|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 0.400736000878807, |
|
"eval_steps": 500, |
|
"global_step": 228, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 2.8571428571428573e-06, |
|
"loss": 2.0419, |
|
"step": 1 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 5.7142857142857145e-06, |
|
"loss": 2.1227, |
|
"step": 2 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 8.571428571428571e-06, |
|
"loss": 2.0401, |
|
"step": 3 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 1.1428571428571429e-05, |
|
"loss": 2.0777, |
|
"step": 4 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 1.4285714285714287e-05, |
|
"loss": 1.9235, |
|
"step": 5 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 1.7142857142857142e-05, |
|
"loss": 1.8527, |
|
"step": 6 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 2e-05, |
|
"loss": 1.8595, |
|
"step": 7 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 1.9998989635940996e-05, |
|
"loss": 1.8005, |
|
"step": 8 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 1.9995958747931083e-05, |
|
"loss": 1.7729, |
|
"step": 9 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 1.9990907948430327e-05, |
|
"loss": 1.7767, |
|
"step": 10 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 1.998383825806799e-05, |
|
"loss": 1.733, |
|
"step": 11 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 1.9974751105436266e-05, |
|
"loss": 1.7177, |
|
"step": 12 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 1.9963648326801653e-05, |
|
"loss": 1.6892, |
|
"step": 13 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 1.9950532165733847e-05, |
|
"loss": 1.6988, |
|
"step": 14 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 1.993540527265239e-05, |
|
"loss": 1.6935, |
|
"step": 15 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 1.9918270704291104e-05, |
|
"loss": 1.7013, |
|
"step": 16 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 1.98991319230804e-05, |
|
"loss": 1.6657, |
|
"step": 17 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 1.9877992796447604e-05, |
|
"loss": 1.663, |
|
"step": 18 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 1.9854857596035476e-05, |
|
"loss": 1.6739, |
|
"step": 19 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 1.982973099683902e-05, |
|
"loss": 1.6262, |
|
"step": 20 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 1.9802618076260784e-05, |
|
"loss": 1.7099, |
|
"step": 21 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 1.9773524313084857e-05, |
|
"loss": 1.6464, |
|
"step": 22 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 1.9742455586369786e-05, |
|
"loss": 1.6538, |
|
"step": 23 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 1.9709418174260523e-05, |
|
"loss": 1.6665, |
|
"step": 24 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 1.9674418752719835e-05, |
|
"loss": 1.6097, |
|
"step": 25 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 1.963746439417924e-05, |
|
"loss": 1.6503, |
|
"step": 26 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 1.959856256610988e-05, |
|
"loss": 1.658, |
|
"step": 27 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 1.9557721129513538e-05, |
|
"loss": 1.6575, |
|
"step": 28 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 1.9514948337334144e-05, |
|
"loss": 1.6055, |
|
"step": 29 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 1.947025283279008e-05, |
|
"loss": 1.6067, |
|
"step": 30 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 1.9423643647627625e-05, |
|
"loss": 1.5934, |
|
"step": 31 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 1.937513020029588e-05, |
|
"loss": 1.6424, |
|
"step": 32 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 1.932472229404356e-05, |
