |
|
"epoch": 1.73, |
|
"learning_rate": 7.774790660436857e-06, |
|
"loss": 2.7616, |
|
"step": 72 |
|
}, |
|
{ |
|
"epoch": 1.76, |
|
"learning_rate": 7.518215581897763e-06, |
|
"loss": 2.6208, |
|
"step": 73 |
|
}, |
|
{ |
|
"epoch": 1.78, |
|
"learning_rate": 7.263370099279173e-06, |
|
"loss": 2.7671, |
|
"step": 74 |
|
}, |
|
{ |
|
"epoch": 1.81, |
|
"learning_rate": 7.010431818542298e-06, |
|
"loss": 2.5627, |
|
"step": 75 |
|
}, |
|
{ |
|
"epoch": 1.83, |
|
"learning_rate": 6.759577016488343e-06, |
|
"loss": 2.5638, |
|
"step": 76 |
|
}, |
|
{ |
|
"epoch": 1.86, |
|
"learning_rate": 6.510980517908334e-06, |
|
"loss": 2.4598, |
|
"step": 77 |
|
}, |
|
{ |
|
"epoch": 1.88, |
|
"learning_rate": 6.264815573744884e-06, |
|
"loss": 2.5585, |
|
"step": 78 |
|
}, |
|
{ |
|
"epoch": 1.9, |
|
"learning_rate": 6.021253740350793e-06, |
|
"loss": 2.7079, |
|
"step": 79 |
|
}, |
|
{ |
|
"epoch": 1.93, |
|
"learning_rate": 5.780464759928623e-06, |
|
"loss": 2.4145, |
|
"step": 80 |
|
}, |
|
{ |
|
"epoch": 1.95, |
|
"learning_rate": 5.542616442234618e-06, |
|
"loss": 2.7065, |
|
"step": 81 |
|
}, |
|
{ |
|
"epoch": 1.98, |
|
"learning_rate": 5.307874547629339e-06, |
|
"loss": 2.588, |
|
"step": 82 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"learning_rate": 5.076402671556578e-06, |
|
"loss": 2.2886, |
|
"step": 83 |
|
}, |
|
{ |
|
"epoch": 2.02, |
|
"learning_rate": 4.848362130531039e-06, |
|
"loss": 1.7128, |
|
"step": 84 |
|
}, |
|
{ |
|
"epoch": 2.05, |
|
"learning_rate": 4.623911849714226e-06, |
|
"loss": 1.7341, |
|
"step": 85 |
|
}, |
|
{ |
|
"epoch": 2.07, |
|
"learning_rate": 4.403208252156921e-06, |
|
"loss": 1.9271, |
|
"step": 86 |
|
}, |
|
{ |
|
"epoch": 2.1, |
|
"learning_rate": 4.186405149785403e-06, |
|
"loss": 1.7111, |
|
"step": 87 |
|
}, |
|
{ |
|
"epoch": 2.12, |
|
"learning_rate": 3.973653636207437e-06, |
|
"loss": 1.7875, |
|
"step": 88 |
|
}, |
|
{ |
|
"epoch": 2.14, |
|
"learning_rate": 3.7651019814126656e-06, |
|
"loss": 1.633, |
|
"step": 89 |
|
}, |
|
{ |
|
"epoch": 2.17, |
|
"learning_rate": 3.560895528440844e-06, |
|
"loss": 1.5634, |
|
"step": 90 |
|
}, |
|
{ |
|
"epoch": 2.19, |
|
"learning_rate": 3.361176592089919e-06, |
|
"loss": 1.6087, |
|
"step": 91 |
|
}, |
|
{ |
|
"epoch": 2.22, |
|
"learning_rate": 3.1660843597345137e-06, |
|
"loss": 1.7095, |
|
"step": 92 |
|
}, |
|
{ |
|
"epoch": 2.24, |
|
"learning_rate": 2.975754794324015e-06, |
|
"loss": 1.7235, |
|
"step": 93 |
|
}, |
|
{ |
|
"epoch": 2.27, |
|
"learning_rate": 2.7903205396277546e-06, |
|
"loss": 1.5522, |
|
"step": 94 |
|
}, |
|
{ |
|
"epoch": 2.29, |
|
"learning_rate": 2.6099108277934105e-06, |
|
"loss": 1.594, |
|
"step": 95 |
|
}, |
|
{ |
|
"epoch": 2.31, |
|
"learning_rate": 2.4346513892830427e-06, |
|
"loss": 1.5743, |
|
"step": 96 |
|
}, |
|
{ |
|
"epoch": 2.34, |
|
"learning_rate": 2.2646643652494693e-06, |
|
"loss": 1.8422, |
|
"step": 97 |
|
}, |
|
{ |
|
"epoch": 2.36, |
|
"learning_rate": 2.100068222414121e-06, |
|
"loss": 1.5189, |
|
"step": 98 |
|
}, |
|
{ |
|
"epoch": 2.39, |
|
"learning_rate": 1.9409776705056514e-06, |
|
"loss": 1.633, |
