|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 5.0, |
|
"eval_steps": 500, |
|
"global_step": 160, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 4e-05, |
|
"loss": 3.9159, |
|
"step": 1 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 8e-05, |
|
"loss": 3.7411, |
|
"step": 2 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 0.00012, |
|
"loss": 2.6559, |
|
"step": 3 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 0.00016, |
|
"loss": 1.0552, |
|
"step": 4 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 0.0002, |
|
"loss": 0.4381, |
|
"step": 5 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 0.00019997946042345127, |
|
"loss": 0.7843, |
|
"step": 6 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 0.00019991785013128923, |
|
"loss": 0.4278, |
|
"step": 7 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 0.0001998151944325001, |
|
"loss": 0.3102, |
|
"step": 8 |
|
}, |
|
{ |
|
"epoch": 0.28, |
|
"learning_rate": 0.00019967153549717553, |
|
"loss": 0.184, |
|
"step": 9 |
|
}, |
|
{ |
|
"epoch": 0.31, |
|
"learning_rate": 0.00019948693233918952, |
|
"loss": 0.2193, |
|
"step": 10 |
|
}, |
|
{ |
|
"epoch": 0.34, |
|
"learning_rate": 0.00019926146079195594, |
|
"loss": 0.1322, |
|
"step": 11 |
|
}, |
|
{ |
|
"epoch": 0.38, |
|
"learning_rate": 0.0001989952134772769, |
|
"loss": 0.1299, |
|
"step": 12 |
|
}, |
|
{ |
|
"epoch": 0.41, |
|
"learning_rate": 0.00019868829976729443, |
|
"loss": 0.1289, |
|
"step": 13 |
|
}, |
|
{ |
|
"epoch": 0.44, |
|
"learning_rate": 0.00019834084573956128, |
|
"loss": 0.1078, |
|
"step": 14 |
|
}, |
|
{ |
|
"epoch": 0.47, |
|
"learning_rate": 0.00019795299412524945, |
|
"loss": 0.0919, |
|
"step": 15 |
|
}, |
|
{ |
|
"epoch": 0.5, |
|
"learning_rate": 0.00019752490425051743, |
|
"loss": 0.0843, |
|
"step": 16 |
|
}, |
|
{ |
|
"epoch": 0.53, |
|
"learning_rate": 0.00019705675197106016, |
|
"loss": 0.0998, |
|
"step": 17 |
|
}, |
|
{ |
|
"epoch": 0.56, |
|
"learning_rate": 0.00019654872959986937, |
|
"loss": 0.0729, |
|
"step": 18 |
|
}, |
|
{ |
|
"epoch": 0.59, |
|
"learning_rate": 0.0001960010458282326, |
|
"loss": 0.111, |
|
"step": 19 |
|
}, |
|
{ |
|
"epoch": 0.62, |
|
"learning_rate": 0.00019541392564000488, |
|
"loss": 0.0915, |
|
"step": 20 |
|
}, |
|
{ |
|
"epoch": 0.66, |
|
"learning_rate": 0.00019478761021918728, |
|
"loss": 0.0905, |
|
"step": 21 |
|
}, |
|
{ |
|
"epoch": 0.69, |
|
"learning_rate": 0.00019412235685085035, |
|
"loss": 0.0977, |
|
"step": 22 |
|
}, |
|
{ |
|
"epoch": 0.72, |
|
"learning_rate": 0.00019341843881544372, |
|
"loss": 0.0709, |
|
"step": 23 |
|
}, |
|
{ |
|
"epoch": 0.75, |
|
