1035 lines
27 KiB
JSON
1035 lines
27 KiB
JSON
{
|
|
"best_global_step": null,
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 0.03548490117455023,
|
|
"eval_steps": 500,
|
|
"global_step": 500,
|
|
"is_hyper_param_search": false,
|
|
"is_local_process_zero": true,
|
|
"is_world_process_zero": true,
|
|
"log_history": [
|
|
{
|
|
"entropy": 10.691987895965577,
|
|
"epoch": 0.00035484901174550227,
|
|
"grad_norm": 11.625,
|
|
"learning_rate": 2e-06,
|
|
"loss": 10.8366,
|
|
"mean_token_accuracy": 0.0,
|
|
"num_tokens": 11280.0,
|
|
"step": 5
|
|
},
|
|
{
|
|
"entropy": 10.691946220397949,
|
|
"epoch": 0.0007096980234910045,
|
|
"grad_norm": 11.5625,
|
|
"learning_rate": 4.5e-06,
|
|
"loss": 10.7735,
|
|
"mean_token_accuracy": 0.00010460250778123736,
|
|
"num_tokens": 21887.0,
|
|
"step": 10
|
|
},
|
|
{
|
|
"entropy": 10.69083604812622,
|
|
"epoch": 0.001064547035236507,
|
|
"grad_norm": 9.6875,
|
|
"learning_rate": 7e-06,
|
|
"loss": 10.5654,
|
|
"mean_token_accuracy": 0.02854402889497578,
|
|
"num_tokens": 32911.0,
|
|
"step": 15
|
|
},
|
|
{
|
|
"entropy": 10.66428632736206,
|
|
"epoch": 0.001419396046982009,
|
|
"grad_norm": 6.0625,
|
|
"learning_rate": 9.5e-06,
|
|
"loss": 10.3144,
|
|
"mean_token_accuracy": 0.04550958257168532,
|
|
"num_tokens": 43994.0,
|
|
"step": 20
|
|
},
|
|
{
|
|
"entropy": 10.494576263427735,
|
|
"epoch": 0.0017742450587275114,
|
|
"grad_norm": 4.28125,
|
|
"learning_rate": 1.2e-05,
|
|
"loss": 10.105,
|
|
"mean_token_accuracy": 0.041560862213373184,
|
|
"num_tokens": 54342.0,
|
|
"step": 25
|
|
},
|
|
{
|
|
"entropy": 10.533498859405517,
|
|
"epoch": 0.002129094070473014,
|
|
"grad_norm": 2.921875,
|
|
"learning_rate": 1.4500000000000002e-05,
|
|
"loss": 9.9509,
|
|
"mean_token_accuracy": 0.044634624011814596,
|
|
"num_tokens": 64768.0,
|
|
"step": 30
|
|
},
|
|
{
|
|
"entropy": 10.53185338973999,
|
|
"epoch": 0.002483943082218516,
|
|
"grad_norm": 2.59375,
|
|
"learning_rate": 1.7000000000000003e-05,
|
|
"loss": 9.9269,
|
|
"mean_token_accuracy": 0.04376727268099785,
|
|
"num_tokens": 75530.0,
|
|
"step": 35
|
|
},
|
|
{
|
|
"entropy": 10.487653255462646,
|
|
"epoch": 0.002838792093964018,
|
|
"grad_norm": 2.40625,
|
|
"learning_rate": 1.95e-05,
|
|
"loss": 9.8511,
|
|
"mean_token_accuracy": 0.045070053450763224,
|
|
"num_tokens": 85983.0,
|
|
"step": 40
|
|
},
|
|
{
|
|
"entropy": 10.522008514404297,
|
|
"epoch": 0.0031936411057095207,
|
|
"grad_norm": 2.28125,
|
|
"learning_rate": 2.2e-05,
|
|
"loss": 9.8344,
|
|
"mean_token_accuracy": 0.04323017336428166,
|
|
"num_tokens": 96168.0,
|
|
"step": 45
|
|
},
|
|
{
|
|
"entropy": 10.482469272613525,
|
|
"epoch": 0.0035484901174550228,
|
|
"grad_norm": 1.8046875,
|
|
"learning_rate": 2.4500000000000003e-05,
|
|
"loss": 9.7891,
|
|
"mean_token_accuracy": 0.05232359655201435,
|
|
"num_tokens": 107899.0,
|
|
"step": 50
|
|
},
|
|
{
|
|
"entropy": 10.414291763305664,
|
|
"epoch": 0.0039033391292005253,
|
|
"grad_norm": 2.03125,
|
|
"learning_rate": 2.7e-05,
|
|
"loss": 9.7448,
|
|
"mean_token_accuracy": 0.05363052934408188,
|
|
"num_tokens": 119073.0,
|
|
"step": 55
|
|
},
|
|
{
|
|
"entropy": 10.483110237121583,
|
|
"epoch": 0.004258188140946028,
|
|
"grad_norm": 2.03125,
|
|
"learning_rate": 2.95e-05,
|
|
"loss": 9.7062,
|
|
