1035 lines
27 KiB
JSON
1035 lines
27 KiB
JSON
{
|
|
"best_global_step": null,
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 0.029265437518290898,
|
|
"eval_steps": 500,
|
|
"global_step": 500,
|
|
"is_hyper_param_search": false,
|
|
"is_local_process_zero": true,
|
|
"is_world_process_zero": true,
|
|
"log_history": [
|
|
{
|
|
"entropy": 10.691977596282959,
|
|
"epoch": 0.000292654375182909,
|
|
"grad_norm": 12.75,
|
|
"learning_rate": 2e-06,
|
|
"loss": 10.8274,
|
|
"mean_token_accuracy": 0.0,
|
|
"num_tokens": 10517.0,
|
|
"step": 5
|
|
},
|
|
{
|
|
"entropy": 10.691955089569092,
|
|
"epoch": 0.000585308750365818,
|
|
"grad_norm": 12.875,
|
|
"learning_rate": 4.5e-06,
|
|
"loss": 10.7547,
|
|
"mean_token_accuracy": 0.00010537407360970974,
|
|
"num_tokens": 20499.0,
|
|
"step": 10
|
|
},
|
|
{
|
|
"entropy": 10.690744304656983,
|
|
"epoch": 0.000877963125548727,
|
|
"grad_norm": 10.375,
|
|
"learning_rate": 7e-06,
|
|
"loss": 10.507,
|
|
"mean_token_accuracy": 0.03299602484330535,
|
|
"num_tokens": 30396.0,
|
|
"step": 15
|
|
},
|
|
{
|
|
"entropy": 10.663488483428955,
|
|
"epoch": 0.001170617500731636,
|
|
"grad_norm": 6.0625,
|
|
"learning_rate": 9.5e-06,
|
|
"loss": 10.196,
|
|
"mean_token_accuracy": 0.05248659364879131,
|
|
"num_tokens": 40862.0,
|
|
"step": 20
|
|
},
|
|
{
|
|
"entropy": 10.466671466827393,
|
|
"epoch": 0.0014632718759145448,
|
|
"grad_norm": 4.125,
|
|
"learning_rate": 1.2e-05,
|
|
"loss": 9.9118,
|
|
"mean_token_accuracy": 0.05587219782173634,
|
|
"num_tokens": 50482.0,
|
|
"step": 25
|
|
},
|
|
{
|
|
"entropy": 10.479176902770996,
|
|
"epoch": 0.001755926251097454,
|
|
"grad_norm": 3.28125,
|
|
"learning_rate": 1.4500000000000002e-05,
|
|
"loss": 9.7666,
|
|
"mean_token_accuracy": 0.05055557526648045,
|
|
"num_tokens": 59557.0,
|
|
"step": 30
|
|
},
|
|
{
|
|
"entropy": 10.499413967132568,
|
|
"epoch": 0.002048580626280363,
|
|
"grad_norm": 2.84375,
|
|
"learning_rate": 1.7000000000000003e-05,
|
|
"loss": 9.7212,
|
|
"mean_token_accuracy": 0.05302470289170742,
|
|
"num_tokens": 69864.0,
|
|
"step": 35
|
|
},
|
|
{
|
|
"entropy": 10.425041770935058,
|
|
"epoch": 0.002341235001463272,
|
|
"grad_norm": 2.34375,
|
|
"learning_rate": 1.95e-05,
|
|
"loss": 9.6782,
|
|
"mean_token_accuracy": 0.05624178126454353,
|
|
"num_tokens": 80632.0,
|
|
"step": 40
|
|
},
|
|
{
|
|
"entropy": 10.474136352539062,
|
|
"epoch": 0.0026338893766461808,
|
|
"grad_norm": 2.453125,
|
|
"learning_rate": 2.2e-05,
|
|
"loss": 9.6706,
|
|
"mean_token_accuracy": 0.04874711595475674,