"learning_rate": 0.00019267614527653488, |
|
"loss": 0.0959, |
|
"step": 24 |
|
}, |
|
{ |
|
"epoch": 0.78, |
|
"learning_rate": 0.00019189578116202307, |
|
"loss": 0.0722, |
|
"step": 25 |
|
}, |
|
{ |
|
"epoch": 0.81, |
|
"learning_rate": 0.00019107766703887764, |
|
"loss": 0.0737, |
|
"step": 26 |
|
}, |
|
{ |
|
"epoch": 0.84, |
|
"learning_rate": 0.00019022213898145176, |
|
"loss": 0.0744, |
|
"step": 27 |
|
}, |
|
{ |
|
"epoch": 0.88, |
|
"learning_rate": 0.00018932954843342591, |
|
"loss": 0.0663, |
|
"step": 28 |
|
}, |
|
{ |
|
"epoch": 0.91, |
|
"learning_rate": 0.00018840026206343784, |
|
"loss": 0.0887, |
|
"step": 29 |
|
}, |
|
{ |
|
"epoch": 0.94, |
|
"learning_rate": 0.00018743466161445823, |
|
"loss": 0.0816, |
|
"step": 30 |
|
}, |
|
{ |
|
"epoch": 0.97, |
|
"learning_rate": 0.00018643314374697378, |
|
"loss": 0.0705, |
|
"step": 31 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 0.00018539611987604258, |
|
"loss": 0.0529, |
|
"step": 32 |
|
}, |
|
{ |
|
"epoch": 1.03, |
|
"learning_rate": 0.00018432401600228823, |
|
"loss": 0.0753, |
|
"step": 33 |
|
}, |
|
{ |
|
"epoch": 1.06, |
|
"learning_rate": 0.0001832172725369024, |
|
"loss": 0.0514, |
|
"step": 34 |
|
}, |
|
{ |
|
"epoch": 1.09, |
|
"learning_rate": 0.00018207634412072764, |
|
"loss": 0.0459, |
|
"step": 35 |
|
}, |
|
{ |
|
"epoch": 1.12, |
|
"learning_rate": 0.00018090169943749476, |
|
"loss": 0.0523, |
|
"step": 36 |
|
}, |
|
{ |
|
"epoch": 1.16, |
|
"learning_rate": 0.0001796938210212915, |
|
"loss": 0.0377, |
|
"step": 37 |
|
}, |
|
{ |
|
"epoch": 1.19, |
|
"learning_rate": 0.00017845320505834175, |
|
"loss": 0.0499, |
|
"step": 38 |
|
}, |
|
{ |
|
"epoch": 1.22, |
|
"learning_rate": 0.0001771803611831762, |
|
"loss": 0.0486, |
|
"step": 39 |
|
}, |
|
{ |
|
"epoch": 1.25, |
|
"learning_rate": 0.0001758758122692791, |
|
"loss": 0.0348, |
|
"step": 40 |
|
}, |
|
{ |
|
"epoch": 1.28, |
|
"learning_rate": 0.00017454009421429597, |
|
"loss": 0.0407, |
|
"step": 41 |
|
}, |
|
{ |
|
"epoch": 1.31, |
|
"learning_rate": 0.00017317375571989158, |
|
"loss": 0.0544, |
|
"step": 42 |
|
}, |
|
{ |
|
"epoch": 1.34, |
|
"learning_rate": 0.00017177735806634789, |
|
"loss": 0.0317, |
|
"step": 43 |
|
}, |
|
{ |
|
"epoch": 1.38, |
|
"learning_rate": 0.00017035147488199482, |
|
"loss": 0.0504, |
|
"step": 44 |
|
}, |
|
{ |
|
"epoch": 1.41, |
|
"learning_rate": 0.00016889669190756868, |
|
"loss": 0.0592, |
|
"step": 45 |
|
}, |
|
{ |
|
"epoch": 1.44, |
|
"learning_rate": 0.00016741360675559473, |
|
"loss": 0.047, |
|
"step": 46 |
|
}, |
|
{ |
|
"epoch": 1.47, |
|
"learning_rate": 0.00016590282866489319, |
|
"loss": 0.0409, |