"mean_token_accuracy": 0.05208716280758381,
|
|
"num_tokens": 129149.0,
|
|
"step": 60
|
|
},
|
|
{
|
|
"entropy": 10.321412563323975,
|
|
"epoch": 0.00461303715269153,
|
|
"grad_norm": 1.890625,
|
|
"learning_rate": 3.2e-05,
|
|
"loss": 9.5992,
|
|
"mean_token_accuracy": 0.05521074235439301,
|
|
"num_tokens": 139045.0,
|
|
"step": 65
|
|
},
|
|
{
|
|
"entropy": 10.363921737670898,
|
|
"epoch": 0.004967886164437032,
|
|
"grad_norm": 1.859375,
|
|
"learning_rate": 3.4500000000000005e-05,
|
|
"loss": 9.6115,
|
|
"mean_token_accuracy": 0.054533609002828595,
|
|
"num_tokens": 150251.0,
|
|
"step": 70
|
|
},
|
|
{
|
|
"entropy": 10.423143291473389,
|
|
"epoch": 0.005322735176182534,
|
|
"grad_norm": 1.984375,
|
|
"learning_rate": 3.7e-05,
|
|
"loss": 9.5269,
|
|
"mean_token_accuracy": 0.05440324060618877,
|
|
"num_tokens": 160230.0,
|
|
"step": 75
|
|
},
|
|
{
|
|
"entropy": 10.341320323944093,
|
|
"epoch": 0.005677584187928036,
|
|
"grad_norm": 1.546875,
|
|
"learning_rate": 3.95e-05,
|
|
"loss": 9.478,
|
|
"mean_token_accuracy": 0.05366469696164131,
|
|
"num_tokens": 170692.0,
|
|
"step": 80
|
|
},
|
|
{
|
|
"entropy": 10.290056896209716,
|
|
"epoch": 0.006032433199673539,
|
|
"grad_norm": 2.078125,
|
|
"learning_rate": 4.2000000000000004e-05,
|
|
"loss": 9.3959,
|
|
"mean_token_accuracy": 0.05197730958461762,
|
|
"num_tokens": 180616.0,
|
|
"step": 85
|
|
},
|
|
{
|
|
"entropy": 10.320787811279297,
|
|
"epoch": 0.006387282211419041,
|
|
"grad_norm": 1.7421875,
|
|
"learning_rate": 4.45e-05,
|
|
"loss": 9.3057,
|
|
"mean_token_accuracy": 0.05711167939007282,
|
|
"num_tokens": 191016.0,
|
|
"step": 90
|
|
},
|
|
{
|
|
"entropy": 10.258602237701416,
|
|
"epoch": 0.0067421312231645434,
|
|
"grad_norm": 1.640625,
|
|
"learning_rate": 4.7000000000000004e-05,
|
|
"loss": 9.2578,
|
|
"mean_token_accuracy": 0.05482340231537819,
|
|
"num_tokens": 202433.0,
|
|
"step": 95
|
|
},
|
|
{
|
|
"entropy": 10.235629463195801,
|
|
"epoch": 0.0070969802349100455,
|
|
"grad_norm": 1.40625,
|
|
"learning_rate": 4.9500000000000004e-05,
|
|
"loss": 9.2401,
|
|
"mean_token_accuracy": 0.05469023734331131,
|
|
"num_tokens": 214586.0,
|
|
"step": 100
|
|
},
|
|
{
|
|
"entropy": 10.223749732971191,
|
|
"epoch": 0.007451829246655548,
|
|
"grad_norm": 1.484375,
|
|
"learning_rate": 5.2e-05,
|
|
"loss": 9.0977,
|
|
"mean_token_accuracy": 0.05747384466230869,
|
|
"num_tokens": 225155.0,
|
|
"step": 105
|
|
},
|
|
{
|
|
"entropy": 10.136597728729248,
|
|
"epoch": 0.007806678258401051,
|
|
"grad_norm": 1.4609375,
|
|
"learning_rate": 5.45e-05,
|
|
"loss": 8.9983,
|
|
"mean_token_accuracy": 0.05849210806190967,
|
|
"num_tokens": 236050.0,
|
|
"step": 110
|
|
},
|
|
{
|
|
"entropy": 10.020172691345214,
|
|
"epoch": 0.008161527270146553,
|
|
"grad_norm": 1.375,
|
|
"learning_rate": 5.7e-05,
|
|
"loss": 8.8941,
|
|
"mean_token_accuracy": 0.05953489430248737,
|
|
"num_tokens": 245825.0,
|
|
"step": 115
|
|
},
|
|
{
|
|
"entropy": 9.979390907287598,
|
|
"epoch": 0.008516376281892056,
|
|
"grad_norm": 1.9453125,
|
|
"learning_rate": 5.9499999999999996e-05,
|
|
"loss": 8.8174,
|
|
"mean_token_accuracy": 0.06152731701731682,
|
|
"num_tokens": 256752.0,
|
|
"step": 120
|
|
},
|
|
{
|
|
"entropy": 9.866498374938965,
|
|
"epoch": 0.008871225293637557,
|
|
"grad_norm": 1.15625,