|
|
"num_tokens": 90889.0,
|
|
"step": 45
|
|
},
|
|
{
|
|
"entropy": 10.464450550079345,
|
|
"epoch": 0.0029265437518290896,
|
|
"grad_norm": 2.328125,
|
|
"learning_rate": 2.4500000000000003e-05,
|
|
"loss": 9.5772,
|
|
"mean_token_accuracy": 0.05688605234026909,
|
|
"num_tokens": 101136.0,
|
|
"step": 50
|
|
},
|
|
{
|
|
"entropy": 10.375558757781983,
|
|
"epoch": 0.003219198127011999,
|
|
"grad_norm": 2.203125,
|
|
"learning_rate": 2.7e-05,
|
|
"loss": 9.5567,
|
|
"mean_token_accuracy": 0.06301373578608035,
|
|
"num_tokens": 111743.0,
|
|
"step": 55
|
|
},
|
|
{
|
|
"entropy": 10.292755794525146,
|
|
"epoch": 0.003511852502194908,
|
|
"grad_norm": 2.09375,
|
|
"learning_rate": 2.95e-05,
|
|
"loss": 9.5051,
|
|
"mean_token_accuracy": 0.06023350767791271,
|
|
"num_tokens": 121968.0,
|
|
"step": 60
|
|
},
|
|
{
|
|
"entropy": 10.296923160552979,
|
|
"epoch": 0.0038045068773778167,
|
|
"grad_norm": 1.9296875,
|
|
"learning_rate": 3.2e-05,
|
|
"loss": 9.4037,
|
|
"mean_token_accuracy": 0.06824495643377304,
|
|
"num_tokens": 131917.0,
|
|
"step": 65
|
|
},
|
|
{
|
|
"entropy": 10.250029182434082,
|
|
"epoch": 0.004097161252560726,
|
|
"grad_norm": 2.0625,
|
|
"learning_rate": 3.4500000000000005e-05,
|
|
"loss": 9.3308,
|
|
"mean_token_accuracy": 0.06861314773559571,
|
|
"num_tokens": 141562.0,
|
|
"step": 70
|
|
},
|
|
{
|
|
"entropy": 10.270350456237793,
|
|
"epoch": 0.004389815627743635,
|
|
"grad_norm": 1.984375,
|
|
"learning_rate": 3.7e-05,
|
|
"loss": 9.323,
|
|
"mean_token_accuracy": 0.06920466646552086,
|
|
"num_tokens": 152163.0,
|
|
"step": 75
|
|
},
|
|
{
|
|
"entropy": 10.280463314056396,
|
|
"epoch": 0.004682470002926544,
|
|
"grad_norm": 2.375,
|
|
"learning_rate": 3.95e-05,
|
|
"loss": 9.301,
|
|
"mean_token_accuracy": 0.06251571290194988,
|
|
"num_tokens": 162684.0,
|
|
"step": 80
|
|
},
|
|
{
|
|
"entropy": 10.195634841918945,
|
|
"epoch": 0.004975124378109453,
|
|
"grad_norm": 1.9921875,
|
|
"learning_rate": 4.2000000000000004e-05,
|
|
"loss": 9.1337,
|
|
"mean_token_accuracy": 0.07077112905681134,
|
|
"num_tokens": 173234.0,
|
|
"step": 85
|
|
},
|
|
{
|
|
"entropy": 10.168700313568115,
|
|
"epoch": 0.0052677787532923615,
|
|
"grad_norm": 1.765625,
|
|
"learning_rate": 4.45e-05,
|
|
"loss": 9.1283,
|
|
"mean_token_accuracy": 0.06726612299680709,
|
|
"num_tokens": 183645.0,
|
|
"step": 90
|
|
},
|
|
{
|
|
"entropy": 10.11437921524048,
|
|
"epoch": 0.00556043312847527,
|
|
"grad_norm": 1.6875,
|
|
"learning_rate": 4.7000000000000004e-05,
|
|