|
"step": 47 |
|
}, |
|
{ |
|
"epoch": 1.5, |
|
"learning_rate": 0.00016436497825030884, |
|
"loss": 0.0341, |
|
"step": 48 |
|
}, |
|
{ |
|
"epoch": 1.53, |
|
"learning_rate": 0.00016280068724776797, |
|
"loss": 0.0783, |
|
"step": 49 |
|
}, |
|
{ |
|
"epoch": 1.56, |
|
"learning_rate": 0.0001612105982547663, |
|
"loss": 0.0318, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 1.59, |
|
"learning_rate": 0.0001595953644663957, |
|
"loss": 0.05, |
|
"step": 51 |
|
}, |
|
{ |
|
"epoch": 1.62, |
|
"learning_rate": 0.000157955649407017, |
|
"loss": 0.0659, |
|
"step": 52 |
|
}, |
|
{ |
|
"epoch": 1.66, |
|
"learning_rate": 0.00015629212665768978, |
|
"loss": 0.0594, |
|
"step": 53 |
|
}, |
|
{ |
|
"epoch": 1.69, |
|
"learning_rate": 0.00015460547957947104, |
|
"loss": 0.0279, |
|
"step": 54 |
|
}, |
|
{ |
|
"epoch": 1.72, |
|
"learning_rate": 0.00015289640103269625, |
|
"loss": 0.0731, |
|
"step": 55 |
|
}, |
|
{ |
|
"epoch": 1.75, |
|
"learning_rate": 0.00015116559309235825, |
|
"loss": 0.047, |
|
"step": 56 |
|
}, |
|
{ |
|
"epoch": 1.78, |
|
"learning_rate": 0.0001494137667597006, |
|
"loss": 0.0523, |
|
"step": 57 |
|
}, |
|
{ |
|
"epoch": 1.81, |
|
"learning_rate": 0.00014764164167014451, |
|
"loss": 0.0604, |
|
"step": 58 |
|
}, |
|
{ |
|
"epoch": 1.84, |
|
"learning_rate": 0.00014584994579766865, |
|
"loss": 0.048, |
|
"step": 59 |
|
}, |
|
{ |
|
"epoch": 1.88, |
|
"learning_rate": 0.00014403941515576344, |
|
"loss": 0.0544, |
|
"step": 60 |
|
}, |
|
{ |
|
"epoch": 1.91, |
|
"learning_rate": 0.0001422107934950832, |
|
"loss": 0.0493, |
|
"step": 61 |
|
}, |
|
{ |
|
"epoch": 1.94, |
|
"learning_rate": 0.00014036483199791948, |
|
"loss": 0.0699, |
|
"step": 62 |
|
}, |
|
{ |
|
"epoch": 1.97, |
|
"learning_rate": 0.0001385022889696218, |
|
"loss": 0.0594, |
|
"step": 63 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"learning_rate": 0.00013662392952709228, |
|
"loss": 0.0531, |
|
"step": 64 |
|
}, |
|
{ |
|
"epoch": 2.03, |
|
"learning_rate": 0.00013473052528448201, |
|
"loss": 0.0336, |
|
"step": 65 |
|
}, |
|
{ |
|
"epoch": 2.06, |
|
"learning_rate": 0.00013282285403621864, |
|
"loss": 0.0474, |
|
"step": 66 |
|
}, |
|
{ |
|
"epoch": 2.09, |
|
"learning_rate": 0.00013090169943749476, |
|
"loss": 0.0288, |
|
"step": 67 |
|
}, |
|
{ |
|
"epoch": 2.12, |
|
"learning_rate": 0.00012896785068234926, |
|
"loss": 0.0277, |
|
"step": 68 |
|
}, |
|
{ |
|
"epoch": 2.16, |
|
"learning_rate": 0.00012702210217947288, |
|
"loss": 0.0386, |
|
"step": 69 |
|
}, |
|
{ |
|
"epoch": 2.19, |
|
"learning_rate": 0.00012506525322587207, |
|
"loss": 0.023, |
|
"step": 70 |
|
}, |
|
{ |
|