|
|
"learning_rate": 6.2e-05,
|
|
"loss": 8.802,
|
|
"mean_token_accuracy": 0.05181486271321774,
|
|
"num_tokens": 267603.0,
|
|
"step": 125
|
|
},
|
|
{
|
|
"entropy": 9.726642417907716,
|
|
"epoch": 0.00922607430538306,
|
|
"grad_norm": 1.1875,
|
|
"learning_rate": 6.450000000000001e-05,
|
|
"loss": 8.7071,
|
|
"mean_token_accuracy": 0.05791006051003933,
|
|
"num_tokens": 277478.0,
|
|
"step": 130
|
|
},
|
|
{
|
|
"entropy": 9.658979320526123,
|
|
"epoch": 0.009580923317128561,
|
|
"grad_norm": 1.2109375,
|
|
"learning_rate": 6.7e-05,
|
|
"loss": 8.66,
|
|
"mean_token_accuracy": 0.05634133592247963,
|
|
"num_tokens": 289710.0,
|
|
"step": 135
|
|
},
|
|
{
|
|
"entropy": 9.384044075012207,
|
|
"epoch": 0.009935772328874064,
|
|
"grad_norm": 1.1796875,
|
|
"learning_rate": 6.950000000000001e-05,
|
|
"loss": 8.5346,
|
|
"mean_token_accuracy": 0.05907244272530079,
|
|
"num_tokens": 299295.0,
|
|
"step": 140
|
|
},
|
|
{
|
|
"entropy": 9.350571346282958,
|
|
"epoch": 0.010290621340619567,
|
|
"grad_norm": 0.9375,
|
|
"learning_rate": 7.2e-05,
|
|
"loss": 8.5165,
|
|
"mean_token_accuracy": 0.05372942052781582,
|
|
"num_tokens": 310544.0,
|
|
"step": 145
|
|
},
|
|
{
|
|
"entropy": 9.217219734191895,
|
|
"epoch": 0.010645470352365068,
|
|
"grad_norm": 1.09375,
|
|
"learning_rate": 7.45e-05,
|
|
"loss": 8.4485,
|
|
"mean_token_accuracy": 0.05611223764717579,
|
|
"num_tokens": 321558.0,
|
|
"step": 150
|
|
},
|
|
{
|
|
"entropy": 9.077949619293213,
|
|
"epoch": 0.011000319364110571,
|
|
"grad_norm": 1.203125,
|
|
"learning_rate": 7.7e-05,
|
|
"loss": 8.44,
|
|
"mean_token_accuracy": 0.05572409965097904,
|
|
"num_tokens": 333265.0,
|
|
"step": 155
|
|
},
|
|
{
|
|
"entropy": 8.919435977935791,
|
|
"epoch": 0.011355168375856073,
|
|
"grad_norm": 1.1328125,
|
|
"learning_rate": 7.950000000000001e-05,
|
|
"loss": 8.3707,
|
|
"mean_token_accuracy": 0.060116075351834296,
|
|
"num_tokens": 343406.0,
|
|
"step": 160
|
|
},
|
|
{
|
|
"entropy": 8.981222343444824,
|
|
"epoch": 0.011710017387601575,
|
|
"grad_norm": 1.140625,
|
|
"learning_rate": 8.2e-05,
|
|
"loss": 8.4388,
|
|
"mean_token_accuracy": 0.05228844713419676,
|
|
"num_tokens": 354934.0,
|
|
"step": 165
|
|
},
|
|
{
|
|
"entropy": 8.788821792602539,
|
|
"epoch": 0.012064866399347078,
|
|
"grad_norm": 0.95703125,
|
|
"learning_rate": 8.450000000000001e-05,
|
|
"loss": 8.3627,
|
|
"mean_token_accuracy": 0.05599985755980015,
|
|
"num_tokens": 364991.0,
|
|
"step": 170
|
|
},
|
|
{
|
|
"entropy": 8.838678550720214,
|
|
"epoch": 0.01241971541109258,
|
|
"grad_norm": 0.87109375,
|
|
"learning_rate": 8.7e-05,
|
|
"loss": 8.3174,
|
|
"mean_token_accuracy": 0.05698671489953995,
|
|
"num_tokens": 375362.0,
|
|
"step": 175
|
|
},
|
|
{
|
|
"entropy": 8.669477367401123,
|
|
"epoch": 0.012774564422838083,
|
|
"grad_norm": 1.046875,
|
|
"learning_rate": 8.95e-05,
|
|
"loss": 8.2929,
|
|
"mean_token_accuracy": 0.06303150560706854,
|
|
"num_tokens": 386758.0,
|
|
"step": 180
|
|
},
|
|
{
|
|
"entropy": 8.66919822692871,
|
|
"epoch": 0.013129413434583584,
|
|
"grad_norm": 0.9140625,
|
|
"learning_rate": 9.2e-05,
|
|
"loss": 8.304,
|
|
"mean_token_accuracy": 0.061275123804807666,
|
|
"num_tokens": 397073.0,
|
|
"step": 185
|
|
},
|
|
{
|
|
"entropy": 8.68841199874878,
|
|
"epoch": 0.013484262446329087,
