2057 lines
54 KiB
JSON
2057 lines
54 KiB
JSON
{
|
|
"best_global_step": null,
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 1.0080685829551186,
|
|
"eval_steps": 500,
|
|
"global_step": 1000,
|
|
"is_hyper_param_search": false,
|
|
"is_local_process_zero": true,
|
|
"is_world_process_zero": true,
|
|
"log_history": [
|
|
{
|
|
"entropy": 10.742588329315186,
|
|
"epoch": 0.005042864346949067,
|
|
"grad_norm": 4.78125,
|
|
"learning_rate": 2e-06,
|
|
"loss": 10.7888,
|
|
"mean_token_accuracy": 6.958942394703626e-05,
|
|
"num_tokens": 11445.0,
|
|
"step": 5
|
|
},
|
|
{
|
|
"entropy": 10.742593383789062,
|
|
"epoch": 0.010085728693898134,
|
|
"grad_norm": 5.25,
|
|
"learning_rate": 4.5e-06,
|
|
"loss": 10.7696,
|
|
"mean_token_accuracy": 0.0,
|
|
"num_tokens": 22095.0,
|
|
"step": 10
|
|
},
|
|
{
|
|
"entropy": 10.742596626281738,
|
|
"epoch": 0.015128593040847202,
|
|
"grad_norm": 4.5625,
|
|
"learning_rate": 7e-06,
|
|
"loss": 10.7579,
|
|
"mean_token_accuracy": 0.00017376195173710585,
|
|
"num_tokens": 33847.0,
|
|
"step": 15
|
|
},
|
|
{
|
|
"entropy": 10.742617321014404,
|
|
"epoch": 0.020171457387796268,
|
|
"grad_norm": 4.53125,
|
|
"learning_rate": 9.5e-06,
|
|
"loss": 10.7022,
|
|
"mean_token_accuracy": 0.0003061072085984051,
|
|
"num_tokens": 44495.0,
|
|
"step": 20
|
|
},
|
|
{
|
|
"entropy": 10.742618179321289,
|
|
"epoch": 0.025214321734745335,
|
|
"grad_norm": 4.0625,
|
|
"learning_rate": 1.2e-05,
|
|
"loss": 10.6161,
|
|
"mean_token_accuracy": 0.0038310963835101576,
|
|
"num_tokens": 55298.0,
|
|
"step": 25
|
|
},
|
|
{
|
|
"entropy": 10.742384719848634,
|
|
"epoch": 0.030257186081694403,
|
|
"grad_norm": 3.859375,
|
|
"learning_rate": 1.4500000000000002e-05,
|
|
"loss": 10.4928,
|
|
"mean_token_accuracy": 0.023839639127254487,
|
|
"num_tokens": 66391.0,
|
|
"step": 30
|
|
},
|
|
{
|
|
"entropy": 10.741421985626221,
|
|
"epoch": 0.03530005042864347,
|
|
"grad_norm": 3.125,
|
|
"learning_rate": 1.7000000000000003e-05,
|
|
"loss": 10.3446,
|
|
"mean_token_accuracy": 0.037043477036058904,
|
|
"num_tokens": 77624.0,
|
|
"step": 35
|
|
},
|
|
{
|
|
"entropy": 10.738873195648193,
|
|
"epoch": 0.040342914775592535,
|
|
"grad_norm": 2.578125,
|
|
"learning_rate": 1.95e-05,
|
|
"loss": 10.2004,
|
|
"mean_token_accuracy": 0.04290186390280724,
|
|
"num_tokens": 89426.0,
|
|
"step": 40
|
|
},
|
|
{
|
|
"entropy": 10.734352970123291,
|
|
"epoch": 0.0453857791225416,
|
|
"grad_norm": 2.1875,
|
|
"learning_rate": 2.2e-05,
|
|
"loss": 10.1008,
|
|
"mean_token_accuracy": 0.04177976287901401,
|
|
"num_tokens": 100252.0,
|
|
"step": 45
|
|
},
|
|
{
|
|
"entropy": 10.731473350524903,
|
|
"epoch": 0.05042864346949067,
|
|
"grad_norm": 2.0625,
|
|
"learning_rate": 2.4500000000000003e-05,
|
|
"loss": 10.0294,
|
|
"mean_token_accuracy": 0.035502530820667746,
|
|
"num_tokens": 110578.0,
|
|
"step": 50
|
|
},
|
|
{
|
|
"entropy": 10.731419467926026,
|
|
"epoch": 0.05547150781643974,
|
|
"grad_norm": 1.9453125,
|
|
"learning_rate": 2.7e-05,
|
|
"loss": 9.9893,
|
|
"mean_token_accuracy": 0.03719575461000204,
|
|
"num_tokens": 121631.0,
|
|
"step": 55
|
|
},
|
|
{
|
|
"entropy": 10.730601024627685,
|
|
"epoch": 0.060514372163388806,
|
|
"grad_norm": 1.96875,
|
|
"learning_rate": 2.95e-05,
|
|
"loss": 9.9237,
|
|
"mean_token_accuracy": 0.043491701036691664,
|
|
"num_tokens": 134254.0,
|
|
"step": 60
|
|
},
|
|
{
|
|
"entropy": 10.727654361724854,
|
|
"epoch": 0.06555723651033787,
|
|
"grad_norm": 1.734375,
|
|
"learning_rate": 3.2e-05,
|
|
"loss": 9.8843,
|
|
"mean_token_accuracy": 0.03803328052163124,
|
|
"num_tokens": 144928.0,
|
|
"step": 65
|
|
},
|
|
{
|
|
"entropy": 10.724302101135255,
|
|
"epoch": 0.07060010085728693,
|
|
"grad_norm": 1.8203125,
|
|
"learning_rate": 3.4500000000000005e-05,
|
|
"loss": 9.7798,
|
|
"mean_token_accuracy": 0.039696192182600495,
|
|
"num_tokens": 155518.0,
|
|
"step": 70
|
|
},
|
|
{
|
|
"entropy": 10.720338535308837,
|
|
"epoch": 0.07564296520423601,
|
|
"grad_norm": 1.9375,
|
|
"learning_rate": 3.7e-05,
|
|
"loss": 9.7038,
|
|
"mean_token_accuracy": 0.04226764794439077,
|
|
"num_tokens": 165850.0,
|
|
"step": 75
|
|
},
|
|
{
|
|
"entropy": 10.71519594192505,
|
|
"epoch": 0.08068582955118507,
|
|
"grad_norm": 1.8828125,
|
|
"learning_rate": 3.95e-05,
|
|
"loss": 9.6932,
|
|
"mean_token_accuracy": 0.04092370942234993,
|
|
"num_tokens": 176787.0,
|
|
"step": 80
|
|
},
|
|
{
|
|
"entropy": 10.709799861907959,
|
|
"epoch": 0.08572869389813415,
|
|
"grad_norm": 1.7578125,
|
|
"learning_rate": 4.2000000000000004e-05,
|
|
"loss": 9.6293,
|
|
"mean_token_accuracy": 0.03919961098581552,
|
|
"num_tokens": 188345.0,
|
|
"step": 85
|
|
},
|
|
{
|
|
"entropy": 10.706261253356933,
|
|
"epoch": 0.0907715582450832,
|
|
"grad_norm": 1.78125,
|
|
"learning_rate": 4.45e-05,
|
|
"loss": 9.5833,
|
|
"mean_token_accuracy": 0.03263202589005232,
|
|
"num_tokens": 199315.0,
|
|
"step": 90
|
|
},
|
|
{
|
|
"entropy": 10.702005004882812,
|
|
"epoch": 0.09581442259203228,
|
|
"grad_norm": 1.7578125,
|
|
"learning_rate": 4.7000000000000004e-05,
|
|
"loss": 9.4401,
|
|
"mean_token_accuracy": 0.04271471928805113,
|
|
"num_tokens": 209627.0,
|
|
"step": 95
|
|
},
|
|
{
|
|
"entropy": 10.688950729370116,
|
|
"epoch": 0.10085728693898134,
|
|
"grad_norm": 1.8125,
|
|
"learning_rate": 4.9500000000000004e-05,
|
|
"loss": 9.3648,
|
|
"mean_token_accuracy": 0.04892551712691784,
|
|
"num_tokens": 220872.0,
|
|
"step": 100
|
|
},
|
|
{
|
|
"entropy": 10.666452217102051,
|
|
"epoch": 0.1059001512859304,
|
|
"grad_norm": 1.8125,
|
|
"learning_rate": 5.2e-05,
|
|
"loss": 9.2706,
|
|
"mean_token_accuracy": 0.04733602069318295,
|
|
"num_tokens": 233272.0,
|
|
"step": 105
|
|
},
|
|
{
|
|
"entropy": 10.643214797973632,
|
|
"epoch": 0.11094301563287948,
|
|
"grad_norm": 1.8125,
|
|
"learning_rate": 5.45e-05,
|
|
"loss": 9.1724,
|
|
"mean_token_accuracy": 0.05065925419330597,
|
|
"num_tokens": 243529.0,
|
|
"step": 110
|
|
},
|
|
{
|
|
"entropy": 10.622024440765381,
|
|
"epoch": 0.11598587997982854,
|
|
"grad_norm": 1.640625,
|
|
"learning_rate": 5.7e-05,
|
|
"loss": 9.0982,
|
|
"mean_token_accuracy": 0.04518871679902077,
|
|
"num_tokens": 254942.0,
|
|
"step": 115
|
|
},
|
|
{
|
|
"entropy": 10.591822242736816,
|
|
"epoch": 0.12102874432677761,
|
|
"grad_norm": 1.7421875,
|
|
"learning_rate": 5.9499999999999996e-05,
|
|
"loss": 8.9746,
|
|
"mean_token_accuracy": 0.05078088715672493,
|
|
"num_tokens": 266552.0,
|
|
"step": 120
|
|
},
|
|
{
|
|
"entropy": 10.53172492980957,
|
|
"epoch": 0.1260716086737267,
|
|
"grad_norm": 1.59375,
|
|
"learning_rate": 6.2e-05,