"epoch": 2.22, |
|
"learning_rate": 0.00012309810767852433, |
|
"loss": 0.0342, |
|
"step": 71 |
|
}, |
|
{ |
|
"epoch": 2.25, |
|
"learning_rate": 0.00012112147362416076, |
|
"loss": 0.023, |
|
"step": 72 |
|
}, |
|
{ |
|
"epoch": 2.28, |
|
"learning_rate": 0.00011913616304731063, |
|
"loss": 0.0242, |
|
"step": 73 |
|
}, |
|
{ |
|
"epoch": 2.31, |
|
"learning_rate": 0.00011714299149674537, |
|
"loss": 0.0322, |
|
"step": 74 |
|
}, |
|
{ |
|
"epoch": 2.34, |
|
"learning_rate": 0.00011514277775045768, |
|
"loss": 0.0404, |
|
"step": 75 |
|
}, |
|
{ |
|
"epoch": 2.38, |
|
"learning_rate": 0.00011313634347931466, |
|
"loss": 0.027, |
|
"step": 76 |
|
}, |
|
{ |
|
"epoch": 2.41, |
|
"learning_rate": 0.00011112451290952237, |
|
"loss": 0.0282, |
|
"step": 77 |
|
}, |
|
{ |
|
"epoch": 2.44, |
|
"learning_rate": 0.00010910811248404065, |
|
"loss": 0.0295, |
|
"step": 78 |
|
}, |
|
{ |
|
"epoch": 2.47, |
|
"learning_rate": 0.0001070879705230873, |
|
"loss": 0.0315, |
|
"step": 79 |
|
}, |
|
{ |
|
"epoch": 2.5, |
|
"learning_rate": 0.00010506491688387127, |
|
"loss": 0.0301, |
|
"step": 80 |
|
}, |
|
{ |
|
"epoch": 2.53, |
|
"learning_rate": 0.0001030397826196943, |
|
"loss": 0.0251, |
|
"step": 81 |
|
}, |
|
{ |
|
"epoch": 2.56, |
|
"learning_rate": 0.00010101339963856111, |
|
"loss": 0.0186, |
|
"step": 82 |
|
}, |
|
{ |
|
"epoch": 2.59, |
|
"learning_rate": 9.898660036143893e-05, |
|
"loss": 0.0238, |
|
"step": 83 |
|
}, |
|
{ |
|
"epoch": 2.62, |
|
"learning_rate": 9.696021738030575e-05, |
|
"loss": 0.0188, |
|
"step": 84 |
|
}, |
|
{ |
|
"epoch": 2.66, |
|
"learning_rate": 9.493508311612874e-05, |
|
"loss": 0.0161, |
|
"step": 85 |
|
}, |
|
{ |
|
"epoch": 2.69, |
|
"learning_rate": 9.291202947691271e-05, |
|
"loss": 0.0227, |
|
"step": 86 |
|
}, |
|
{ |
|
"epoch": 2.72, |
|
"learning_rate": 9.089188751595936e-05, |
|
"loss": 0.0276, |
|
"step": 87 |
|
}, |
|
{ |
|
"epoch": 2.75, |
|
"learning_rate": 8.887548709047764e-05, |
|
"loss": 0.035, |
|
"step": 88 |
|
}, |
|
{ |
|
"epoch": 2.78, |
|
"learning_rate": 8.686365652068535e-05, |
|
"loss": 0.0165, |
|
"step": 89 |
|
}, |
|
{ |
|
"epoch": 2.81, |
|
"learning_rate": 8.485722224954237e-05, |
|
"loss": 0.0225, |
|
"step": 90 |
|
}, |
|
{ |
|
"epoch": 2.84, |
|
"learning_rate": 8.285700850325467e-05, |
|
"loss": 0.0171, |
|
"step": 91 |
|
}, |
|
{ |
|
"epoch": 2.88, |
|
"learning_rate": 8.086383695268938e-05, |
|
"loss": 0.031, |
|
"step": 92 |
|
}, |
|
{ |
|
"epoch": 2.91, |
|
"learning_rate": 7.887852637583926e-05, |
|
"loss": 0.0291, |
|
"step": 93 |
|
}, |
|
{ |
|
"epoch": 2.94, |
|
"learning_rate": 7.690189232147566e-05, |
|
"loss": 0.0346, |