|
"loss": 1.6179, |
|
"step": 33 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 1.9272430114938018e-05, |
|
"loss": 1.5971, |
|
"step": 34 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 1.9218264229806917e-05, |
|
"loss": 1.5947, |
|
"step": 35 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 1.9162235584102973e-05, |
|
"loss": 1.5637, |
|
"step": 36 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 1.9104355499692166e-05, |
|
"loss": 1.6186, |
|
"step": 37 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 1.9044635672565898e-05, |
|
"loss": 1.5754, |
|
"step": 38 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 1.8983088170477556e-05, |
|
"loss": 1.6139, |
|
"step": 39 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 1.8919725430503946e-05, |
|
"loss": 1.5928, |
|
"step": 40 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 1.8854560256532098e-05, |
|
"loss": 1.5804, |
|
"step": 41 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 1.8787605816671956e-05, |
|
"loss": 1.6165, |
|
"step": 42 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 1.8718875640595432e-05, |
|
"loss": 1.6021, |
|
"step": 43 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 1.864838361680247e-05, |
|
"loss": 1.5869, |
|
"step": 44 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 1.8576143989814524e-05, |
|
"loss": 1.5575, |
|
"step": 45 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 1.8502171357296144e-05, |
|
"loss": 1.6145, |
|
"step": 46 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 1.8426480667105178e-05, |
|
"loss": 1.5553, |
|
"step": 47 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 1.8349087214272222e-05, |
|
"loss": 1.5679, |
|
"step": 48 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 1.8270006637909907e-05, |
|
"loss": 1.5606, |
|
"step": 49 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 1.818925491805265e-05, |
|
"loss": 1.5488, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 1.810684837242755e-05, |
|
"loss": 1.5561, |
|
"step": 51 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 1.8022803653156983e-05, |
|
"loss": 1.5796, |
|
"step": 52 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 1.7937137743393695e-05, |
|
"loss": 1.565, |
|
"step": 53 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 1.784986795388895e-05, |
|
"loss": 1.5969, |
|
"step": 54 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 1.776101191949449e-05, |
|
"loss": 1.5392, |
|
"step": 55 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 1.7670587595599034e-05, |
|
"loss": 1.5849, |
|
"step": 56 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 1.757861325449997e-05, |
|
"loss": 1.582, |
|
"step": 57 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 1.7485107481711014e-05, |
|
"loss": 1.5919, |
|
"step": 58 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 1.7390089172206594e-05, |
|
"loss": 1.5639, |
|
"step": 59 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 1.7293577526603684e-05, |
|
"loss": 1.5829, |
|
"step": 60 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 1.719559204728188e-05, |
|
"loss": 1.5662, |
|
"step": 61 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 1.7096152534442515e-05, |
|
"loss": 1.5844, |
|
"step": 62 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 1.6995279082107537e-05, |
|
"loss": 1.5202, |
|
"step": 63 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 1.689299207405911e-05, |
|
"loss": 1.5726, |
|
"step": 64 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 1.678931217972055e-05, |
|
"loss": 1.5644, |
|
"step": 65 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 1.6684260349979637e-05, |
|
"loss": 1.5389, |