|
|
"grad_norm": 0.85546875,
|
|
"learning_rate": 9.45e-05,
|
|
"loss": 8.4007,
|
|
"mean_token_accuracy": 0.05668871849775314,
|
|
"num_tokens": 408098.0,
|
|
"step": 190
|
|
},
|
|
{
|
|
"entropy": 8.577757740020752,
|
|
"epoch": 0.01383911145807459,
|
|
"grad_norm": 1.2109375,
|
|
"learning_rate": 9.7e-05,
|
|
"loss": 8.2425,
|
|
"mean_token_accuracy": 0.06331006735563278,
|
|
"num_tokens": 418910.0,
|
|
"step": 195
|
|
},
|
|
{
|
|
"entropy": 8.616193675994873,
|
|
"epoch": 0.014193960469820091,
|
|
"grad_norm": 1.0546875,
|
|
"learning_rate": 9.95e-05,
|
|
"loss": 8.3266,
|
|
"mean_token_accuracy": 0.059384917840361595,
|
|
"num_tokens": 429328.0,
|
|
"step": 200
|
|
},
|
|
{
|
|
"entropy": 8.564495944976807,
|
|
"epoch": 0.014548809481565594,
|
|
"grad_norm": 1.96875,
|
|
"learning_rate": 0.000102,
|
|
"loss": 8.286,
|
|
"mean_token_accuracy": 0.058925506100058554,
|
|
"num_tokens": 440293.0,
|
|
"step": 205
|
|
},
|
|
{
|
|
"entropy": 8.546427249908447,
|
|
"epoch": 0.014903658493311095,
|
|
"grad_norm": 1.28125,
|
|
"learning_rate": 0.00010449999999999999,
|
|
"loss": 8.2513,
|
|
"mean_token_accuracy": 0.057647600024938586,
|
|
"num_tokens": 451509.0,
|
|
"step": 210
|
|
},
|
|
{
|
|
"entropy": 8.532197761535645,
|
|
"epoch": 0.015258507505056598,
|
|
"grad_norm": 0.91015625,
|
|
"learning_rate": 0.000107,
|
|
"loss": 8.258,
|
|
"mean_token_accuracy": 0.058545221015810965,
|
|
"num_tokens": 461660.0,
|
|
"step": 215
|
|
},
|
|
{
|
|
"entropy": 8.52504529953003,
|
|
"epoch": 0.015613356516802101,
|
|
"grad_norm": 1.1953125,
|
|
"learning_rate": 0.0001095,
|
|
"loss": 8.2467,
|
|
"mean_token_accuracy": 0.0674739558249712,
|
|
"num_tokens": 471362.0,
|
|
"step": 220
|
|
},
|
|
{
|
|
"entropy": 8.518447113037109,
|
|
"epoch": 0.015968205528547604,
|
|
"grad_norm": 1.15625,
|
|
"learning_rate": 0.000112,
|
|
"loss": 8.1115,
|
|
"mean_token_accuracy": 0.07247221358120441,
|
|
"num_tokens": 482036.0,
|
|
"step": 225
|
|
},
|
|
{
|
|
"entropy": 8.392018604278565,
|
|
"epoch": 0.016323054540293105,
|
|
"grad_norm": 1.0703125,
|
|
"learning_rate": 0.0001145,
|
|
"loss": 8.2038,
|
|
"mean_token_accuracy": 0.06354649700224399,
|
|
"num_tokens": 492633.0,
|
|
"step": 230
|
|
},
|
|
{
|
|
"entropy": 8.478435707092284,
|
|
"epoch": 0.016677903552038607,
|
|
"grad_norm": 1.140625,
|
|
"learning_rate": 0.00011700000000000001,
|
|
"loss": 8.2366,
|
|
"mean_token_accuracy": 0.06092475391924381,
|
|
"num_tokens": 502613.0,
|
|
"step": 235
|
|
},
|
|
{
|
|
"entropy": 8.442237091064452,
|
|
"epoch": 0.01703275256378411,
|
|
"grad_norm": 1.1328125,
|
|
"learning_rate": 0.00011949999999999999,
|
|
"loss": 8.26,
|
|
"mean_token_accuracy": 0.06470727063715458,
|
|
"num_tokens": 513256.0,
|
|
"step": 240
|
|
},
|
|
{
|
|
"entropy": 8.596701526641846,
|
|
"epoch": 0.017387601575529613,
|
|
"grad_norm": 1.0625,
|
|
"learning_rate": 0.000122,
|
|
"loss": 8.2988,
|
|
"mean_token_accuracy": 0.061711058393120764,
|
|
"num_tokens": 523355.0,
|
|
"step": 245
|
|
},
|
|
{
|
|
"entropy": 8.451476860046387,
|
|
"epoch": 0.017742450587275114,
|
|
"grad_norm": 1.015625,
|
|
"learning_rate": 0.0001245,
|
|
"loss": 8.2216,
|
|
"mean_token_accuracy": 0.06455190740525722,
|
|
"num_tokens": 535195.0,
|
|
"step": 250
|
|
},
|
|
{
|
|