|
"step": 94 |
|
}, |
|
{ |
|
"epoch": 2.97, |
|
"learning_rate": 7.493474677412794e-05, |
|
"loss": 0.0327, |
|
"step": 95 |
|
}, |
|
{ |
|
"epoch": 3.0, |
|
"learning_rate": 7.297789782052717e-05, |
|
"loss": 0.0151, |
|
"step": 96 |
|
}, |
|
{ |
|
"epoch": 3.03, |
|
"learning_rate": 7.10321493176508e-05, |
|
"loss": 0.0179, |
|
"step": 97 |
|
}, |
|
{ |
|
"epoch": 3.06, |
|
"learning_rate": 6.909830056250527e-05, |
|
"loss": 0.0101, |
|
"step": 98 |
|
}, |
|
{ |
|
"epoch": 3.09, |
|
"learning_rate": 6.717714596378137e-05, |
|
"loss": 0.0207, |
|
"step": 99 |
|
}, |
|
{ |
|
"epoch": 3.12, |
|
"learning_rate": 6.526947471551798e-05, |
|
"loss": 0.0093, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 3.16, |
|
"learning_rate": 6.337607047290774e-05, |
|
"loss": 0.0135, |
|
"step": 101 |
|
}, |
|
{ |
|
"epoch": 3.19, |
|
"learning_rate": 6.149771103037821e-05, |
|
"loss": 0.0149, |
|
"step": 102 |
|
}, |
|
{ |
|
"epoch": 3.22, |
|
"learning_rate": 5.9635168002080564e-05, |
|
"loss": 0.0147, |
|
"step": 103 |
|
}, |
|
{ |
|
"epoch": 3.25, |
|
"learning_rate": 5.7789206504916816e-05, |
|
"loss": 0.0109, |
|
"step": 104 |
|
}, |
|
{ |
|
"epoch": 3.28, |
|
"learning_rate": 5.596058484423656e-05, |
|
"loss": 0.0107, |
|
"step": 105 |
|
}, |
|
{ |
|
"epoch": 3.31, |
|
"learning_rate": 5.415005420233141e-05, |
|
"loss": 0.0099, |
|
"step": 106 |
|
}, |
|
{ |
|
"epoch": 3.34, |
|
"learning_rate": 5.2358358329855516e-05, |
|
"loss": 0.0172, |
|
"step": 107 |
|
}, |
|
{ |
|
"epoch": 3.38, |
|
"learning_rate": 5.058623324029944e-05, |
|
"loss": 0.0173, |
|
"step": 108 |
|
}, |
|
{ |
|
"epoch": 3.41, |
|
"learning_rate": 4.8834406907641784e-05, |
|
"loss": 0.0126, |
|
"step": 109 |
|
}, |
|
{ |
|
"epoch": 3.44, |
|
"learning_rate": 4.710359896730379e-05, |
|
"loss": 0.0144, |
|
"step": 110 |
|
}, |
|
{ |
|
"epoch": 3.47, |
|
"learning_rate": 4.539452042052901e-05, |
|
"loss": 0.012, |
|
"step": 111 |
|
}, |
|
{ |
|
"epoch": 3.5, |
|
"learning_rate": 4.3707873342310254e-05, |
|
"loss": 0.0146, |
|
"step": 112 |
|
}, |
|
{ |
|
"epoch": 3.53, |
|
"learning_rate": 4.204435059298303e-05, |
|
"loss": 0.0133, |
|
"step": 113 |
|
}, |
|
{ |
|
"epoch": 3.56, |
|
"learning_rate": 4.040463553360431e-05, |
|
"loss": 0.0065, |
|
"step": 114 |
|
}, |
|
{ |
|
"epoch": 3.59, |
|
"learning_rate": 3.878940174523371e-05, |
|
"loss": 0.0144, |
|
"step": 115 |
|
}, |
|
{ |
|
"epoch": 3.62, |
|
"learning_rate": 3.719931275223205e-05, |
|
"loss": 0.0066, |
|
"step": 116 |
|
}, |
|
{ |
|
"epoch": 3.66, |
|
"learning_rate": 3.5635021749691166e-05, |
|
"loss": 0.0168, |
|
"step": 117 |
|
}, |
|
{ |
|
"epoch": 3.69, |