|
"step": 66 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 1.6577857812954994e-05, |
|
"loss": 1.5473, |
|
"step": 67 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 1.6470126069706456e-05, |
|
"loss": 1.5519, |
|
"step": 68 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 1.6361086889890307e-05, |
|
"loss": 1.5158, |
|
"step": 69 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 1.6250762307360206e-05, |
|
"loss": 1.5203, |
|
"step": 70 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 1.6139174615714753e-05, |
|
"loss": 1.5331, |
|
"step": 71 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 1.6026346363792565e-05, |
|
"loss": 1.5399, |
|
"step": 72 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 1.591230035111576e-05, |
|
"loss": 1.5499, |
|
"step": 73 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 1.5797059623282787e-05, |
|
"loss": 1.5821, |
|
"step": 74 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 1.568064746731156e-05, |
|
"loss": 1.5463, |
|
"step": 75 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 1.5563087406933762e-05, |
|
"loss": 1.5557, |
|
"step": 76 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 1.5444403197841345e-05, |
|
"loss": 1.5475, |
|
"step": 77 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 1.5324618822886167e-05, |
|
"loss": 1.5414, |
|
"step": 78 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 1.5203758487233677e-05, |
|
"loss": 1.5411, |
|
"step": 79 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 1.5081846613471736e-05, |
|
"loss": 1.5498, |
|
"step": 80 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 1.4958907836675467e-05, |
|
"loss": 1.5434, |
|
"step": 81 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 1.4834966999429179e-05, |
|
"loss": 1.5594, |
|
"step": 82 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 1.4710049146806348e-05, |
|
"loss": 1.538, |
|
"step": 83 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 1.4584179521308703e-05, |
|
"loss": 1.5279, |
|
"step": 84 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 1.4457383557765385e-05, |
|
"loss": 1.5677, |
|
"step": 85 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 1.4329686878193271e-05, |
|
"loss": 1.5504, |
|
"step": 86 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 1.4201115286619464e-05, |
|
"loss": 1.5561, |
|
"step": 87 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 1.4071694763866988e-05, |
|
"loss": 1.5122, |
|
"step": 88 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 1.3941451462304778e-05, |
|
"loss": 1.5893, |
|
"step": 89 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 1.3810411700563005e-05, |
|
"loss": 1.5561, |
|
"step": 90 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 1.3678601958214779e-05, |
|
"loss": 1.5287, |
|
"step": 91 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 1.3546048870425356e-05, |
|
"loss": 1.5305, |
|
"step": 92 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 1.3412779222569907e-05, |
|
"loss": 1.5191, |
|
"step": 93 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 1.3278819944820893e-05, |
|
"loss": 1.5404, |
|
"step": 94 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 1.314419810670624e-05, |
|
"loss": 1.5356, |
|
"step": 95 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 1.3008940911639302e-05, |
|
"loss": 1.5509, |
|
"step": 96 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 1.2873075691421808e-05, |
|
"loss": 1.5729, |
|
"step": 97 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 1.2736629900720832e-05, |
|
"loss": 1.5401, |
|
"step": 98 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 1.2599631111520956e-05, |
|
"loss": 1.5466, |
|