|
|
"loss": 8.8367,
|
|
"mean_token_accuracy": 0.049286961928009985,
|
|
"num_tokens": 277906.0,
|
|
"step": 125
|
|
},
|
|
{
|
|
"entropy": 10.461231136322022,
|
|
"epoch": 0.13111447302067575,
|
|
"grad_norm": 1.6328125,
|
|
"learning_rate": 6.450000000000001e-05,
|
|
"loss": 8.7177,
|
|
"mean_token_accuracy": 0.05140590891242027,
|
|
"num_tokens": 289079.0,
|
|
"step": 130
|
|
},
|
|
{
|
|
"entropy": 10.385406398773194,
|
|
"epoch": 0.1361573373676248,
|
|
"grad_norm": 1.5,
|
|
"learning_rate": 6.7e-05,
|
|
"loss": 8.6441,
|
|
"mean_token_accuracy": 0.04429629724472761,
|
|
"num_tokens": 300107.0,
|
|
"step": 135
|
|
},
|
|
{
|
|
"entropy": 10.312572288513184,
|
|
"epoch": 0.14120020171457387,
|
|
"grad_norm": 1.5234375,
|
|
"learning_rate": 6.950000000000001e-05,
|
|
"loss": 8.4806,
|
|
"mean_token_accuracy": 0.05319313518702984,
|
|
"num_tokens": 310779.0,
|
|
"step": 140
|
|
},
|
|
{
|
|
"entropy": 10.203822803497314,
|
|
"epoch": 0.14624306606152296,
|
|
"grad_norm": 1.453125,
|
|
"learning_rate": 7.2e-05,
|
|
"loss": 8.4202,
|
|
"mean_token_accuracy": 0.051876432821154596,
|
|
"num_tokens": 323478.0,
|
|
"step": 145
|
|
},
|
|
{
|
|
"entropy": 10.096873664855957,
|
|
"epoch": 0.15128593040847202,
|
|
"grad_norm": 1.3515625,
|
|
"learning_rate": 7.45e-05,
|
|
"loss": 8.2552,
|
|
"mean_token_accuracy": 0.05630794763565063,
|
|
"num_tokens": 335255.0,
|
|
"step": 150
|
|
},
|
|
{
|
|
"entropy": 9.960623264312744,
|
|
"epoch": 0.15632879475542108,
|
|
"grad_norm": 1.3046875,
|
|
"learning_rate": 7.7e-05,
|
|
"loss": 8.1385,
|
|
"mean_token_accuracy": 0.05476163513958454,
|
|
"num_tokens": 344824.0,
|
|
"step": 155
|
|
},
|
|
{
|
|
"entropy": 9.771032524108886,
|
|
"epoch": 0.16137165910237014,
|
|
"grad_norm": 1.265625,
|
|
"learning_rate": 7.950000000000001e-05,
|
|
"loss": 8.0754,
|
|
"mean_token_accuracy": 0.05333587229251861,
|
|
"num_tokens": 356283.0,
|
|
"step": 160
|
|
},
|
|
{
|
|
"entropy": 9.623997592926026,
|
|
"epoch": 0.1664145234493192,
|
|
"grad_norm": 1.015625,
|
|
"learning_rate": 8.2e-05,
|
|
"loss": 7.996,
|
|
"mean_token_accuracy": 0.05669583305716515,
|
|
"num_tokens": 368098.0,
|
|
"step": 165
|
|
},
|
|
{
|
|
"entropy": 9.383370685577393,
|
|
"epoch": 0.1714573877962683,
|
|
"grad_norm": 1.03125,
|
|
"learning_rate": 8.450000000000001e-05,
|
|
"loss": 7.8796,
|
|
"mean_token_accuracy": 0.059235493466258046,
|
|
"num_tokens": 380948.0,
|
|
"step": 170
|
|
},
|
|
{
|
|
"entropy": 9.174373054504395,
|
|
"epoch": 0.17650025214321735,
|
|
"grad_norm": 1.0234375,
|
|
"learning_rate": 8.7e-05,
|
|
"loss": 7.7633,
|
|
"mean_token_accuracy": 0.05942847430706024,
|
|
"num_tokens": 390392.0,
|
|
"step": 175
|
|
},
|
|
{
|
|
"entropy": 8.980836009979248,
|
|
"epoch": 0.1815431164901664,
|
|
"grad_norm": 1.3203125,
|
|
"learning_rate": 8.95e-05,
|
|
"loss": 7.7496,
|
|
"mean_token_accuracy": 0.06362583562731743,
|
|
"num_tokens": 400328.0,
|
|
"step": 180
|
|
},
|
|
{
|
|
"entropy": 8.7655029296875,
|
|
"epoch": 0.18658598083711547,
|
|
"grad_norm": 0.828125,
|
|
"learning_rate": 9.2e-05,
|
|
"loss": 7.7544,
|
|
"mean_token_accuracy": 0.06182343065738678,
|
|
"num_tokens": 410909.0,
|
|
"step": 185
|
|
},
|
|
{
|
|
"entropy": 8.639219570159913,
|
|
"epoch": 0.19162884518406456,
|
|
"grad_norm": 0.9375,
|
|
"learning_rate": 9.45e-05,
|
|
"loss": 7.691,
|
|
"mean_token_accuracy": 0.06425628773868083,
|
|
"num_tokens": 422726.0,
|
|
"step": 190
|
|
},
|
|
{
|
|
"entropy": 8.559172058105469,
|
|
"epoch": 0.19667170953101362,
|
|
"grad_norm": 0.91796875,
|
|
"learning_rate": 9.7e-05,
|
|
"loss": 7.5594,
|
|
"mean_token_accuracy": 0.07219541147351265,
|
|
"num_tokens": 433376.0,
|
|
"step": 195
|
|
},
|
|
{
|
|
"entropy": 8.397376251220702,
|
|
"epoch": 0.20171457387796268,
|
|
"grad_norm": 1.0078125,
|
|
"learning_rate": 9.95e-05,
|
|
"loss": 7.6275,
|
|
"mean_token_accuracy": 0.06337097361683845,
|
|
"num_tokens": 443940.0,
|
|
"step": 200
|
|
},
|
|
{
|
|
"entropy": 8.361763286590577,
|
|
"epoch": 0.20675743822491174,
|
|
"grad_norm": 0.78125,
|
|
"learning_rate": 0.000102,
|
|
"loss": 7.5213,
|
|
"mean_token_accuracy": 0.06769266054034233,
|
|
"num_tokens": 455016.0,
|
|
"step": 205
|
|
},
|
|
{
|
|
"entropy": 8.303247165679931,
|
|
"epoch": 0.2118003025718608,
|
|
"grad_norm": 0.92578125,
|
|
"learning_rate": 0.00010449999999999999,
|
|
"loss": 7.553,
|
|
"mean_token_accuracy": 0.06669275388121605,
|
|
"num_tokens": 466380.0,
|
|
"step": 210
|
|
},
|
|
{
|
|
"entropy": 8.23162612915039,
|
|
"epoch": 0.2168431669188099,
|
|
"grad_norm": 1.140625,
|
|
"learning_rate": 0.000107,
|
|
"loss": 7.4065,
|
|
"mean_token_accuracy": 0.07161542177200317,
|
|
"num_tokens": 476915.0,
|
|
"step": 215
|
|
},
|
|
{
|
|
"entropy": 8.226816177368164,
|
|
"epoch": 0.22188603126575895,
|
|
"grad_norm": 1.1015625,
|
|
"learning_rate": 0.0001095,
|
|
"loss": 7.4817,
|
|
"mean_token_accuracy": 0.07145784981548786,
|
|
"num_tokens": 488924.0,
|
|
"step": 220
|
|
},
|
|
{
|
|
"entropy": 8.18234987258911,
|
|
"epoch": 0.22692889561270801,
|
|
"grad_norm": 0.83984375,
|
|
"learning_rate": 0.000112,
|
|
"loss": 7.4263,
|
|
"mean_token_accuracy": 0.07183554545044898,
|
|
"num_tokens": 498965.0,
|
|
"step": 225
|
|
},
|
|
{
|
|
"entropy": 8.137568473815918,
|
|
"epoch": 0.23197175995965708,
|
|
"grad_norm": 0.828125,
|
|
"learning_rate": 0.0001145,
|
|
"loss": 7.4409,
|
|
"mean_token_accuracy": 0.07232684977352619,
|
|
"num_tokens": 509755.0,
|
|
"step": 230
|
|
},
|
|
{
|
|
"entropy": 8.12470645904541,
|
|
"epoch": 0.23701462430660616,
|
|
"grad_norm": 0.796875,
|
|
"learning_rate": 0.00011700000000000001,
|
|
"loss": 7.4064,
|
|
"mean_token_accuracy": 0.07139695808291435,
|
|
"num_tokens": 521221.0,
|
|
"step": 235
|
|
},
|
|
{
|
|
"entropy": 8.080301761627197,
|
|
"epoch": 0.24205748865355523,
|
|
"grad_norm": 0.91015625,
|
|
"learning_rate": 0.00011949999999999999,
|
|
"loss": 7.503,
|
|
"mean_token_accuracy": 0.06776785589754582,
|
|
"num_tokens": 533595.0,
|
|
"step": 240
|
|
},
|
|
{
|
|
"entropy": 8.111075401306152,
|
|
"epoch": 0.24710035300050429,
|
|
"grad_norm": 0.94140625,
|
|
"learning_rate": 0.000122,
|
|
"loss": 7.4348,
|
|
"mean_token_accuracy": 0.07060522697865963,
|
|
"num_tokens": 543672.0,
|
|
"step": 245
|
|
},
|
|
{
|
|
"entropy": 8.046464586257935,
|
|
"epoch": 0.2521432173474534,
|
|
"grad_norm": 0.7734375,
|
|
"learning_rate": 0.0001245,
|
|
"loss": 7.3208,
|
|
"mean_token_accuracy": 0.07489579617977142,
|
|
"num_tokens": 554906.0,
|
|
"step": 250
|
|
},
|
|
{
|
|
"entropy": 7.984986257553101,
|
|
"epoch": 0.25718608169440244,