|
"step": 99 |
|
}, |
|
{ |
|
"epoch": 2.41, |
|
"learning_rate": 1.7875035823168641e-06, |
|
"loss": 1.5292, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 2.43, |
|
"learning_rate": 1.6397529164356606e-06, |
|
"loss": 1.5055, |
|
"step": 101 |
|
}, |
|
{ |
|
"epoch": 2.46, |
|
"learning_rate": 1.4978286427038602e-06, |
|
"loss": 1.6929, |
|
"step": 102 |
|
}, |
|
{ |
|
"epoch": 2.48, |
|
"learning_rate": 1.3618296704558364e-06, |
|
"loss": 1.6472, |
|
"step": 103 |
|
}, |
|
{ |
|
"epoch": 2.51, |
|
"learning_rate": 1.2318507795870138e-06, |
|
"loss": 1.5057, |
|
"step": 104 |
|
}, |
|
{ |
|
"epoch": 2.53, |
|
"learning_rate": 1.1079825545001887e-06, |
|
"loss": 1.4482, |
|
"step": 105 |
|
}, |
|
{ |
|
"epoch": 2.55, |
|
"learning_rate": 9.903113209758098e-07, |
|
"loss": 1.7398, |
|
"step": 106 |
|
}, |
|
{ |
|
"epoch": 2.58, |
|
"learning_rate": 8.789190860101226e-07, |
|
"loss": 1.604, |
|
"step": 107 |
|
}, |
|
{ |
|
"epoch": 2.6, |
|
"learning_rate": 7.738834806631712e-07, |
|
"loss": 1.6135, |
|
"step": 108 |
|
}, |
|
{ |
|
"epoch": 2.63, |
|
"learning_rate": 6.752777059564431e-07, |
|
"loss": 1.5833, |
|
"step": 109 |
|
}, |
|
{ |
|
"epoch": 2.65, |
|
"learning_rate": 5.831704818578842e-07, |
|
"loss": 1.8117, |
|
"step": 110 |
|
}, |
|
{ |
|
"epoch": 2.67, |
|
"learning_rate": 4.976259993898503e-07, |
|
"loss": 1.5287, |
|
"step": 111 |
|
}, |
|
{ |
|
"epoch": 2.7, |
|
"learning_rate": 4.187038758933204e-07, |
|
"loss": 1.5151, |
|
"step": 112 |
|
}, |
|
{ |
|
"epoch": 2.72, |
|
"learning_rate": 3.4645911347961357e-07, |
|
"loss": 1.5188, |
|
"step": 113 |
|
}, |
|
{ |
|
"epoch": 2.75, |
|
"learning_rate": 2.809420606985236e-07, |
|
"loss": 1.5203, |
|
"step": 114 |
|
}, |
|
{ |
|
"epoch": 2.77, |
|
"learning_rate": 2.2219837744959284e-07, |
|
"loss": 1.6247, |
|
"step": 115 |
|
}, |
|
{ |
|
"epoch": 2.8, |
|
"learning_rate": 1.7026900316098217e-07, |
|
"loss": 1.4147, |
|
"step": 116 |
|
}, |
|
{ |
|
"epoch": 2.82, |
|
"learning_rate": 1.2519012825812804e-07, |
|
"loss": 1.6317, |
|
"step": 117 |
|
}, |
|
{ |
|
"epoch": 2.84, |
|
"learning_rate": 8.699316894203225e-08, |
|
"loss": 1.6046, |
|
"step": 118 |
|
}, |
|
{ |
|
"epoch": 2.87, |
|
"learning_rate": 5.5704745294815624e-08, |
|
"loss": 1.5425, |
|
"step": 119 |
|
}, |
|
{ |
|
"epoch": 2.89, |
|
"learning_rate": 3.134666272774034e-08, |
|
"loss": 1.622, |
|
"step": 120 |
|
}, |
|
{ |
|
"epoch": 2.92, |
|
"learning_rate": 1.3935896784663671e-08, |
|
"loss": 1.597, |
|
"step": 121 |
|
}, |
|
{ |
|
"epoch": 2.94, |
|
"learning_rate": 3.4845813115114147e-09, |
|
"loss": 1.6068, |
|
"step": 122 |
|
}, |
|
{ |
|
"epoch": 2.96, |
|
"learning_rate": 0.0, |
|
"loss": 1.5153, |
|
"step": 123 |
|
}, |
|
{ |
|
"epoch": 2.96, |
|
"step": 123, |
|
"total_flos": 2.008420388228301e+16, |
|
"train_loss": 3.741002580983852, |
|
"train_runtime": 2257.2991, |
|
"train_samples_per_second": 7.05, |
|
"train_steps_per_second": 0.054 |
|
} |
|
], |
|
"max_steps": 123, |
|
"num_train_epochs": 3, |
|
"total_flos": 2.008420388228301e+16, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|