"step": 99 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 1.2462107007552726e-05, |
|
"loss": 1.5614, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 1.2324085378698529e-05, |
|
"loss": 1.5177, |
|
"step": 101 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 1.2185594115376991e-05, |
|
"loss": 1.534, |
|
"step": 102 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 1.2046661202907101e-05, |
|
"loss": 1.5641, |
|
"step": 103 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 1.1907314715853138e-05, |
|
"loss": 1.5361, |
|
"step": 104 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 1.176758281235155e-05, |
|
"loss": 1.5613, |
|
"step": 105 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 1.1627493728420978e-05, |
|
"loss": 1.5139, |
|
"step": 106 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 1.1487075772256517e-05, |
|
"loss": 1.5168, |
|
"step": 107 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 1.1346357318509395e-05, |
|
"loss": 1.5248, |
|
"step": 108 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 1.1205366802553231e-05, |
|
"loss": 1.5336, |
|
"step": 109 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 1.1064132714738024e-05, |
|
"loss": 1.5257, |
|
"step": 110 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 1.092268359463302e-05, |
|
"loss": 1.5574, |
|
"step": 111 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 1.0781048025259648e-05, |
|
"loss": 1.5435, |
|
"step": 112 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 1.0639254627315658e-05, |
|
"loss": 1.5381, |
|
"step": 113 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 1.049733205339167e-05, |
|
"loss": 1.5247, |
|
"step": 114 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 1.0355308982181254e-05, |
|
"loss": 1.5423, |
|
"step": 115 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 1.0213214112685747e-05, |
|
"loss": 1.5186, |
|
"step": 116 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 1.0071076158414977e-05, |
|
"loss": 1.5536, |
|
"step": 117 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 9.928923841585025e-06, |
|
"loss": 1.5147, |
|
"step": 118 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 9.786785887314255e-06, |
|
"loss": 1.5233, |
|
"step": 119 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 9.644691017818752e-06, |
|
"loss": 1.549, |
|
"step": 120 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 9.502667946608332e-06, |
|
"loss": 1.5046, |
|
"step": 121 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 9.360745372684346e-06, |
|
"loss": 1.5089, |
|
"step": 122 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 9.218951974740354e-06, |
|
"loss": 1.5225, |
|
"step": 123 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 9.07731640536698e-06, |
|
"loss": 1.5566, |
|
"step": 124 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 8.935867285261977e-06, |
|
"loss": 1.4955, |
|
"step": 125 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 8.79463319744677e-06, |
|
"loss": 1.5357, |
|
"step": 126 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 8.653642681490608e-06, |
|
"loss": 1.5426, |
|
"step": 127 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 8.512924227743482e-06, |
|
"loss": 1.5394, |
|
"step": 128 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 8.372506271579022e-06, |
|
"loss": 1.5105, |
|
"step": 129 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 8.232417187648454e-06, |
|
"loss": 1.5238, |
|
"step": 130 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 8.092685284146865e-06, |
|
"loss": 1.5085, |
|
"step": 131 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 7.953338797092902e-06, |
|
"loss": 1.4983, |
|