|
|
"grad_norm": 0.90234375,
|
|
"learning_rate": 0.000127,
|
|
"loss": 7.3941,
|
|
"mean_token_accuracy": 0.07390029206871987,
|
|
"num_tokens": 566419.0,
|
|
"step": 255
|
|
},
|
|
{
|
|
"entropy": 8.038315868377685,
|
|
"epoch": 0.2622289460413515,
|
|
"grad_norm": 1.1171875,
|
|
"learning_rate": 0.0001295,
|
|
"loss": 7.3625,
|
|
"mean_token_accuracy": 0.07229974381625652,
|
|
"num_tokens": 578848.0,
|
|
"step": 260
|
|
},
|
|
{
|
|
"entropy": 7.896354866027832,
|
|
"epoch": 0.26727181038830056,
|
|
"grad_norm": 0.92578125,
|
|
"learning_rate": 0.000132,
|
|
"loss": 7.305,
|
|
"mean_token_accuracy": 0.07605222910642624,
|
|
"num_tokens": 590601.0,
|
|
"step": 265
|
|
},
|
|
{
|
|
"entropy": 7.9424131393432615,
|
|
"epoch": 0.2723146747352496,
|
|
"grad_norm": 0.9375,
|
|
"learning_rate": 0.00013450000000000002,
|
|
"loss": 7.3548,
|
|
"mean_token_accuracy": 0.07202886901795864,
|
|
"num_tokens": 602021.0,
|
|
"step": 270
|
|
},
|
|
{
|
|
"entropy": 8.004773712158203,
|
|
"epoch": 0.2773575390821987,
|
|
"grad_norm": 0.95703125,
|
|
"learning_rate": 0.00013700000000000002,
|
|
"loss": 7.4443,
|
|
"mean_token_accuracy": 0.06653593964874745,
|
|
"num_tokens": 613966.0,
|
|
"step": 275
|
|
},
|
|
{
|
|
"entropy": 7.990764474868774,
|
|
"epoch": 0.28240040342914774,
|
|
"grad_norm": 1.5390625,
|
|
"learning_rate": 0.0001395,
|
|
"loss": 7.2762,
|
|
"mean_token_accuracy": 0.07293831780552865,
|
|
"num_tokens": 624851.0,
|
|
"step": 280
|
|
},
|
|
{
|
|
"entropy": 7.808479070663452,
|
|
"epoch": 0.2874432677760968,
|
|
"grad_norm": 0.8515625,
|
|
"learning_rate": 0.00014199999999999998,
|
|
"loss": 7.3518,
|
|
"mean_token_accuracy": 0.07462373450398445,
|
|
"num_tokens": 635580.0,
|
|
"step": 285
|
|
},
|
|
{
|
|
"entropy": 7.965742826461792,
|
|
"epoch": 0.2924861321230459,
|
|
"grad_norm": 0.9296875,
|
|
"learning_rate": 0.0001445,
|
|
"loss": 7.4071,
|
|
"mean_token_accuracy": 0.0715030949562788,
|
|
"num_tokens": 646927.0,
|
|
"step": 290
|
|
},
|
|
{
|
|
"entropy": 7.943162631988526,
|
|
"epoch": 0.297528996469995,
|
|
"grad_norm": 1.0234375,
|
|
"learning_rate": 0.000147,
|
|
"loss": 7.2414,
|
|
"mean_token_accuracy": 0.07769636139273643,
|
|
"num_tokens": 658116.0,
|
|
"step": 295
|
|
},
|
|
{
|
|
"entropy": 7.861419916152954,
|
|
"epoch": 0.30257186081694404,
|
|
"grad_norm": 0.9453125,
|
|
"learning_rate": 0.0001495,
|
|
"loss": 7.2404,
|
|
"mean_token_accuracy": 0.07680558413267136,
|
|
"num_tokens": 669054.0,
|
|
"step": 300
|
|
},
|
|
{
|
|
"entropy": 7.859039735794068,
|
|
"epoch": 0.3076147251638931,
|
|
"grad_norm": 1.203125,
|
|
"learning_rate": 0.000152,
|
|
"loss": 7.2846,
|
|
"mean_token_accuracy": 0.07930284813046455,
|
|
"num_tokens": 680439.0,
|
|
"step": 305
|
|
},
|
|
{
|
|
"entropy": 7.81926383972168,
|
|
"epoch": 0.31265758951084216,
|
|
"grad_norm": 1.0859375,
|
|
"learning_rate": 0.00015450000000000001,
|
|
"loss": 7.2209,
|
|
"mean_token_accuracy": 0.0794371448457241,
|
|
"num_tokens": 690909.0,
|
|
"step": 310
|
|
},
|
|
{
|
|
"entropy": 7.75442066192627,
|
|
"epoch": 0.3177004538577912,
|
|
"grad_norm": 1.0234375,
|
|
"learning_rate": 0.000157,
|
|
"loss": 7.209,
|
|
"mean_token_accuracy": 0.07925689667463302,
|
|
"num_tokens": 701744.0,
|
|
"step": 315
|
|
},
|
|
{
|
|
"entropy": 7.746679449081421,
|
|
"epoch": 0.3227433182047403,
|
|
"grad_norm": 0.9296875,
|
|
"learning_rate": 0.0001595,
|
|
"loss": 7.2338,
|
|
"mean_token_accuracy": 0.08032149039208888,
|
|
"num_tokens": 712268.0,
|
|
"step": 320
|
|
},
|
|
{
|
|
"entropy": 7.765822124481201,
|
|
"epoch": 0.32778618255168934,
|
|
"grad_norm": 0.9765625,
|
|
"learning_rate": 0.000162,
|
|
"loss": 7.2609,
|
|
"mean_token_accuracy": 0.07947171069681644,
|
|
"num_tokens": 722978.0,
|
|
"step": 325
|
|
},
|
|
{
|
|
"entropy": 7.870586299896241,
|
|
"epoch": 0.3328290468986384,
|
|
"grad_norm": 0.94921875,
|
|
"learning_rate": 0.00016450000000000001,
|
|
"loss": 7.253,
|
|
"mean_token_accuracy": 0.07748774662613869,
|
|
"num_tokens": 734278.0,
|
|
"step": 330
|
|
},
|
|
{
|
|
"entropy": 7.8009439468383786,
|
|
"epoch": 0.3378719112455875,
|
|
"grad_norm": 0.8515625,
|
|
"learning_rate": 0.00016700000000000002,
|
|
"loss": 7.1867,
|
|
"mean_token_accuracy": 0.07406368069350719,
|
|
"num_tokens": 746109.0,
|
|
"step": 335
|
|
},
|
|
{
|
|
"entropy": 7.751057529449463,
|
|
"epoch": 0.3429147755925366,
|
|
"grad_norm": 1.0703125,
|
|
"learning_rate": 0.00016950000000000003,
|
|
"loss": 7.1104,
|
|
"mean_token_accuracy": 0.07908297292888164,
|
|
"num_tokens": 757519.0,
|
|
"step": 340
|
|
},
|
|
{
|
|
"entropy": 7.711837816238403,
|
|
"epoch": 0.34795763993948564,
|
|
"grad_norm": 1.0234375,
|
|
"learning_rate": 0.00017199999999999998,
|
|
"loss": 7.0784,
|
|
"mean_token_accuracy": 0.08319444209337234,
|
|
"num_tokens": 767665.0,
|
|
"step": 345
|
|
},
|
|
{
|
|
"entropy": 7.729198932647705,
|
|
"epoch": 0.3530005042864347,
|
|
"grad_norm": 1.125,
|
|
"learning_rate": 0.00017449999999999999,
|
|
"loss": 7.118,
|
|
"mean_token_accuracy": 0.08570092767477036,
|
|
"num_tokens": 778880.0,
|
|
"step": 350
|
|
},
|
|
{
|
|
"entropy": 7.589126014709473,
|
|
"epoch": 0.35804336863338376,
|
|
"grad_norm": 0.94140625,
|
|
"learning_rate": 0.000177,
|
|
"loss": 7.1275,
|
|
"mean_token_accuracy": 0.08408626839518547,
|
|
"num_tokens": 790360.0,
|
|
"step": 355
|
|
},
|
|
{
|
|
"entropy": 7.650224447250366,
|
|
"epoch": 0.3630862329803328,
|
|
"grad_norm": 1.1796875,
|
|
"learning_rate": 0.0001795,
|
|
"loss": 7.0884,
|
|
"mean_token_accuracy": 0.08589479774236679,
|
|
"num_tokens": 800889.0,
|
|
"step": 360
|
|
},
|
|
{
|
|
"entropy": 7.698587417602539,
|
|
"epoch": 0.3681290973272819,
|
|
"grad_norm": 1.1015625,
|
|
"learning_rate": 0.000182,
|
|
"loss": 7.0774,
|
|
"mean_token_accuracy": 0.08224210813641548,
|
|
"num_tokens": 811414.0,
|
|
"step": 365
|
|
},
|
|
{
|
|
"entropy": 7.551970529556274,
|
|
"epoch": 0.37317196167423095,
|
|
"grad_norm": 0.96484375,
|
|
"learning_rate": 0.0001845,
|
|
"loss": 7.0729,
|
|
"mean_token_accuracy": 0.08361641615629196,
|
|
"num_tokens": 822822.0,
|
|
"step": 370
|
|
},
|
|
{
|
|
"entropy": 7.61205792427063,
|
|
"epoch": 0.37821482602118,
|
|
"grad_norm": 1.09375,
|
|
"learning_rate": 0.000187,
|
|
"loss": 6.9949,
|
|
"mean_token_accuracy": 0.09048431143164634,
|
|
"num_tokens": 833145.0,
|
|
"step": 375
|
|
},
|
|
{
|
|
"entropy": 7.606620168685913,
|
|
"epoch": 0.3832576903681291,
|
|
"grad_norm": 1.046875,
|
|
"learning_rate": 0.0001895,
|
|
"loss": 7.1067,
|
|
"mean_token_accuracy": 0.08064279481768608,
|
|