"entropy": 8.469265937805176,
|
|
"epoch": 0.018097299599020615,
|
|
"grad_norm": 1.046875,
|
|
"learning_rate": 0.000127,
|
|
"loss": 8.1852,
|
|
"mean_token_accuracy": 0.06743233427405357,
|
|
"num_tokens": 544831.0,
|
|
"step": 255
|
|
},
|
|
{
|
|
"entropy": 8.445653438568115,
|
|
"epoch": 0.01845214861076612,
|
|
"grad_norm": 1.2421875,
|
|
"learning_rate": 0.0001295,
|
|
"loss": 8.2694,
|
|
"mean_token_accuracy": 0.059878384508192536,
|
|
"num_tokens": 555835.0,
|
|
"step": 260
|
|
},
|
|
{
|
|
"entropy": 8.478738117218018,
|
|
"epoch": 0.01880699762251162,
|
|
"grad_norm": 1.1796875,
|
|
"learning_rate": 0.000132,
|
|
"loss": 8.2247,
|
|
"mean_token_accuracy": 0.06131057888269424,
|
|
"num_tokens": 566875.0,
|
|
"step": 265
|
|
},
|
|
{
|
|
"entropy": 8.375457763671875,
|
|
"epoch": 0.019161846634257122,
|
|
"grad_norm": 1.09375,
|
|
"learning_rate": 0.00013450000000000002,
|
|
"loss": 8.237,
|
|
"mean_token_accuracy": 0.06333270035684109,
|
|
"num_tokens": 577462.0,
|
|
"step": 270
|
|
},
|
|
{
|
|
"entropy": 8.501999855041504,
|
|
"epoch": 0.019516695646002627,
|
|
"grad_norm": 1.0703125,
|
|
"learning_rate": 0.00013700000000000002,
|
|
"loss": 8.1721,
|
|
"mean_token_accuracy": 0.06697900146245957,
|
|
"num_tokens": 587569.0,
|
|
"step": 275
|
|
},
|
|
{
|
|
"entropy": 8.420077037811279,
|
|
"epoch": 0.019871544657748128,
|
|
"grad_norm": 0.921875,
|
|
"learning_rate": 0.0001395,
|
|
"loss": 8.2355,
|
|
"mean_token_accuracy": 0.05800015553832054,
|
|
"num_tokens": 599322.0,
|
|
"step": 280
|
|
},
|
|
{
|
|
"entropy": 8.452305221557618,
|
|
"epoch": 0.02022639366949363,
|
|
"grad_norm": 0.921875,
|
|
"learning_rate": 0.00014199999999999998,
|
|
"loss": 8.2397,
|
|
"mean_token_accuracy": 0.06376027651131153,
|
|
"num_tokens": 610060.0,
|
|
"step": 285
|
|
},
|
|
{
|
|
"entropy": 8.481741237640382,
|
|
"epoch": 0.020581242681239134,
|
|
"grad_norm": 1.234375,
|
|
"learning_rate": 0.0001445,
|
|
"loss": 8.295,
|
|
"mean_token_accuracy": 0.06520596966147423,
|
|
"num_tokens": 620091.0,
|
|
"step": 290
|
|
},
|
|
{
|
|
"entropy": 8.525186729431152,
|
|
"epoch": 0.020936091692984635,
|
|
"grad_norm": 2.140625,
|
|
"learning_rate": 0.000147,
|
|
"loss": 8.12,
|
|
"mean_token_accuracy": 0.0748389147222042,
|
|
"num_tokens": 631440.0,
|
|
"step": 295
|
|
},
|
|
{
|
|
"entropy": 8.37361249923706,
|
|
"epoch": 0.021290940704730137,
|
|
"grad_norm": 1.015625,
|
|
"learning_rate": 0.0001495,
|
|
"loss": 8.0801,
|
|
"mean_token_accuracy": 0.06787928678095341,
|
|
"num_tokens": 641374.0,
|
|
"step": 300
|
|
},
|
|
{
|
|
"entropy": 8.305605411529541,
|
|
"epoch": 0.02164578971647564,
|
|
"grad_norm": 1.0,
|
|
"learning_rate": 0.000152,
|
|
"loss": 8.1269,
|
|
"mean_token_accuracy": 0.06664356999099255,
|
|
"num_tokens": 652772.0,
|
|
"step": 305
|
|
},
|
|
{
|
|
"entropy": 8.341912460327148,
|
|
"epoch": 0.022000638728221143,
|
|
"grad_norm": 1.1328125,
|
|
"learning_rate": 0.00015450000000000001,
|
|
"loss": 8.1348,
|
|
"mean_token_accuracy": 0.07087188474833965,
|
|
"num_tokens": 664052.0,
|
|
"step": 310
|
|
},
|
|
{
|
|
"entropy": 8.363450813293458,
|
|
"epoch": 0.022355487739966644,
|
|
"grad_norm": 1.03125,
|
|
"learning_rate": 0.000157,
|
|
"loss": 8.1412,
|
|
"mean_token_accuracy": 0.07019479162991046,
|
|
"num_tokens": 675937.0,