"step": 132 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 7.814405884623012e-06, |
|
"loss": 1.5384, |
|
"step": 133 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 7.675914621301476e-06, |
|
"loss": 1.5294, |
|
"step": 134 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 7.5378929924472735e-06, |
|
"loss": 1.558, |
|
"step": 135 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 7.400368888479048e-06, |
|
"loss": 1.5188, |
|
"step": 136 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 7.263370099279173e-06, |
|
"loss": 1.5527, |
|
"step": 137 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 7.126924308578196e-06, |
|
"loss": 1.5368, |
|
"step": 138 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 6.9910590883607e-06, |
|
"loss": 1.5541, |
|
"step": 139 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 6.855801893293765e-06, |
|
"loss": 1.5142, |
|
"step": 140 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 6.721180055179113e-06, |
|
"loss": 1.524, |
|
"step": 141 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 6.587220777430097e-06, |
|
"loss": 1.5234, |
|
"step": 142 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 6.453951129574644e-06, |
|
"loss": 1.5979, |
|
"step": 143 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 6.321398041785225e-06, |
|
"loss": 1.588, |
|
"step": 144 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 6.189588299436997e-06, |
|
"loss": 1.4905, |
|
"step": 145 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 6.058548537695225e-06, |
|
"loss": 1.5129, |
|
"step": 146 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 5.928305236133016e-06, |
|
"loss": 1.5519, |
|
"step": 147 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 5.798884713380542e-06, |
|
"loss": 1.5366, |
|
"step": 148 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 5.67031312180673e-06, |
|
"loss": 1.5418, |
|
"step": 149 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 5.542616442234618e-06, |
|
"loss": 1.5212, |
|
"step": 150 |
|
}, |
|
{ |
|
"epoch": 0.27, |
|
"learning_rate": 5.415820478691301e-06, |
|
"loss": 1.5012, |
|
"step": 151 |
|
}, |
|
{ |
|
"epoch": 0.27, |
|
"learning_rate": 5.2899508531936526e-06, |
|
"loss": 1.542, |
|
"step": 152 |
|
}, |
|
{ |
|
"epoch": 0.27, |
|
"learning_rate": 5.165033000570825e-06, |
|
"loss": 1.5637, |
|
"step": 153 |
|
}, |
|
{ |
|
"epoch": 0.27, |
|
"learning_rate": 5.041092163324537e-06, |
|
"loss": 1.5441, |
|
"step": 154 |
|
}, |
|
{ |
|
"epoch": 0.27, |
|
"learning_rate": 4.918153386528271e-06, |
|
"loss": 1.5215, |
|
"step": 155 |
|
}, |
|
{ |
|
"epoch": 0.27, |
|
"learning_rate": 4.7962415127663265e-06, |
|
"loss": 1.5035, |
|
"step": 156 |
|
}, |
|
{ |
|
"epoch": 0.28, |
|
"learning_rate": 4.675381177113837e-06, |
|
"loss": 1.5253, |
|
"step": 157 |
|
}, |
|
{ |
|
"epoch": 0.28, |
|
"learning_rate": 4.555596802158653e-06, |
|
"loss": 1.5, |
|
"step": 158 |
|
}, |
|
{ |
|
"epoch": 0.28, |
|
"learning_rate": 4.436912593066241e-06, |
|
"loss": 1.5098, |
|
"step": 159 |
|
}, |
|
{ |
|
"epoch": 0.28, |
|
"learning_rate": 4.319352532688444e-06, |
|
"loss": 1.5607, |
|
"step": 160 |
|
}, |
|
{ |
|
"epoch": 0.28, |
|
"learning_rate": 4.2029403767172175e-06, |
|
"loss": 1.5356, |
|
"step": 161 |
|
}, |
|
{ |
|
"epoch": 0.28, |
|
"learning_rate": 4.087699648884248e-06, |
|
"loss": 1.5478, |
|
"step": 162 |
|
}, |
|
{ |
|
"epoch": 0.29, |
|
"learning_rate": 3.973653636207437e-06, |
|
"loss": 1.5812, |
|
"step": 163 |
|
}, |
|
{ |
|
"epoch": 0.29, |
|
"learning_rate": 3.860825384285247e-06, |
|
"loss": 1.4965, |
|
"step": 164 |
|
}, |
|
{ |
|
"epoch": 0.29, |
|
"learning_rate": 3.7492376926397966e-06, |
|
"loss": 1.5407, |
|
"step": 165 |
|
}, |
|
{ |
|
"epoch": 0.29, |
|
"learning_rate": 3.6389131101096953e-06, |