"num_tokens": 844015.0,
|
|
"step": 380
|
|
},
|
|
{
|
|
"entropy": 7.537594938278199,
|
|
"epoch": 0.3883005547150782,
|
|
"grad_norm": 0.8984375,
|
|
"learning_rate": 0.000192,
|
|
"loss": 7.0227,
|
|
"mean_token_accuracy": 0.08781845718622208,
|
|
"num_tokens": 855806.0,
|
|
"step": 385
|
|
},
|
|
{
|
|
"entropy": 7.601389598846436,
|
|
"epoch": 0.39334341906202724,
|
|
"grad_norm": 1.09375,
|
|
"learning_rate": 0.0001945,
|
|
"loss": 6.952,
|
|
"mean_token_accuracy": 0.08714486509561539,
|
|
"num_tokens": 865750.0,
|
|
"step": 390
|
|
},
|
|
{
|
|
"entropy": 7.603587245941162,
|
|
"epoch": 0.3983862834089763,
|
|
"grad_norm": 0.9296875,
|
|
"learning_rate": 0.00019700000000000002,
|
|
"loss": 7.0798,
|
|
"mean_token_accuracy": 0.08171837627887726,
|
|
"num_tokens": 877630.0,
|
|
"step": 395
|
|
},
|
|
{
|
|
"entropy": 7.46077446937561,
|
|
"epoch": 0.40342914775592537,
|
|
"grad_norm": 0.921875,
|
|
"learning_rate": 0.00019950000000000002,
|
|
"loss": 6.9621,
|
|
"mean_token_accuracy": 0.08956382796168327,
|
|
"num_tokens": 888667.0,
|
|
"step": 400
|
|
},
|
|
{
|
|
"entropy": 7.612736940383911,
|
|
"epoch": 0.4084720121028744,
|
|
"grad_norm": 1.0625,
|
|
"learning_rate": 0.000202,
|
|
"loss": 7.0383,
|
|
"mean_token_accuracy": 0.0815612755715847,
|
|
"num_tokens": 900989.0,
|
|
"step": 405
|
|
},
|
|
{
|
|
"entropy": 7.4917103290557865,
|
|
"epoch": 0.4135148764498235,
|
|
"grad_norm": 0.94921875,
|
|
"learning_rate": 0.00020449999999999998,
|
|
"loss": 6.9247,
|
|
"mean_token_accuracy": 0.08268933743238449,
|
|
"num_tokens": 911479.0,
|
|
"step": 410
|
|
},
|
|
{
|
|
"entropy": 7.538983345031738,
|
|
"epoch": 0.41855774079677255,
|
|
"grad_norm": 0.96875,
|
|
"learning_rate": 0.000207,
|
|
"loss": 6.9579,
|
|
"mean_token_accuracy": 0.0860539086163044,
|
|
"num_tokens": 922434.0,
|
|
"step": 415
|
|
},
|
|
{
|
|
"entropy": 7.502885150909424,
|
|
"epoch": 0.4236006051437216,
|
|
"grad_norm": 1.0859375,
|
|
"learning_rate": 0.0002095,
|
|
"loss": 6.9417,
|
|
"mean_token_accuracy": 0.08750939816236496,
|
|
"num_tokens": 933405.0,
|
|
"step": 420
|
|
},
|
|
{
|
|
"entropy": 7.503132390975952,
|
|
"epoch": 0.4286434694906707,
|
|
"grad_norm": 1.09375,
|
|
"learning_rate": 0.000212,
|
|
"loss": 6.9414,
|
|
"mean_token_accuracy": 0.0863698624074459,
|
|
"num_tokens": 943567.0,
|
|
"step": 425
|
|
},
|
|
{
|
|
"entropy": 7.433593511581421,
|
|
"epoch": 0.4336863338376198,
|
|
"grad_norm": 1.015625,
|
|
"learning_rate": 0.0002145,
|
|
"loss": 6.9729,
|
|
"mean_token_accuracy": 0.08398969918489456,
|
|
"num_tokens": 955123.0,
|
|
"step": 430
|
|
},
|
|
{
|
|
"entropy": 7.4864178657531735,
|
|
"epoch": 0.43872919818456885,
|
|
"grad_norm": 1.140625,
|
|
"learning_rate": 0.00021700000000000002,
|
|
"loss": 6.9878,
|
|
"mean_token_accuracy": 0.08637007214128971,
|
|
"num_tokens": 965565.0,
|
|
"step": 435
|
|
},
|
|
{
|
|
"entropy": 7.533525228500366,
|
|
"epoch": 0.4437720625315179,
|
|
"grad_norm": 0.890625,
|
|
"learning_rate": 0.0002195,
|
|
"loss": 6.9331,
|
|
"mean_token_accuracy": 0.09186888933181762,
|
|
"num_tokens": 977101.0,
|
|
"step": 440
|
|
},
|
|
{
|
|
"entropy": 7.447479343414306,
|
|
"epoch": 0.44881492687846697,
|
|
"grad_norm": 0.96484375,
|
|
"learning_rate": 0.000222,
|
|
"loss": 6.9676,
|
|
"mean_token_accuracy": 0.08773500472307205,
|
|
"num_tokens": 987445.0,
|
|
"step": 445
|
|
},
|
|
{
|
|
"entropy": 7.413919162750244,
|
|
"epoch": 0.45385779122541603,
|
|
"grad_norm": 0.98828125,
|
|
"learning_rate": 0.0002245,
|
|
"loss": 6.8527,
|
|
"mean_token_accuracy": 0.0943599946796894,
|
|
"num_tokens": 999379.0,
|
|
"step": 450
|
|
},
|
|
{
|
|
"entropy": 7.366581916809082,
|
|
"epoch": 0.4589006555723651,
|
|
"grad_norm": 1.0625,
|
|
"learning_rate": 0.00022700000000000002,
|
|
"loss": 6.7887,
|
|
"mean_token_accuracy": 0.09886589571833611,
|
|
"num_tokens": 1010744.0,
|
|
"step": 455
|
|
},
|
|
{
|
|
"entropy": 7.395628499984741,
|
|
"epoch": 0.46394351991931415,
|
|
"grad_norm": 1.0390625,
|
|
"learning_rate": 0.00022950000000000002,
|
|
"loss": 6.8751,
|
|
"mean_token_accuracy": 0.09002586975693702,
|
|
"num_tokens": 1021808.0,
|
|
"step": 460
|
|
},
|
|
{
|
|
"entropy": 7.436336374282837,
|
|
"epoch": 0.4689863842662632,
|
|
"grad_norm": 1.0703125,
|
|
"learning_rate": 0.00023200000000000003,
|
|
"loss": 6.9197,
|
|
"mean_token_accuracy": 0.08965835645794869,
|
|
"num_tokens": 1033694.0,
|
|
"step": 465
|
|
},
|
|
{
|
|
"entropy": 7.366260051727295,
|
|
"epoch": 0.47402924861321233,
|
|
"grad_norm": 1.1875,
|
|
"learning_rate": 0.00023449999999999998,
|
|
"loss": 6.8866,
|
|
"mean_token_accuracy": 0.08940717950463295,
|
|
"num_tokens": 1044690.0,
|
|
"step": 470
|
|
},
|
|
{
|
|
"entropy": 7.3759284019470215,
|
|
"epoch": 0.4790721129601614,
|
|
"grad_norm": 0.96484375,
|
|
"learning_rate": 0.000237,
|
|
"loss": 6.883,
|
|
"mean_token_accuracy": 0.09023785665631294,
|
|
"num_tokens": 1055557.0,
|
|
"step": 475
|
|
},
|
|
{
|
|
"entropy": 7.412348127365112,
|
|
"epoch": 0.48411497730711045,
|
|
"grad_norm": 1.0234375,
|
|
"learning_rate": 0.0002395,
|
|
"loss": 6.8225,
|
|
"mean_token_accuracy": 0.09234220460057259,
|
|
"num_tokens": 1067304.0,
|
|
"step": 480
|
|
},
|
|
{
|
|
"entropy": 7.280447006225586,
|
|
"epoch": 0.4891578416540595,
|
|
"grad_norm": 1.15625,
|
|
"learning_rate": 0.000242,
|
|
"loss": 6.8498,
|
|
"mean_token_accuracy": 0.09138206690549851,
|
|
"num_tokens": 1078955.0,
|
|
"step": 485
|
|
},
|
|
{
|
|
"entropy": 7.339125728607177,
|
|
"epoch": 0.49420070600100857,
|
|
"grad_norm": 1.0078125,
|
|
"learning_rate": 0.0002445,
|
|
"loss": 6.8907,
|
|
"mean_token_accuracy": 0.09167812019586563,
|
|
"num_tokens": 1091405.0,
|
|
"step": 490
|
|
},
|
|
{
|
|
"entropy": 7.334241151809692,
|
|
"epoch": 0.49924357034795763,
|
|
"grad_norm": 1.1640625,
|
|
"learning_rate": 0.000247,
|
|
"loss": 6.8625,
|
|
"mean_token_accuracy": 0.09363722130656242,
|
|
"num_tokens": 1103227.0,
|
|
"step": 495
|
|
},
|
|
{
|
|
"entropy": 7.351650381088257,
|
|
"epoch": 0.5042864346949067,
|
|
"grad_norm": 0.890625,
|
|
"learning_rate": 0.0002495,
|
|
"loss": 6.7968,
|
|
"mean_token_accuracy": 0.09819836765527726,
|
|
"num_tokens": 1114485.0,
|
|
"step": 500
|
|
},
|
|
{
|
|
"epoch": 0.5042864346949067,
|
|
"eval_entropy": 7.210145644559205,
|
|
"eval_loss": 6.899664402008057,
|
|
"eval_mean_token_accuracy": 0.09479910408267538,
|
|
"eval_num_tokens": 1114485.0,
|
|
"eval_runtime": 2.1175,
|
|
"eval_samples_per_second": 1481.964,
|
|
"eval_steps_per_second": 185.6,
|
|
"step": 500
|
|
},
|
|
{
|
|
"entropy": 7.260896873474121,
|
|