|
|
"step": 315
|
|
},
|
|
{
|
|
"entropy": 8.314255905151366,
|
|
"epoch": 0.022710336751712145,
|
|
"grad_norm": 0.93359375,
|
|
"learning_rate": 0.0001595,
|
|
"loss": 8.065,
|
|
"mean_token_accuracy": 0.07636438310146332,
|
|
"num_tokens": 687149.0,
|
|
"step": 320
|
|
},
|
|
{
|
|
"entropy": 8.342943286895752,
|
|
"epoch": 0.02306518576345765,
|
|
"grad_norm": 1.0390625,
|
|
"learning_rate": 0.000162,
|
|
"loss": 8.0868,
|
|
"mean_token_accuracy": 0.06923852488398552,
|
|
"num_tokens": 697646.0,
|
|
"step": 325
|
|
},
|
|
{
|
|
"entropy": 8.419994449615478,
|
|
"epoch": 0.02342003477520315,
|
|
"grad_norm": 1.1484375,
|
|
"learning_rate": 0.00016450000000000001,
|
|
"loss": 8.2002,
|
|
"mean_token_accuracy": 0.06634289510548115,
|
|
"num_tokens": 708612.0,
|
|
"step": 330
|
|
},
|
|
{
|
|
"entropy": 8.354696655273438,
|
|
"epoch": 0.023774883786948652,
|
|
"grad_norm": 1.0546875,
|
|
"learning_rate": 0.00016700000000000002,
|
|
"loss": 8.1205,
|
|
"mean_token_accuracy": 0.07153847180306912,
|
|
"num_tokens": 719705.0,
|
|
"step": 335
|
|
},
|
|
{
|
|
"entropy": 8.35772638320923,
|
|
"epoch": 0.024129732798694157,
|
|
"grad_norm": 0.890625,
|
|
"learning_rate": 0.00016950000000000003,
|
|
"loss": 8.0862,
|
|
"mean_token_accuracy": 0.06683175601065158,
|
|
"num_tokens": 730552.0,
|
|
"step": 340
|
|
},
|
|
{
|
|
"entropy": 8.334082317352294,
|
|
"epoch": 0.024484581810439658,
|
|
"grad_norm": 1.140625,
|
|
"learning_rate": 0.00017199999999999998,
|
|
"loss": 8.0935,
|
|
"mean_token_accuracy": 0.06866642013192177,
|
|
"num_tokens": 741651.0,
|
|
"step": 345
|
|
},
|
|
{
|
|
"entropy": 8.407394886016846,
|
|
"epoch": 0.02483943082218516,
|
|
"grad_norm": 1.03125,
|
|
"learning_rate": 0.00017449999999999999,
|
|
"loss": 8.1023,
|
|
"mean_token_accuracy": 0.06832008063793182,
|
|
"num_tokens": 751820.0,
|
|
"step": 350
|
|
},
|
|
{
|
|
"entropy": 8.351481819152832,
|
|
"epoch": 0.025194279833930664,
|
|
"grad_norm": 1.21875,
|
|
"learning_rate": 0.000177,
|
|
"loss": 8.1154,
|
|
"mean_token_accuracy": 0.06219382137060166,
|
|
"num_tokens": 762125.0,
|
|
"step": 355
|
|
},
|
|
{
|
|
"entropy": 8.352156829833984,
|
|
"epoch": 0.025549128845676165,
|
|
"grad_norm": 0.92578125,
|
|
"learning_rate": 0.0001795,
|
|
"loss": 8.0332,
|
|
"mean_token_accuracy": 0.07175925448536873,
|
|
"num_tokens": 773137.0,
|
|
"step": 360
|
|
},
|
|
{
|
|
"entropy": 8.26557502746582,
|
|
"epoch": 0.025903977857421667,
|
|
"grad_norm": 1.015625,
|
|
"learning_rate": 0.000182,
|
|
"loss": 8.117,
|
|
"mean_token_accuracy": 0.06638101302087307,
|
|
"num_tokens": 784702.0,
|
|
"step": 365
|
|
},
|
|
{
|
|
"entropy": 8.30114049911499,
|
|
"epoch": 0.026258826869167168,
|
|
"grad_norm": 0.9296875,
|
|
"learning_rate": 0.0001845,
|
|
"loss": 8.0759,
|
|
"mean_token_accuracy": 0.06737305559217929,
|
|
"num_tokens": 795806.0,
|
|
"step": 370
|
|
},
|
|
{
|
|
"entropy": 8.324819755554199,
|
|
"epoch": 0.026613675880912673,
|
|
"grad_norm": 0.8515625,
|
|
"learning_rate": 0.000187,
|
|
"loss": 8.0175,
|
|
"mean_token_accuracy": 0.06811637170612812,
|
|
"num_tokens": 806006.0,
|
|
"step": 375
|
|
},
|
|
{
|
|
"entropy": 8.294619274139404,
|
|
"epoch": 0.026968524892658174,
|
|
"grad_norm": 0.98828125,
|
|
"learning_rate": 0.0001895,
|
|
"loss": 8.0306,