|
"learning_rate": 3.4097171335106824e-05, |
|
"loss": 0.0151, |
|
"step": 118 |
|
}, |
|
{ |
|
"epoch": 3.72, |
|
"learning_rate": 3.258639324440527e-05, |
|
"loss": 0.0151, |
|
"step": 119 |
|
}, |
|
{ |
|
"epoch": 3.75, |
|
"learning_rate": 3.110330809243134e-05, |
|
"loss": 0.0112, |
|
"step": 120 |
|
}, |
|
{ |
|
"epoch": 3.78, |
|
"learning_rate": 2.964852511800519e-05, |
|
"loss": 0.0287, |
|
"step": 121 |
|
}, |
|
{ |
|
"epoch": 3.81, |
|
"learning_rate": 2.8222641933652117e-05, |
|
"loss": 0.0103, |
|
"step": 122 |
|
}, |
|
{ |
|
"epoch": 3.84, |
|
"learning_rate": 2.6826244280108437e-05, |
|
"loss": 0.0167, |
|
"step": 123 |
|
}, |
|
{ |
|
"epoch": 3.88, |
|
"learning_rate": 2.5459905785704042e-05, |
|
"loss": 0.0073, |
|
"step": 124 |
|
}, |
|
{ |
|
"epoch": 3.91, |
|
"learning_rate": 2.4124187730720917e-05, |
|
"loss": 0.0065, |
|
"step": 125 |
|
}, |
|
{ |
|
"epoch": 3.94, |
|
"learning_rate": 2.2819638816823797e-05, |
|
"loss": 0.0132, |
|
"step": 126 |
|
}, |
|
{ |
|
"epoch": 3.97, |
|
"learning_rate": 2.154679494165829e-05, |
|
"loss": 0.0032, |
|
"step": 127 |
|
}, |
|
{ |
|
"epoch": 4.0, |
|
"learning_rate": 2.0306178978708514e-05, |
|
"loss": 0.0041, |
|
"step": 128 |
|
}, |
|
{ |
|
"epoch": 4.03, |
|
"learning_rate": 1.9098300562505266e-05, |
|
"loss": 0.0031, |
|
"step": 129 |
|
}, |
|
{ |
|
"epoch": 4.06, |
|
"learning_rate": 1.7923655879272393e-05, |
|
"loss": 0.0034, |
|
"step": 130 |
|
}, |
|
{ |
|
"epoch": 4.09, |
|
"learning_rate": 1.6782727463097624e-05, |
|
"loss": 0.0091, |
|
"step": 131 |
|
}, |
|
{ |
|
"epoch": 4.12, |
|
"learning_rate": 1.5675983997711795e-05, |
|
"loss": 0.0089, |
|
"step": 132 |
|
}, |
|
{ |
|
"epoch": 4.16, |
|
"learning_rate": 1.4603880123957447e-05, |
|
"loss": 0.0129, |
|
"step": 133 |
|
}, |
|
{ |
|
"epoch": 4.19, |
|
"learning_rate": 1.356685625302625e-05, |
|
"loss": 0.0129, |
|
"step": 134 |
|
}, |
|
{ |
|
"epoch": 4.22, |
|
"learning_rate": 1.2565338385541792e-05, |
|
"loss": 0.0032, |
|
"step": 135 |
|
}, |
|
{ |
|
"epoch": 4.25, |
|
"learning_rate": 1.1599737936562149e-05, |
|
"loss": 0.0118, |
|
"step": 136 |
|
}, |
|
{ |
|
"epoch": 4.28, |
|
"learning_rate": 1.0670451566574102e-05, |
|
"loss": 0.0062, |
|
"step": 137 |
|
}, |
|
{ |
|
"epoch": 4.31, |
|
"learning_rate": 9.777861018548251e-06, |
|
"loss": 0.0068, |
|
"step": 138 |
|
}, |
|
{ |
|
"epoch": 4.34, |
|
"learning_rate": 8.92233296112236e-06, |
|
"loss": 0.0086, |
|
"step": 139 |
|
}, |
|
{ |
|
"epoch": 4.38, |
|
"learning_rate": 8.10421883797694e-06, |
|
"loss": 0.0049, |
|
"step": 140 |
|
}, |
|
{ |
|
"epoch": 4.41, |
|
"learning_rate": 7.32385472346514e-06, |
|