"epoch": 0.5093292990418558,
|
|
"grad_norm": 0.96484375,
|
|
"learning_rate": 0.000252,
|
|
"loss": 6.8181,
|
|
"mean_token_accuracy": 0.09441837519407273,
|
|
"num_tokens": 1126789.0,
|
|
"step": 505
|
|
},
|
|
{
|
|
"entropy": 7.352414274215699,
|
|
"epoch": 0.5143721633888049,
|
|
"grad_norm": 0.93359375,
|
|
"learning_rate": 0.0002545,
|
|
"loss": 6.7989,
|
|
"mean_token_accuracy": 0.09330016896128654,
|
|
"num_tokens": 1138192.0,
|
|
"step": 510
|
|
},
|
|
{
|
|
"entropy": 7.2082147121429445,
|
|
"epoch": 0.5194150277357539,
|
|
"grad_norm": 1.1328125,
|
|
"learning_rate": 0.000257,
|
|
"loss": 6.7411,
|
|
"mean_token_accuracy": 0.09130406081676483,
|
|
"num_tokens": 1148526.0,
|
|
"step": 515
|
|
},
|
|
{
|
|
"entropy": 7.369548559188843,
|
|
"epoch": 0.524457892082703,
|
|
"grad_norm": 1.0,
|
|
"learning_rate": 0.0002595,
|
|
"loss": 6.8496,
|
|
"mean_token_accuracy": 0.08768555521965027,
|
|
"num_tokens": 1159194.0,
|
|
"step": 520
|
|
},
|
|
{
|
|
"entropy": 7.23188009262085,
|
|
"epoch": 0.529500756429652,
|
|
"grad_norm": 0.9609375,
|
|
"learning_rate": 0.000262,
|
|
"loss": 6.7433,
|
|
"mean_token_accuracy": 0.1001415841281414,
|
|
"num_tokens": 1170601.0,
|
|
"step": 525
|
|
},
|
|
{
|
|
"entropy": 7.288579416275025,
|
|
"epoch": 0.5345436207766011,
|
|
"grad_norm": 1.09375,
|
|
"learning_rate": 0.00026450000000000003,
|
|
"loss": 6.7419,
|
|
"mean_token_accuracy": 0.0954306311905384,
|
|
"num_tokens": 1182441.0,
|
|
"step": 530
|
|
},
|
|
{
|
|
"entropy": 7.17895131111145,
|
|
"epoch": 0.5395864851235502,
|
|
"grad_norm": 1.1953125,
|
|
"learning_rate": 0.00026700000000000004,
|
|
"loss": 6.6349,
|
|
"mean_token_accuracy": 0.10441381633281707,
|
|
"num_tokens": 1192530.0,
|
|
"step": 535
|
|
},
|
|
{
|
|
"entropy": 7.13337082862854,
|
|
"epoch": 0.5446293494704992,
|
|
"grad_norm": 1.0234375,
|
|
"learning_rate": 0.00026950000000000005,
|
|
"loss": 6.7521,
|
|
"mean_token_accuracy": 0.0956033818423748,
|
|
"num_tokens": 1203831.0,
|
|
"step": 540
|
|
},
|
|
{
|
|
"entropy": 7.2454320907592775,
|
|
"epoch": 0.5496722138174484,
|
|
"grad_norm": 1.0859375,
|
|
"learning_rate": 0.00027200000000000005,
|
|
"loss": 6.6923,
|
|
"mean_token_accuracy": 0.0969244122505188,
|
|
"num_tokens": 1214164.0,
|
|
"step": 545
|
|
},
|
|
{
|
|
"entropy": 7.150829601287842,
|
|
"epoch": 0.5547150781643974,
|
|
"grad_norm": 0.98828125,
|
|
"learning_rate": 0.0002745,
|
|
"loss": 6.7235,
|
|
"mean_token_accuracy": 0.09548334404826164,
|
|
"num_tokens": 1226932.0,
|
|
"step": 550
|
|
},
|
|
{
|
|
"entropy": 7.283726549148559,
|
|
"epoch": 0.5597579425113465,
|
|
"grad_norm": 1.1328125,
|
|
"learning_rate": 0.000277,
|
|
"loss": 6.79,
|
|
"mean_token_accuracy": 0.08873454928398132,
|
|
"num_tokens": 1237212.0,
|
|
"step": 555
|
|
},
|
|
{
|
|
"entropy": 7.143116188049317,
|
|
"epoch": 0.5648008068582955,
|
|
"grad_norm": 1.0859375,
|
|
"learning_rate": 0.0002795,
|
|
"loss": 6.6907,
|
|
"mean_token_accuracy": 0.09660674557089806,
|
|
"num_tokens": 1249340.0,
|
|
"step": 560
|
|
},
|
|
{
|
|
"entropy": 7.123078393936157,
|
|
"epoch": 0.5698436712052446,
|
|
"grad_norm": 1.03125,
|
|
"learning_rate": 0.00028199999999999997,
|
|
"loss": 6.6907,
|
|
"mean_token_accuracy": 0.09954180717468261,
|
|
"num_tokens": 1260778.0,
|
|
"step": 565
|
|
},
|
|
{
|
|
"entropy": 7.1541259765625,
|
|
"epoch": 0.5748865355521936,
|
|
"grad_norm": 0.91015625,
|
|
"learning_rate": 0.0002845,
|
|
"loss": 6.6986,
|
|
"mean_token_accuracy": 0.09604307189583779,
|
|
"num_tokens": 1272216.0,
|
|
"step": 570
|
|
},
|
|
{
|
|
"entropy": 7.223128890991211,
|
|
"epoch": 0.5799293998991427,
|
|
"grad_norm": 1.46875,
|
|
"learning_rate": 0.000287,
|
|
"loss": 6.7725,
|
|
"mean_token_accuracy": 0.09934922829270362,
|
|
"num_tokens": 1283558.0,
|
|
"step": 575
|
|
},
|
|
{
|
|
"entropy": 7.209529066085816,
|
|
"epoch": 0.5849722642460918,
|
|
"grad_norm": 1.2890625,
|
|
"learning_rate": 0.0002895,
|
|
"loss": 6.834,
|
|
"mean_token_accuracy": 0.08799525387585164,
|
|
"num_tokens": 1294638.0,
|
|
"step": 580
|
|
},
|
|
{
|
|
"entropy": 7.27526068687439,
|
|
"epoch": 0.5900151285930408,
|
|
"grad_norm": 1.15625,
|
|
"learning_rate": 0.000292,
|
|
"loss": 6.7472,
|
|
"mean_token_accuracy": 0.09616359025239944,
|
|
"num_tokens": 1304697.0,
|
|
"step": 585
|
|
},
|
|
{
|
|
"entropy": 7.173431634902954,
|
|
"epoch": 0.59505799293999,
|
|
"grad_norm": 1.1484375,
|
|
"learning_rate": 0.0002945,
|
|
"loss": 6.7033,
|
|
"mean_token_accuracy": 0.09485772103071213,
|
|
"num_tokens": 1316618.0,
|
|
"step": 590
|
|
},
|
|
{
|
|
"entropy": 7.151233100891114,
|
|
"epoch": 0.600100857286939,
|
|
"grad_norm": 0.91015625,
|
|
"learning_rate": 0.000297,
|
|
"loss": 6.8116,
|
|
"mean_token_accuracy": 0.08478758260607719,
|
|
"num_tokens": 1328134.0,
|
|
"step": 595
|
|
},
|
|
{
|
|
"entropy": 7.242454576492309,
|
|
"epoch": 0.6051437216338881,
|
|
"grad_norm": 0.8671875,
|
|
"learning_rate": 0.0002995,
|
|
"loss": 6.7227,
|
|
"mean_token_accuracy": 0.0935085467994213,
|
|
"num_tokens": 1339101.0,
|
|
"step": 600
|
|
},
|
|
{
|
|
"entropy": 7.143788623809814,
|
|
"epoch": 0.6101865859808371,
|
|
"grad_norm": 0.984375,
|
|
"learning_rate": 0.000302,
|
|
"loss": 6.5973,
|
|
"mean_token_accuracy": 0.09975283592939377,
|
|
"num_tokens": 1349181.0,
|
|
"step": 605
|
|
},
|
|
{
|
|
"entropy": 7.0980949878692625,
|
|
"epoch": 0.6152294503277862,
|
|
"grad_norm": 0.8828125,
|
|
"learning_rate": 0.0003045,
|
|
"loss": 6.7221,
|
|
"mean_token_accuracy": 0.09565178751945495,
|
|
"num_tokens": 1361443.0,
|
|
"step": 610
|
|
},
|
|
{
|
|
"entropy": 7.098438501358032,
|
|
"epoch": 0.6202723146747352,
|
|
"grad_norm": 1.0078125,
|
|
"learning_rate": 0.000307,
|
|
"loss": 6.6778,
|
|
"mean_token_accuracy": 0.10161929428577424,
|
|
"num_tokens": 1372807.0,
|
|
"step": 615
|
|
},
|
|
{
|
|
"entropy": 7.160452890396118,
|
|
"epoch": 0.6253151790216843,
|
|
"grad_norm": 0.9765625,
|
|
"learning_rate": 0.0003095,
|
|
"loss": 6.7426,
|
|
"mean_token_accuracy": 0.09811687916517257,
|
|
"num_tokens": 1385509.0,
|
|
"step": 620
|
|
},
|
|
{
|
|
"entropy": 7.133578205108643,
|
|
"epoch": 0.6303580433686334,
|
|
"grad_norm": 1.1328125,
|
|
"learning_rate": 0.000312,
|
|
"loss": 6.6405,
|
|
"mean_token_accuracy": 0.0968807429075241,
|
|
"num_tokens": 1396035.0,
|
|
"step": 625
|
|
},
|
|
{
|
|
"entropy": 7.047602128982544,
|
|
"epoch": 0.6354009077155824,
|
|
"grad_norm": 1.046875,
|
|
"learning_rate": 0.0003145,
|
|
"loss": 6.5876,
|
|
"mean_token_accuracy": 0.10460086688399314,
|
|
"num_tokens": 1406839.0,
|
|
"step": 630
|
|