|
|
"mean_token_accuracy": 0.0675454068928957,
|
|
"num_tokens": 816345.0,
|
|
"step": 380
|
|
},
|
|
{
|
|
"entropy": 8.203885173797607,
|
|
"epoch": 0.027323373904403675,
|
|
"grad_norm": 0.92578125,
|
|
"learning_rate": 0.000192,
|
|
"loss": 8.0311,
|
|
"mean_token_accuracy": 0.06778408177196979,
|
|
"num_tokens": 827038.0,
|
|
"step": 385
|
|
},
|
|
{
|
|
"entropy": 8.261708545684815,
|
|
"epoch": 0.02767822291614918,
|
|
"grad_norm": 0.84765625,
|
|
"learning_rate": 0.0001945,
|
|
"loss": 8.0325,
|
|
"mean_token_accuracy": 0.07145289778709411,
|
|
"num_tokens": 837465.0,
|
|
"step": 390
|
|
},
|
|
{
|
|
"entropy": 8.322563171386719,
|
|
"epoch": 0.02803307192789468,
|
|
"grad_norm": 1.1015625,
|
|
"learning_rate": 0.00019700000000000002,
|
|
"loss": 8.0101,
|
|
"mean_token_accuracy": 0.0660801563411951,
|
|
"num_tokens": 848461.0,
|
|
"step": 395
|
|
},
|
|
{
|
|
"entropy": 8.286761474609374,
|
|
"epoch": 0.028387920939640182,
|
|
"grad_norm": 1.5390625,
|
|
"learning_rate": 0.00019950000000000002,
|
|
"loss": 8.0912,
|
|
"mean_token_accuracy": 0.06686392053961754,
|
|
"num_tokens": 859215.0,
|
|
"step": 400
|
|
},
|
|
{
|
|
"entropy": 8.153140926361084,
|
|
"epoch": 0.028742769951385687,
|
|
"grad_norm": 0.95703125,
|
|
"learning_rate": 0.000202,
|
|
"loss": 7.9714,
|
|
"mean_token_accuracy": 0.06791464723646641,
|
|
"num_tokens": 869165.0,
|
|
"step": 405
|
|
},
|
|
{
|
|
"entropy": 8.278606796264649,
|
|
"epoch": 0.029097618963131188,
|
|
"grad_norm": 1.1796875,
|
|
"learning_rate": 0.00020449999999999998,
|
|
"loss": 8.0299,
|
|
"mean_token_accuracy": 0.06314935013651848,
|
|
"num_tokens": 879737.0,
|
|
"step": 410
|
|
},
|
|
{
|
|
"entropy": 8.317267990112304,
|
|
"epoch": 0.02945246797487669,
|
|
"grad_norm": 0.97265625,
|
|
"learning_rate": 0.000207,
|
|
"loss": 8.0687,
|
|
"mean_token_accuracy": 0.06716468073427677,
|
|
"num_tokens": 889960.0,
|
|
"step": 415
|
|
},
|
|
{
|
|
"entropy": 8.284847354888916,
|
|
"epoch": 0.02980731698662219,
|
|
"grad_norm": 1.0390625,
|
|
"learning_rate": 0.0002095,
|
|
"loss": 8.0315,
|
|
"mean_token_accuracy": 0.06850776262581348,
|
|
"num_tokens": 902004.0,
|
|
"step": 420
|
|
},
|
|
{
|
|
"entropy": 8.195465469360352,
|
|
"epoch": 0.030162165998367695,
|
|
"grad_norm": 1.015625,
|
|
"learning_rate": 0.000212,
|
|
"loss": 7.9605,
|
|
"mean_token_accuracy": 0.06964701861143112,
|
|
"num_tokens": 912448.0,
|
|
"step": 425
|
|
},
|
|
{
|
|
"entropy": 8.246010208129883,
|
|
"epoch": 0.030517015010113197,
|
|
"grad_norm": 1.1015625,
|
|
"learning_rate": 0.0002145,
|
|
"loss": 8.0425,
|
|
"mean_token_accuracy": 0.06564139947295189,
|
|
"num_tokens": 922295.0,
|
|
"step": 430
|
|
},
|
|
{
|
|
"entropy": 8.212500381469727,
|
|
"epoch": 0.030871864021858698,
|
|
"grad_norm": 1.15625,
|
|
"learning_rate": 0.00021700000000000002,
|
|
"loss": 7.9675,
|
|
"mean_token_accuracy": 0.07113610915839672,
|
|
"num_tokens": 932849.0,
|
|
"step": 435
|
|
},
|
|
{
|
|
"entropy": 8.252998352050781,
|
|
"epoch": 0.031226713033604202,
|
|
"grad_norm": 1.1640625,
|
|
"learning_rate": 0.0002195,
|
|
"loss": 7.9716,
|
|
"mean_token_accuracy": 0.06684075817465782,
|
|
"num_tokens": 943570.0,
|
|
"step": 440
|
|
},
|
|
{
|
|
"entropy": 8.093912363052368,
|
|
"epoch": 0.031581562045349704,
|
|