|
"loss": 1.4986, |
|
"step": 166 |
|
}, |
|
{ |
|
"epoch": 0.29, |
|
"learning_rate": 3.529873930293546e-06, |
|
"loss": 1.5161, |
|
"step": 167 |
|
}, |
|
{ |
|
"epoch": 0.3, |
|
"learning_rate": 3.422142187045011e-06, |
|
"loss": 1.5216, |
|
"step": 168 |
|
}, |
|
{ |
|
"epoch": 0.3, |
|
"learning_rate": 3.3157396500203655e-06, |
|
"loss": 1.5343, |
|
"step": 169 |
|
}, |
|
{ |
|
"epoch": 0.3, |
|
"learning_rate": 3.2106878202794513e-06, |
|
"loss": 1.4906, |
|
"step": 170 |
|
}, |
|
{ |
|
"epoch": 0.3, |
|
"learning_rate": 3.1070079259408934e-06, |
|
"loss": 1.5339, |
|
"step": 171 |
|
}, |
|
{ |
|
"epoch": 0.3, |
|
"learning_rate": 3.004720917892464e-06, |
|
"loss": 1.5335, |
|
"step": 172 |
|
}, |
|
{ |
|
"epoch": 0.3, |
|
"learning_rate": 2.90384746555749e-06, |
|
"loss": 1.4958, |
|
"step": 173 |
|
}, |
|
{ |
|
"epoch": 0.31, |
|
"learning_rate": 2.804407952718119e-06, |
|
"loss": 1.4891, |
|
"step": 174 |
|
}, |
|
{ |
|
"epoch": 0.31, |
|
"learning_rate": 2.7064224733963197e-06, |
|
"loss": 1.5082, |
|
"step": 175 |
|
}, |
|
{ |
|
"epoch": 0.31, |
|
"learning_rate": 2.6099108277934105e-06, |
|
"loss": 1.5197, |
|
"step": 176 |
|
}, |
|
{ |
|
"epoch": 0.31, |
|
"learning_rate": 2.514892518288988e-06, |
|
"loss": 1.5156, |
|
"step": 177 |
|
}, |
|
{ |
|
"epoch": 0.31, |
|
"learning_rate": 2.421386745500034e-06, |
|
"loss": 1.5117, |
|
"step": 178 |
|
}, |
|
{ |
|
"epoch": 0.31, |
|
"learning_rate": 2.329412404400969e-06, |
|
"loss": 1.5217, |
|
"step": 179 |
|
}, |
|
{ |
|
"epoch": 0.32, |
|
"learning_rate": 2.238988080505513e-06, |
|
"loss": 1.5377, |
|
"step": 180 |
|
}, |
|
{ |
|
"epoch": 0.32, |
|
"learning_rate": 2.150132046111054e-06, |
|
"loss": 1.5183, |
|
"step": 181 |
|
}, |
|
{ |
|
"epoch": 0.32, |
|
"learning_rate": 2.0628622566063063e-06, |
|
"loss": 1.5288, |
|
"step": 182 |
|
}, |
|
{ |
|
"epoch": 0.32, |
|
"learning_rate": 1.977196346843019e-06, |
|
"loss": 1.5509, |
|
"step": 183 |
|
}, |
|
{ |
|
"epoch": 0.32, |
|
"learning_rate": 1.8931516275724527e-06, |
|
"loss": 1.4979, |
|
"step": 184 |
|
}, |
|
{ |
|
"epoch": 0.33, |
|
"learning_rate": 1.8107450819473505e-06, |
|
"loss": 1.5506, |
|
"step": 185 |
|
}, |
|
{ |
|
"epoch": 0.33, |
|
"learning_rate": 1.7299933620900945e-06, |
|
"loss": 1.5076, |
|
"step": 186 |
|
}, |
|
{ |
|
"epoch": 0.33, |
|
"learning_rate": 1.6509127857277784e-06, |
|
"loss": 1.5278, |
|
"step": 187 |
|
}, |
|
{ |
|
"epoch": 0.33, |
|
"learning_rate": 1.573519332894824e-06, |
|
"loss": 1.5406, |
|
"step": 188 |
|
}, |
|
{ |
|
"epoch": 0.33, |
|
"learning_rate": 1.4978286427038602e-06, |
|
"loss": 1.5436, |
|
"step": 189 |
|
}, |
|
{ |
|
"epoch": 0.33, |
|
"learning_rate": 1.4238560101854815e-06, |
|
"loss": 1.4894, |
|
"step": 190 |
|
}, |
|
{ |
|
"epoch": 0.34, |
|
"learning_rate": 1.3516163831975337e-06, |
|
"loss": 1.4934, |
|
"step": 191 |
|
}, |
|
{ |
|
"epoch": 0.34, |
|
"learning_rate": 1.2811243594045697e-06, |
|
"loss": 1.495, |
|
"step": 192 |
|
}, |
|
{ |
|
"epoch": 0.34, |
|
"learning_rate": 1.2123941833280472e-06, |
|
"loss": 1.5111, |
|
"step": 193 |
|
}, |
|
{ |
|
"epoch": 0.34, |
|
"learning_rate": 1.1454397434679022e-06, |
|
"loss": 1.5202, |
|
"step": 194 |
|
}, |
|
{ |
|
"epoch": 0.34, |
|
"learning_rate": 1.080274569496057e-06, |
|
"loss": 1.5215, |
|
"step": 195 |
|
}, |
|
{ |
|
"epoch": 0.34, |
|
"learning_rate": 1.0169118295224488e-06, |
|
"loss": 1.5132, |
|
"step": 196 |
|
}, |
|
{ |
|
"epoch": 0.35, |
|
"learning_rate": 9.55364327434105e-07, |
|
"loss": 1.5496, |
|
"step": 197 |
|
}, |
|
{ |
|
"epoch": 0.35, |
|
"learning_rate": 8.956445003078351e-07, |
|
"loss": 1.5646, |
|
"step": 198 |
|
}, |
|
{ |
|
"epoch": 0.35, |
|
"learning_rate": 8.377644158970277e-07, |
|
"loss": 1.5008, |
|