},
|
|
{
|
|
"entropy": 6.976920652389526,
|
|
"epoch": 0.6404437720625316,
|
|
"grad_norm": 0.95703125,
|
|
"learning_rate": 0.000317,
|
|
"loss": 6.559,
|
|
"mean_token_accuracy": 0.09928513914346696,
|
|
"num_tokens": 1417708.0,
|
|
"step": 635
|
|
},
|
|
{
|
|
"entropy": 7.116953611373901,
|
|
"epoch": 0.6454866364094806,
|
|
"grad_norm": 1.2265625,
|
|
"learning_rate": 0.0003195,
|
|
"loss": 6.6878,
|
|
"mean_token_accuracy": 0.0931871697306633,
|
|
"num_tokens": 1428985.0,
|
|
"step": 640
|
|
},
|
|
{
|
|
"entropy": 7.110161256790161,
|
|
"epoch": 0.6505295007564297,
|
|
"grad_norm": 1.25,
|
|
"learning_rate": 0.000322,
|
|
"loss": 6.7068,
|
|
"mean_token_accuracy": 0.10163840278983116,
|
|
"num_tokens": 1440232.0,
|
|
"step": 645
|
|
},
|
|
{
|
|
"entropy": 6.962800979614258,
|
|
"epoch": 0.6555723651033787,
|
|
"grad_norm": 1.0390625,
|
|
"learning_rate": 0.00032450000000000003,
|
|
"loss": 6.4498,
|
|
"mean_token_accuracy": 0.1053794801235199,
|
|
"num_tokens": 1451735.0,
|
|
"step": 650
|
|
},
|
|
{
|
|
"entropy": 7.015639686584473,
|
|
"epoch": 0.6606152294503278,
|
|
"grad_norm": 1.015625,
|
|
"learning_rate": 0.00032700000000000003,
|
|
"loss": 6.6221,
|
|
"mean_token_accuracy": 0.09871228262782097,
|
|
"num_tokens": 1463477.0,
|
|
"step": 655
|
|
},
|
|
{
|
|
"entropy": 7.01046838760376,
|
|
"epoch": 0.6656580937972768,
|
|
"grad_norm": 1.09375,
|
|
"learning_rate": 0.00032950000000000004,
|
|
"loss": 6.6263,
|
|
"mean_token_accuracy": 0.09373534545302391,
|
|
"num_tokens": 1473802.0,
|
|
"step": 660
|
|
},
|
|
{
|
|
"entropy": 7.009602022171021,
|
|
"epoch": 0.6707009581442259,
|
|
"grad_norm": 1.109375,
|
|
"learning_rate": 0.00033200000000000005,
|
|
"loss": 6.6264,
|
|
"mean_token_accuracy": 0.09858302772045135,
|
|
"num_tokens": 1485519.0,
|
|
"step": 665
|
|
},
|
|
{
|
|
"entropy": 7.069497060775757,
|
|
"epoch": 0.675743822491175,
|
|
"grad_norm": 1.015625,
|
|
"learning_rate": 0.00033450000000000005,
|
|
"loss": 6.6323,
|
|
"mean_token_accuracy": 0.09767952933907509,
|
|
"num_tokens": 1496957.0,
|
|
"step": 670
|
|
},
|
|
{
|
|
"entropy": 7.0134529113769535,
|
|
"epoch": 0.680786686838124,
|
|
"grad_norm": 1.1953125,
|
|
"learning_rate": 0.000337,
|
|
"loss": 6.6387,
|
|
"mean_token_accuracy": 0.10315296724438668,
|
|
"num_tokens": 1508336.0,
|
|
"step": 675
|
|
},
|
|
{
|
|
"entropy": 6.98244080543518,
|
|
"epoch": 0.6858295511850732,
|
|
"grad_norm": 0.97265625,
|
|
"learning_rate": 0.0003395,
|
|
"loss": 6.6118,
|
|
"mean_token_accuracy": 0.09836572855710983,
|
|
"num_tokens": 1520016.0,
|
|
"step": 680
|
|
},
|
|
{
|
|
"entropy": 6.9399433612823485,
|
|
"epoch": 0.6908724155320222,
|
|
"grad_norm": 0.9609375,
|
|
"learning_rate": 0.000342,
|
|
"loss": 6.4545,
|
|
"mean_token_accuracy": 0.10555818825960159,
|
|
"num_tokens": 1531877.0,
|
|
"step": 685
|
|
},
|
|
{
|
|
"entropy": 6.861659812927246,
|
|
"epoch": 0.6959152798789713,
|
|
"grad_norm": 1.09375,
|
|
"learning_rate": 0.00034449999999999997,
|
|
"loss": 6.4507,
|
|
"mean_token_accuracy": 0.10879426971077918,
|
|
"num_tokens": 1541553.0,
|
|
"step": 690
|
|
},
|
|
{
|
|
"entropy": 6.983822345733643,
|
|
"epoch": 0.7009581442259203,
|
|
"grad_norm": 1.015625,
|
|
"learning_rate": 0.000347,
|
|
"loss": 6.6146,
|
|
"mean_token_accuracy": 0.10361509844660759,
|
|
"num_tokens": 1552491.0,
|
|
"step": 695
|
|
},
|
|
{
|
|
"entropy": 6.899546098709107,
|
|
"epoch": 0.7060010085728694,
|
|
"grad_norm": 1.0078125,
|
|
"learning_rate": 0.0003495,
|
|
"loss": 6.5076,
|
|
"mean_token_accuracy": 0.10220511853694916,
|
|
"num_tokens": 1563695.0,
|
|
"step": 700
|
|
},
|
|
{
|
|
"entropy": 6.986747407913208,
|
|
"epoch": 0.7110438729198184,
|
|
"grad_norm": 1.140625,
|
|
"learning_rate": 0.000352,
|
|
"loss": 6.5457,
|
|
"mean_token_accuracy": 0.1040324755012989,
|
|
"num_tokens": 1573467.0,
|
|
"step": 705
|
|
},
|
|
{
|
|
"entropy": 6.899123620986939,
|
|
"epoch": 0.7160867372667675,
|
|
"grad_norm": 0.953125,
|
|
"learning_rate": 0.0003545,
|
|
"loss": 6.523,
|
|
"mean_token_accuracy": 0.1107451319694519,
|
|
"num_tokens": 1584367.0,
|
|
"step": 710
|
|
},
|
|
{
|
|
"entropy": 6.902140855789185,
|
|
"epoch": 0.7211296016137166,
|
|
"grad_norm": 1.0625,
|
|
"learning_rate": 0.000357,
|
|
"loss": 6.4647,
|
|
"mean_token_accuracy": 0.11195559501647949,
|
|
"num_tokens": 1595301.0,
|
|
"step": 715
|
|
},
|
|
{
|
|
"entropy": 6.908828496932983,
|
|
"epoch": 0.7261724659606656,
|
|
"grad_norm": 1.0234375,
|
|
"learning_rate": 0.0003595,
|
|
"loss": 6.5362,
|
|
"mean_token_accuracy": 0.10281007215380669,
|
|
"num_tokens": 1605916.0,
|
|
"step": 720
|
|
},
|
|
{
|
|
"entropy": 6.960860681533814,
|
|
"epoch": 0.7312153303076148,
|
|
"grad_norm": 1.1328125,
|
|
"learning_rate": 0.000362,
|
|
"loss": 6.5308,
|
|
"mean_token_accuracy": 0.10332813784480095,
|
|
"num_tokens": 1616455.0,
|
|
"step": 725
|
|
},
|
|
{
|
|
"entropy": 6.853022289276123,
|
|
"epoch": 0.7362581946545638,
|
|
"grad_norm": 1.0,
|
|
"learning_rate": 0.0003645,
|
|
"loss": 6.4979,
|
|
"mean_token_accuracy": 0.10751453042030334,
|
|
"num_tokens": 1627396.0,
|
|
"step": 730
|
|
},
|
|
{
|
|
"entropy": 7.0525219440460205,
|
|
"epoch": 0.7413010590015129,
|
|
"grad_norm": 1.1171875,
|
|
"learning_rate": 0.000367,
|
|
"loss": 6.6317,
|
|
"mean_token_accuracy": 0.09941819235682488,
|
|
"num_tokens": 1639073.0,
|
|
"step": 735
|
|
},
|
|
{
|
|
"entropy": 6.803125476837158,
|
|
"epoch": 0.7463439233484619,
|
|
"grad_norm": 1.03125,
|
|
"learning_rate": 0.0003695,
|
|
"loss": 6.4978,
|
|
"mean_token_accuracy": 0.10190683454275132,
|
|
"num_tokens": 1649787.0,
|
|
"step": 740
|
|
},
|
|
{
|
|
"entropy": 6.935564804077148,
|
|
"epoch": 0.751386787695411,
|
|
"grad_norm": 0.9765625,
|
|
"learning_rate": 0.000372,
|
|
"loss": 6.4383,
|
|
"mean_token_accuracy": 0.10585329979658127,
|
|
"num_tokens": 1658883.0,
|
|
"step": 745
|
|
},
|
|
{
|
|
"entropy": 6.7616808891296385,
|
|
"epoch": 0.75642965204236,
|
|
"grad_norm": 1.140625,
|
|
"learning_rate": 0.0003745,
|
|
"loss": 6.4186,
|
|
"mean_token_accuracy": 0.11256190687417984,
|
|
"num_tokens": 1669099.0,
|
|
"step": 750
|
|
},
|
|
{
|
|
"entropy": 6.824008131027222,
|
|
"epoch": 0.7614725163893091,
|
|
"grad_norm": 0.9453125,
|
|
"learning_rate": 0.000377,
|
|
"loss": 6.4199,
|
|
"mean_token_accuracy": 0.10491297394037247,
|
|
"num_tokens": 1680227.0,
|
|
"step": 755
|
|
},
|
|
{
|
|
"entropy": 6.830388927459717,
|
|
"epoch": 0.7665153807362582,
|
|
"grad_norm": 1.1015625,
|
|
"learning_rate": 0.0003795,
|
|
"loss": 6.4178,
|
|
"mean_token_accuracy": 0.10797212198376656,