"grad_norm": 1.0625,
|
|
"learning_rate": 0.000222,
|
|
"loss": 8.0064,
|
|
"mean_token_accuracy": 0.06446249783039093,
|
|
"num_tokens": 953681.0,
|
|
"step": 445
|
|
},
|
|
{
|
|
"entropy": 8.259466552734375,
|
|
"epoch": 0.03193641105709521,
|
|
"grad_norm": 1.0390625,
|
|
"learning_rate": 0.0002245,
|
|
"loss": 8.0026,
|
|
"mean_token_accuracy": 0.06529812254011631,
|
|
"num_tokens": 965027.0,
|
|
"step": 450
|
|
},
|
|
{
|
|
"entropy": 8.120300006866454,
|
|
"epoch": 0.032291260068840706,
|
|
"grad_norm": 1.0703125,
|
|
"learning_rate": 0.00022700000000000002,
|
|
"loss": 7.9335,
|
|
"mean_token_accuracy": 0.06843561045825482,
|
|
"num_tokens": 975757.0,
|
|
"step": 455
|
|
},
|
|
{
|
|
"entropy": 8.253607749938965,
|
|
"epoch": 0.03264610908058621,
|
|
"grad_norm": 0.8359375,
|
|
"learning_rate": 0.00022950000000000002,
|
|
"loss": 7.975,
|
|
"mean_token_accuracy": 0.07162793055176735,
|
|
"num_tokens": 987354.0,
|
|
"step": 460
|
|
},
|
|
{
|
|
"entropy": 8.038683032989502,
|
|
"epoch": 0.033000958092331716,
|
|
"grad_norm": 1.0625,
|
|
"learning_rate": 0.00023200000000000003,
|
|
"loss": 7.9604,
|
|
"mean_token_accuracy": 0.06958170719444752,
|
|
"num_tokens": 997844.0,
|
|
"step": 465
|
|
},
|
|
{
|
|
"entropy": 8.277831172943115,
|
|
"epoch": 0.03335580710407721,
|
|
"grad_norm": 1.1484375,
|
|
"learning_rate": 0.00023449999999999998,
|
|
"loss": 8.0444,
|
|
"mean_token_accuracy": 0.07072999998927117,
|
|
"num_tokens": 1008748.0,
|
|
"step": 470
|
|
},
|
|
{
|
|
"entropy": 8.073792266845704,
|
|
"epoch": 0.03371065611582272,
|
|
"grad_norm": 1.078125,
|
|
"learning_rate": 0.000237,
|
|
"loss": 7.8659,
|
|
"mean_token_accuracy": 0.0738751970231533,
|
|
"num_tokens": 1018612.0,
|
|
"step": 475
|
|
},
|
|
{
|
|
"entropy": 8.137950706481934,
|
|
"epoch": 0.03406550512756822,
|
|
"grad_norm": 1.2578125,
|
|
"learning_rate": 0.0002395,
|
|
"loss": 7.9242,
|
|
"mean_token_accuracy": 0.07282396517693997,
|
|
"num_tokens": 1029431.0,
|
|
"step": 480
|
|
},
|
|
{
|
|
"entropy": 8.098453617095947,
|
|
"epoch": 0.03442035413931372,
|
|
"grad_norm": 1.1328125,
|
|
"learning_rate": 0.000242,
|
|
"loss": 7.9572,
|
|
"mean_token_accuracy": 0.06277543865144253,
|
|
"num_tokens": 1041255.0,
|
|
"step": 485
|
|
},
|
|
{
|
|
"entropy": 8.044496870040893,
|
|
"epoch": 0.034775203151059225,
|
|
"grad_norm": 1.375,
|
|
"learning_rate": 0.0002445,
|
|
"loss": 7.7917,
|
|
"mean_token_accuracy": 0.08647977113723755,
|
|
"num_tokens": 1052416.0,
|
|
"step": 490
|
|
},
|
|
{
|
|
"entropy": 8.117148971557617,
|
|
"epoch": 0.03513005216280473,
|
|
"grad_norm": 0.98046875,
|
|
"learning_rate": 0.000247,
|
|
"loss": 7.9132,
|
|
"mean_token_accuracy": 0.07025011256337166,
|
|
"num_tokens": 1063801.0,
|
|
"step": 495
|
|
},
|
|
{
|
|
"entropy": 8.052232360839843,
|
|
"epoch": 0.03548490117455023,
|
|
"grad_norm": 1.1015625,
|
|
"learning_rate": 0.0002495,
|
|
"loss": 7.9351,
|
|
"mean_token_accuracy": 0.06856431551277638,
|
|
"num_tokens": 1074984.0,
|
|
"step": 500
|
|
}
|
|
],
|
|
"logging_steps": 5,
|
|
"max_steps": 4000,
|
|
"num_input_tokens_seen": 0,
|
|
"num_train_epochs": 1,
|
|
"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": 1584891316224000.0,
|
|
"train_batch_size": 16,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|