"loss": 0.0103, |
|
"step": 141 |
|
}, |
|
{ |
|
"epoch": 4.44, |
|
"learning_rate": 6.581561184556295e-06, |
|
"loss": 0.0086, |
|
"step": 142 |
|
}, |
|
{ |
|
"epoch": 4.47, |
|
"learning_rate": 5.87764314914967e-06, |
|
"loss": 0.0053, |
|
"step": 143 |
|
}, |
|
{ |
|
"epoch": 4.5, |
|
"learning_rate": 5.212389780812732e-06, |
|
"loss": 0.007, |
|
"step": 144 |
|
}, |
|
{ |
|
"epoch": 4.53, |
|
"learning_rate": 4.586074359995119e-06, |
|
"loss": 0.0064, |
|
"step": 145 |
|
}, |
|
{ |
|
"epoch": 4.56, |
|
"learning_rate": 3.998954171767422e-06, |
|
"loss": 0.0023, |
|
"step": 146 |
|
}, |
|
{ |
|
"epoch": 4.59, |
|
"learning_rate": 3.451270400130646e-06, |
|
"loss": 0.0045, |
|
"step": 147 |
|
}, |
|
{ |
|
"epoch": 4.62, |
|
"learning_rate": 2.943248028939838e-06, |
|
"loss": 0.0035, |
|
"step": 148 |
|
}, |
|
{ |
|
"epoch": 4.66, |
|
"learning_rate": 2.4750957494826033e-06, |
|
"loss": 0.005, |
|
"step": 149 |
|
}, |
|
{ |
|
"epoch": 4.69, |
|
"learning_rate": 2.0470058747505516e-06, |
|
"loss": 0.0124, |
|
"step": 150 |
|
}, |
|
{ |
|
"epoch": 4.72, |
|
"learning_rate": 1.6591542604387445e-06, |
|
"loss": 0.0027, |
|
"step": 151 |
|
}, |
|
{ |
|
"epoch": 4.75, |
|
"learning_rate": 1.3117002327055927e-06, |
|
"loss": 0.0055, |
|
"step": 152 |
|
}, |
|
{ |
|
"epoch": 4.78, |
|
"learning_rate": 1.004786522723089e-06, |
|
"loss": 0.002, |
|
"step": 153 |
|
}, |
|
{ |
|
"epoch": 4.81, |
|
"learning_rate": 7.385392080440534e-07, |
|
"loss": 0.0038, |
|
"step": 154 |
|
}, |
|
{ |
|
"epoch": 4.84, |
|
"learning_rate": 5.130676608104845e-07, |
|
"loss": 0.013, |
|
"step": 155 |
|
}, |
|
{ |
|
"epoch": 4.88, |
|
"learning_rate": 3.2846450282447703e-07, |
|
"loss": 0.0073, |
|
"step": 156 |
|
}, |
|
{ |
|
"epoch": 4.91, |
|
"learning_rate": 1.8480556749991274e-07, |
|
"loss": 0.0113, |
|
"step": 157 |
|
}, |
|
{ |
|
"epoch": 4.94, |
|
"learning_rate": 8.214986871076802e-08, |
|
"loss": 0.0066, |
|
"step": 158 |
|
}, |
|
{ |
|
"epoch": 4.97, |
|
"learning_rate": 2.0539576548717076e-08, |
|
"loss": 0.0031, |
|
"step": 159 |
|
}, |
|
{ |
|
"epoch": 5.0, |
|
"learning_rate": 0.0, |
|
"loss": 0.003, |
|
"step": 160 |
|
}, |
|
{ |
|
"epoch": 5.0, |
|
"step": 160, |
|
"total_flos": 32634127908864.0, |
|
"train_loss": 0.11789017441187752, |
|
"train_runtime": 1826.6084, |
|
"train_samples_per_second": 10.949, |
|
"train_steps_per_second": 0.088 |
|
} |
|
], |
|
"logging_steps": 1.0, |
|
"max_steps": 160, |
|
"num_input_tokens_seen": 0, |
|
"num_train_epochs": 5, |
|
"save_steps": 72, |
|
"total_flos": 32634127908864.0, |
|
"train_batch_size": 16, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|