|
|
"num_tokens": 1689901.0,
|
|
"step": 760
|
|
},
|
|
{
|
|
"entropy": 6.790120553970337,
|
|
"epoch": 0.7715582450832073,
|
|
"grad_norm": 1.0625,
|
|
"learning_rate": 0.000382,
|
|
"loss": 6.4932,
|
|
"mean_token_accuracy": 0.10378503128886223,
|
|
"num_tokens": 1700498.0,
|
|
"step": 765
|
|
},
|
|
{
|
|
"entropy": 6.854589986801147,
|
|
"epoch": 0.7766011094301564,
|
|
"grad_norm": 1.0703125,
|
|
"learning_rate": 0.0003845,
|
|
"loss": 6.3993,
|
|
"mean_token_accuracy": 0.1122807152569294,
|
|
"num_tokens": 1710715.0,
|
|
"step": 770
|
|
},
|
|
{
|
|
"entropy": 6.860204267501831,
|
|
"epoch": 0.7816439737771054,
|
|
"grad_norm": 1.046875,
|
|
"learning_rate": 0.00038700000000000003,
|
|
"loss": 6.5261,
|
|
"mean_token_accuracy": 0.106892429292202,
|
|
"num_tokens": 1723858.0,
|
|
"step": 775
|
|
},
|
|
{
|
|
"entropy": 6.753562068939209,
|
|
"epoch": 0.7866868381240545,
|
|
"grad_norm": 1.03125,
|
|
"learning_rate": 0.00038950000000000003,
|
|
"loss": 6.491,
|
|
"mean_token_accuracy": 0.10414958521723747,
|
|
"num_tokens": 1734678.0,
|
|
"step": 780
|
|
},
|
|
{
|
|
"entropy": 6.837999296188355,
|
|
"epoch": 0.7917297024710035,
|
|
"grad_norm": 1.0546875,
|
|
"learning_rate": 0.00039200000000000004,
|
|
"loss": 6.3922,
|
|
"mean_token_accuracy": 0.11342072263360023,
|
|
"num_tokens": 1743869.0,
|
|
"step": 785
|
|
},
|
|
{
|
|
"entropy": 6.905572748184204,
|
|
"epoch": 0.7967725668179526,
|
|
"grad_norm": 0.9609375,
|
|
"learning_rate": 0.00039450000000000005,
|
|
"loss": 6.4633,
|
|
"mean_token_accuracy": 0.10736536532640457,
|
|
"num_tokens": 1755266.0,
|
|
"step": 790
|
|
},
|
|
{
|
|
"entropy": 6.693229627609253,
|
|
"epoch": 0.8018154311649016,
|
|
"grad_norm": 1.0625,
|
|
"learning_rate": 0.00039700000000000005,
|
|
"loss": 6.4254,
|
|
"mean_token_accuracy": 0.1047304928302765,
|
|
"num_tokens": 1766089.0,
|
|
"step": 795
|
|
},
|
|
{
|
|
"entropy": 6.800949335098267,
|
|
"epoch": 0.8068582955118507,
|
|
"grad_norm": 1.28125,
|
|
"learning_rate": 0.0003995,
|
|
"loss": 6.441,
|
|
"mean_token_accuracy": 0.10456803143024444,
|
|
"num_tokens": 1776057.0,
|
|
"step": 800
|
|
},
|
|
{
|
|
"entropy": 6.839222145080567,
|
|
"epoch": 0.8119011598587998,
|
|
"grad_norm": 1.046875,
|
|
"learning_rate": 0.000402,
|
|
"loss": 6.4781,
|
|
"mean_token_accuracy": 0.10709379687905311,
|
|
"num_tokens": 1786729.0,
|
|
"step": 805
|
|
},
|
|
{
|
|
"entropy": 6.80726170539856,
|
|
"epoch": 0.8169440242057489,
|
|
"grad_norm": 1.1640625,
|
|
"learning_rate": 0.0004045,
|
|
"loss": 6.408,
|
|
"mean_token_accuracy": 0.10803752541542053,
|
|
"num_tokens": 1797708.0,
|
|
"step": 810
|
|
},
|
|
{
|
|
"entropy": 6.7316183090209964,
|
|
"epoch": 0.821986888552698,
|
|
"grad_norm": 0.96875,
|
|
"learning_rate": 0.00040699999999999997,
|
|
"loss": 6.4408,
|
|
"mean_token_accuracy": 0.11508865803480148,
|
|
"num_tokens": 1807326.0,
|
|
"step": 815
|
|
},
|
|
{
|
|
"entropy": 6.7365195751190186,
|
|
"epoch": 0.827029752899647,
|
|
"grad_norm": 0.984375,
|
|
"learning_rate": 0.0004095,
|
|
"loss": 6.3771,
|
|
"mean_token_accuracy": 0.11316760107874871,
|
|
"num_tokens": 1817266.0,
|
|
"step": 820
|
|
},
|
|
{
|
|
"entropy": 6.765561532974243,
|
|
"epoch": 0.8320726172465961,
|
|
"grad_norm": 0.890625,
|
|
"learning_rate": 0.000412,
|
|
"loss": 6.3975,
|
|
"mean_token_accuracy": 0.11348253637552261,
|
|
"num_tokens": 1827840.0,
|
|
"step": 825
|
|
},
|
|
{
|
|
"entropy": 6.61966724395752,
|
|
"epoch": 0.8371154815935451,
|
|
"grad_norm": 0.953125,
|
|
"learning_rate": 0.0004145,
|
|
"loss": 6.3751,
|
|
"mean_token_accuracy": 0.10820141658186913,
|
|
"num_tokens": 1838158.0,
|
|
"step": 830
|
|
},
|
|
{
|
|
"entropy": 6.799835348129273,
|
|
"epoch": 0.8421583459404942,
|
|
"grad_norm": 1.03125,
|
|
"learning_rate": 0.000417,
|
|
"loss": 6.4544,
|
|
"mean_token_accuracy": 0.10612540319561958,
|
|
"num_tokens": 1848317.0,
|
|
"step": 835
|
|
},
|
|
{
|
|
"entropy": 6.711783790588379,
|
|
"epoch": 0.8472012102874432,
|
|
"grad_norm": 1.0234375,
|
|
"learning_rate": 0.0004195,
|
|
"loss": 6.3855,
|
|
"mean_token_accuracy": 0.10564651787281036,
|
|
"num_tokens": 1858751.0,
|
|
"step": 840
|
|
},
|
|
{
|
|
"entropy": 6.7946678638458256,
|
|
"epoch": 0.8522440746343923,
|
|
"grad_norm": 0.96875,
|
|
"learning_rate": 0.000422,
|
|
"loss": 6.3903,
|
|
"mean_token_accuracy": 0.1101529210805893,
|
|
"num_tokens": 1870739.0,
|
|
"step": 845
|
|
},
|
|
{
|
|
"entropy": 6.8033802032470705,
|
|
"epoch": 0.8572869389813415,
|
|
"grad_norm": 1.046875,
|
|
"learning_rate": 0.0004245,
|
|
"loss": 6.5132,
|
|
"mean_token_accuracy": 0.11032365560531616,
|
|
"num_tokens": 1880995.0,
|
|
"step": 850
|
|
},
|
|
{
|
|
"entropy": 6.705736064910889,
|
|
"epoch": 0.8623298033282905,
|
|
"grad_norm": 1.09375,
|
|
"learning_rate": 0.000427,
|
|
"loss": 6.4509,
|
|
"mean_token_accuracy": 0.10846404731273651,
|
|
"num_tokens": 1892633.0,
|
|
"step": 855
|
|
},
|
|
{
|
|
"entropy": 6.8639147758483885,
|
|
"epoch": 0.8673726676752396,
|
|
"grad_norm": 1.0234375,
|
|
"learning_rate": 0.0004295,
|
|
"loss": 6.5176,
|
|
"mean_token_accuracy": 0.10401368141174316,
|
|
"num_tokens": 1904802.0,
|
|
"step": 860
|
|
},
|
|
{
|
|
"entropy": 6.778040790557862,
|
|
"epoch": 0.8724155320221886,
|
|
"grad_norm": 0.91015625,
|
|
"learning_rate": 0.000432,
|
|
"loss": 6.5199,
|
|
"mean_token_accuracy": 0.10730656683444977,
|
|
"num_tokens": 1915755.0,
|
|
"step": 865
|
|
},
|
|
{
|
|
"entropy": 6.688451814651489,
|
|
"epoch": 0.8774583963691377,
|
|
"grad_norm": 0.9453125,
|
|
"learning_rate": 0.0004345,
|
|
"loss": 6.2652,
|
|
"mean_token_accuracy": 0.11115367412567138,
|
|
"num_tokens": 1926098.0,
|
|
"step": 870
|
|
},
|
|
{
|
|
"entropy": 6.744540882110596,
|
|
"epoch": 0.8825012607160867,
|
|
"grad_norm": 1.03125,
|
|
"learning_rate": 0.000437,
|
|
"loss": 6.4883,
|
|
"mean_token_accuracy": 0.11085206493735314,
|
|
"num_tokens": 1936607.0,
|
|
"step": 875
|
|
},
|
|
{
|
|
"entropy": 6.703429174423218,
|
|
"epoch": 0.8875441250630358,
|
|
"grad_norm": 1.0234375,
|
|
"learning_rate": 0.0004395,
|
|
"loss": 6.377,
|
|
"mean_token_accuracy": 0.10893332436680794,
|
|
"num_tokens": 1948596.0,
|
|
"step": 880
|
|
},
|
|
{
|
|
"entropy": 6.669404411315918,
|
|
"epoch": 0.8925869894099848,
|
|
"grad_norm": 0.98828125,
|
|
"learning_rate": 0.000442,
|
|
"loss": 6.4526,
|
|
"mean_token_accuracy": 0.11500444263219833,
|
|
"num_tokens": 1960322.0,
|
|
"step": 885
|
|
},
|
|
{
|
|
"entropy": 6.710722208023071,
|
|
"epoch": 0.8976298537569339,
|
|
"grad_norm": 0.953125,
|
|
"learning_rate": 0.0004445,
|
|