"step": 199 |
|
}, |
|
{ |
|
"epoch": 0.35, |
|
"learning_rate": 7.81735770193085e-07, |
|
"loss": 1.5344, |
|
"step": 200 |
|
}, |
|
{ |
|
"epoch": 0.35, |
|
"learning_rate": 7.275698850619861e-07, |
|
"loss": 1.51, |
|
"step": 201 |
|
}, |
|
{ |
|
"epoch": 0.36, |
|
"learning_rate": 6.752777059564431e-07, |
|
"loss": 1.5111, |
|
"step": 202 |
|
}, |
|
{ |
|
"epoch": 0.36, |
|
"learning_rate": 6.248697997041219e-07, |
|
"loss": 1.5183, |
|
"step": 203 |
|
}, |
|
{ |
|
"epoch": 0.36, |
|
"learning_rate": 5.763563523723769e-07, |
|
"loss": 1.5033, |
|
"step": 204 |
|
}, |
|
{ |
|
"epoch": 0.36, |
|
"learning_rate": 5.29747167209923e-07, |
|
"loss": 1.5569, |
|
"step": 205 |
|
}, |
|
{ |
|
"epoch": 0.36, |
|
"learning_rate": 4.850516626658585e-07, |
|
"loss": 1.5635, |
|
"step": 206 |
|
}, |
|
{ |
|
"epoch": 0.36, |
|
"learning_rate": 4.4227887048646335e-07, |
|
"loss": 1.5331, |
|
"step": 207 |
|
}, |
|
{ |
|
"epoch": 0.37, |
|
"learning_rate": 4.014374338901206e-07, |
|
"loss": 1.5119, |
|
"step": 208 |
|
}, |
|
{ |
|
"epoch": 0.37, |
|
"learning_rate": 3.6253560582076075e-07, |
|
"loss": 1.5153, |
|
"step": 209 |
|
}, |
|
{ |
|
"epoch": 0.37, |
|
"learning_rate": 3.255812472801689e-07, |
|
"loss": 1.5168, |
|
"step": 210 |
|
}, |
|
{ |
|
"epoch": 0.37, |
|
"learning_rate": 2.905818257394799e-07, |
|
"loss": 1.5449, |
|
"step": 211 |
|
}, |
|
{ |
|
"epoch": 0.37, |
|
"learning_rate": 2.5754441363021854e-07, |
|
"loss": 1.5279, |
|
"step": 212 |
|
}, |
|
{ |
|
"epoch": 0.37, |
|
"learning_rate": 2.264756869151441e-07, |
|
"loss": 1.5379, |
|
"step": 213 |
|
}, |
|
{ |
|
"epoch": 0.38, |
|
"learning_rate": 1.973819237392205e-07, |
|
"loss": 1.5277, |
|
"step": 214 |
|
}, |
|
{ |
|
"epoch": 0.38, |
|
"learning_rate": 1.7026900316098217e-07, |
|
"loss": 1.5191, |
|
"step": 215 |
|
}, |
|
{ |
|
"epoch": 0.38, |
|
"learning_rate": 1.4514240396452438e-07, |
|
"loss": 1.4924, |
|
"step": 216 |
|
}, |
|
{ |
|
"epoch": 0.38, |
|
"learning_rate": 1.2200720355239893e-07, |
|
"loss": 1.5482, |
|
"step": 217 |
|
}, |
|
{ |
|
"epoch": 0.38, |
|
"learning_rate": 1.0086807691960243e-07, |
|
"loss": 1.5033, |
|
"step": 218 |
|
}, |
|
{ |
|
"epoch": 0.38, |
|
"learning_rate": 8.172929570889553e-08, |
|
"loss": 1.5469, |
|
"step": 219 |
|
}, |
|
{ |
|
"epoch": 0.39, |
|
"learning_rate": 6.459472734760997e-08, |
|
"loss": 1.5517, |
|
"step": 220 |
|
}, |
|
{ |
|
"epoch": 0.39, |
|
"learning_rate": 4.9467834266154756e-08, |
|
"loss": 1.5006, |
|
"step": 221 |
|
}, |
|
{ |
|
"epoch": 0.39, |
|
"learning_rate": 3.635167319834709e-08, |
|
"loss": 1.5133, |
|
"step": 222 |
|
}, |
|
{ |
|
"epoch": 0.39, |
|
"learning_rate": 2.524889456373525e-08, |
|
"loss": 1.5053, |
|
"step": 223 |
|
}, |
|
{ |
|
"epoch": 0.39, |
|
"learning_rate": 1.6161741932017026e-08, |
|
"loss": 1.4884, |
|
"step": 224 |
|
}, |
|
{ |
|
"epoch": 0.4, |
|
"learning_rate": 9.092051569674632e-09, |
|
"loss": 1.4829, |
|
"step": 225 |
|
}, |
|
{ |
|
"epoch": 0.4, |
|
"learning_rate": 4.041252068918145e-09, |
|
"loss": 1.5512, |
|
"step": 226 |
|
}, |
|
{ |
|
"epoch": 0.4, |
|
"learning_rate": 1.0103640590064524e-09, |
|
"loss": 1.5033, |
|
"step": 227 |
|
}, |
|
{ |
|
"epoch": 0.4, |
|
"learning_rate": 0.0, |
|
"loss": 1.514, |
|
"step": 228 |
|
}, |
|
{ |
|
"epoch": 0.4, |
|
"step": 228, |
|
"total_flos": 2.7771263153025843e+17, |
|
"train_loss": 1.567653910632719, |
|
"train_runtime": 13699.6849, |
|
"train_samples_per_second": 17.011, |
|
"train_steps_per_second": 0.017 |
|
} |
|
], |
|
"logging_steps": 1.0, |
|
"max_steps": 228, |
|
"num_input_tokens_seen": 0, |
|
"num_train_epochs": 1, |
|
"save_steps": 50000, |
|
"total_flos": 2.7771263153025843e+17, |
|
"train_batch_size": 2, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|