"loss": 6.379,
|
|
"mean_token_accuracy": 0.11063228026032448,
|
|
"num_tokens": 1971052.0,
|
|
"step": 890
|
|
},
|
|
{
|
|
"entropy": 6.781415891647339,
|
|
"epoch": 0.902672718103883,
|
|
"grad_norm": 1.0078125,
|
|
"learning_rate": 0.000447,
|
|
"loss": 6.3932,
|
|
"mean_token_accuracy": 0.10573994740843773,
|
|
"num_tokens": 1980917.0,
|
|
"step": 895
|
|
},
|
|
{
|
|
"entropy": 6.649546766281128,
|
|
"epoch": 0.9077155824508321,
|
|
"grad_norm": 1.234375,
|
|
"learning_rate": 0.00044950000000000003,
|
|
"loss": 6.3419,
|
|
"mean_token_accuracy": 0.11238374263048172,
|
|
"num_tokens": 1991035.0,
|
|
"step": 900
|
|
},
|
|
{
|
|
"entropy": 6.629901695251465,
|
|
"epoch": 0.9127584467977812,
|
|
"grad_norm": 1.015625,
|
|
"learning_rate": 0.00045200000000000004,
|
|
"loss": 6.3659,
|
|
"mean_token_accuracy": 0.11216280534863472,
|
|
"num_tokens": 2000587.0,
|
|
"step": 905
|
|
},
|
|
{
|
|
"entropy": 6.720604801177979,
|
|
"epoch": 0.9178013111447302,
|
|
"grad_norm": 1.109375,
|
|
"learning_rate": 0.00045450000000000004,
|
|
"loss": 6.3292,
|
|
"mean_token_accuracy": 0.11438163220882416,
|
|
"num_tokens": 2011087.0,
|
|
"step": 910
|
|
},
|
|
{
|
|
"entropy": 6.651438570022583,
|
|
"epoch": 0.9228441754916793,
|
|
"grad_norm": 1.109375,
|
|
"learning_rate": 0.00045700000000000005,
|
|
"loss": 6.3782,
|
|
"mean_token_accuracy": 0.10697766989469529,
|
|
"num_tokens": 2023472.0,
|
|
"step": 915
|
|
},
|
|
{
|
|
"entropy": 6.697429943084717,
|
|
"epoch": 0.9278870398386283,
|
|
"grad_norm": 0.98828125,
|
|
"learning_rate": 0.00045950000000000006,
|
|
"loss": 6.3639,
|
|
"mean_token_accuracy": 0.11164266243577003,
|
|
"num_tokens": 2033783.0,
|
|
"step": 920
|
|
},
|
|
{
|
|
"entropy": 6.520638275146484,
|
|
"epoch": 0.9329299041855774,
|
|
"grad_norm": 0.86328125,
|
|
"learning_rate": 0.000462,
|
|
"loss": 6.3795,
|
|
"mean_token_accuracy": 0.11479249447584153,
|
|
"num_tokens": 2045882.0,
|
|
"step": 925
|
|
},
|
|
{
|
|
"entropy": 6.7621557235717775,
|
|
"epoch": 0.9379727685325264,
|
|
"grad_norm": 0.87109375,
|
|
"learning_rate": 0.0004645,
|
|
"loss": 6.3894,
|
|
"mean_token_accuracy": 0.10624322891235352,
|
|
"num_tokens": 2056549.0,
|
|
"step": 930
|
|
},
|
|
{
|
|
"entropy": 6.654461240768432,
|
|
"epoch": 0.9430156328794755,
|
|
"grad_norm": 1.0390625,
|
|
"learning_rate": 0.000467,
|
|
"loss": 6.3845,
|
|
"mean_token_accuracy": 0.11624199599027633,
|
|
"num_tokens": 2067540.0,
|
|
"step": 935
|
|
},
|
|
{
|
|
"entropy": 6.602401876449585,
|
|
"epoch": 0.9480584972264247,
|
|
"grad_norm": 1.1875,
|
|
"learning_rate": 0.0004695,
|
|
"loss": 6.2956,
|
|
"mean_token_accuracy": 0.11275828555226326,
|
|
"num_tokens": 2077386.0,
|
|
"step": 940
|
|
},
|
|
{
|
|
"entropy": 6.646829652786255,
|
|
"epoch": 0.9531013615733737,
|
|
"grad_norm": 1.125,
|
|
"learning_rate": 0.000472,
|
|
"loss": 6.2697,
|
|
"mean_token_accuracy": 0.11356851384043694,
|
|
"num_tokens": 2087777.0,
|
|
"step": 945
|
|
},
|
|
{
|
|
"entropy": 6.494188833236694,
|
|
"epoch": 0.9581442259203228,
|
|
"grad_norm": 1.09375,
|
|
"learning_rate": 0.0004745,
|
|
"loss": 6.2604,
|
|
"mean_token_accuracy": 0.11282371133565902,
|
|
"num_tokens": 2099071.0,
|
|
"step": 950
|
|
},
|
|
{
|
|
"entropy": 6.53692774772644,
|
|
"epoch": 0.9631870902672718,
|
|
"grad_norm": 1.1171875,
|
|
"learning_rate": 0.000477,
|
|
"loss": 6.3258,
|
|
"mean_token_accuracy": 0.1121332198381424,
|
|
"num_tokens": 2109168.0,
|
|
"step": 955
|
|
},
|
|
{
|
|
"entropy": 6.835064029693603,
|
|
"epoch": 0.9682299546142209,
|
|
"grad_norm": 1.1171875,
|
|
"learning_rate": 0.0004795,
|
|
"loss": 6.425,
|
|
"mean_token_accuracy": 0.10913640931248665,
|
|
"num_tokens": 2120696.0,
|
|
"step": 960
|
|
},
|
|
{
|
|
"entropy": 6.537368297576904,
|
|
"epoch": 0.9732728189611699,
|
|
"grad_norm": 1.171875,
|
|
"learning_rate": 0.000482,
|
|
"loss": 6.371,
|
|
"mean_token_accuracy": 0.10968417152762414,
|
|
"num_tokens": 2132427.0,
|
|
"step": 965
|
|
},
|
|
{
|
|
"entropy": 6.598920631408691,
|
|
"epoch": 0.978315683308119,
|
|
"grad_norm": 1.0703125,
|
|
"learning_rate": 0.0004845,
|
|
"loss": 6.3618,
|
|
"mean_token_accuracy": 0.11373768299818039,
|
|
"num_tokens": 2143120.0,
|
|
"step": 970
|
|
},
|
|
{
|
|
"entropy": 6.740956020355225,
|
|
"epoch": 0.983358547655068,
|
|
"grad_norm": 0.94921875,
|
|
"learning_rate": 0.000487,
|
|
"loss": 6.349,
|
|
"mean_token_accuracy": 0.10921334773302079,
|
|
"num_tokens": 2154154.0,
|
|
"step": 975
|
|
},
|
|
{
|
|
"entropy": 6.541984796524048,
|
|
"epoch": 0.9884014120020171,
|
|
"grad_norm": 0.8984375,
|
|
"learning_rate": 0.0004895,
|
|
"loss": 6.4499,
|
|
"mean_token_accuracy": 0.10598269626498222,
|
|
"num_tokens": 2167000.0,
|
|
"step": 980
|
|
},
|
|
{
|
|
"entropy": 6.663674449920654,
|
|
"epoch": 0.9934442763489663,
|
|
"grad_norm": 1.0390625,
|
|
"learning_rate": 0.000492,
|
|
"loss": 6.3823,
|
|
"mean_token_accuracy": 0.11338229328393937,
|
|
"num_tokens": 2178100.0,
|
|
"step": 985
|
|
},
|
|
{
|
|
"entropy": 6.6093058586120605,
|
|
"epoch": 0.9984871406959153,
|
|
"grad_norm": 1.09375,
|
|
"learning_rate": 0.0004945,
|
|
"loss": 6.331,
|
|
"mean_token_accuracy": 0.10886227712035179,
|
|
"num_tokens": 2189063.0,
|
|
"step": 990
|
|
},
|
|
{
|
|
"entropy": 6.706958293914795,
|
|
"epoch": 1.0030257186081695,
|
|
"grad_norm": 1.046875,
|
|
"learning_rate": 0.000497,
|
|
"loss": 6.2723,
|
|
"mean_token_accuracy": 0.11505810419718425,
|
|
"num_tokens": 2198302.0,
|
|
"step": 995
|
|
},
|
|
{
|
|
"entropy": 6.57634973526001,
|
|
"epoch": 1.0080685829551186,
|
|
"grad_norm": 1.0859375,
|
|
"learning_rate": 0.0004995,
|
|
"loss": 6.2311,
|
|
"mean_token_accuracy": 0.1161844253540039,
|
|
"num_tokens": 2209648.0,
|
|
"step": 1000
|
|
},
|
|
{
|
|
"epoch": 1.0080685829551186,
|
|
"eval_entropy": 6.5141722082181746,
|
|
"eval_loss": 6.3961334228515625,
|
|
"eval_mean_token_accuracy": 0.11549569743166443,
|
|
"eval_num_tokens": 2209648.0,
|
|
"eval_runtime": 2.054,
|
|
"eval_samples_per_second": 1527.74,
|
|
"eval_steps_per_second": 191.333,
|
|
"step": 1000
|
|
}
|
|
],
|
|
"logging_steps": 5,
|
|
"max_steps": 9910,
|
|
"num_input_tokens_seen": 0,
|
|
"num_train_epochs": 10,
|
|
"save_steps": 500,
|
|
"stateful_callbacks": {
|
|
"TrainerControl": {
|
|
"args": {
|
|
"should_epoch_stop": false,
|
|
"should_evaluate": false,
|
|
"should_log": false,
|
|
"should_save": true,
|
|
"should_training_stop": false
|
|
},
|
|
"attributes": {}
|
|
}
|
|
},
|
|
"total_flos": 547568094320640.0,
|
|
"train_batch_size": 16,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|