"loss": 9.0472,
|
|
"mean_token_accuracy": 0.06981992349028587,
|
|
"num_tokens": 193874.0,
|
|
"step": 95
|
|
},
|
|
{
|
|
"entropy": 10.146056175231934,
|
|
"epoch": 0.005853087503658179,
|
|
"grad_norm": 1.6640625,
|
|
"learning_rate": 4.9500000000000004e-05,
|
|
"loss": 8.9144,
|
|
"mean_token_accuracy": 0.0724099788814783,
|
|
"num_tokens": 203965.0,
|
|
"step": 100
|
|
},
|
|
{
|
|
"entropy": 10.023962688446044,
|
|
"epoch": 0.006145741878841089,
|
|
"grad_norm": 1.3984375,
|
|
"learning_rate": 5.2e-05,
|
|
"loss": 8.8606,
|
|
"mean_token_accuracy": 0.07363765798509121,
|
|
"num_tokens": 214334.0,
|
|
"step": 105
|
|
},
|
|
{
|
|
"entropy": 9.978809070587157,
|
|
"epoch": 0.006438396254023998,
|
|
"grad_norm": 1.390625,
|
|
"learning_rate": 5.45e-05,
|
|
"loss": 8.7784,
|
|
"mean_token_accuracy": 0.07168939225375652,
|
|
"num_tokens": 225435.0,
|
|
"step": 110
|
|
},
|
|
{
|
|
"entropy": 9.947597980499268,
|
|
"epoch": 0.006731050629206907,
|
|
"grad_norm": 1.5703125,
|
|
"learning_rate": 5.7e-05,
|
|
"loss": 8.6424,
|
|
"mean_token_accuracy": 0.06688602603971958,
|
|
"num_tokens": 235107.0,
|
|
"step": 115
|
|
},
|
|
{
|
|
"entropy": 9.726509189605713,
|
|
"epoch": 0.007023705004389816,
|
|
"grad_norm": 1.3828125,
|
|
"learning_rate": 5.9499999999999996e-05,
|
|
"loss": 8.5683,
|
|
"mean_token_accuracy": 0.07460153996944427,
|
|
"num_tokens": 245660.0,
|
|
"step": 120
|
|
},
|
|
{
|
|
"entropy": 9.732983684539795,
|
|
"epoch": 0.0073163593795727245,
|
|
"grad_norm": 1.3828125,
|
|
"learning_rate": 6.2e-05,
|
|
"loss": 8.4869,
|
|
"mean_token_accuracy": 0.07104945406317711,
|
|
"num_tokens": 256077.0,
|
|
"step": 125
|
|
},
|
|
{
|
|
"entropy": 9.572216033935547,
|
|
"epoch": 0.007609013754755633,
|
|
"grad_norm": 1.171875,
|
|
"learning_rate": 6.450000000000001e-05,
|
|
"loss": 8.4352,
|
|
"mean_token_accuracy": 0.06920053586363792,
|
|
"num_tokens": 265702.0,
|
|
"step": 130
|
|
},
|
|
{
|
|
"entropy": 9.483925437927246,
|
|
"epoch": 0.007901668129938543,
|
|
"grad_norm": 1.34375,
|
|
"learning_rate": 6.7e-05,
|
|
"loss": 8.3711,
|
|
"mean_token_accuracy": 0.0720779214054346,
|
|
"num_tokens": 276739.0,
|
|
"step": 135
|
|
},
|
|
{
|
|
"entropy": 9.266712093353272,
|
|
"epoch": 0.008194322505121452,
|
|
"grad_norm": 1.171875,
|
|
"learning_rate": 6.950000000000001e-05,
|
|
"loss": 8.2541,
|
|
"mean_token_accuracy": 0.07680930271744728,
|
|
"num_tokens": 285370.0,
|
|
"step": 140
|
|
},
|
|
{
|
|
"entropy": 9.12420482635498,
|
|
"epoch": 0.008486976880304361,
|
|
"grad_norm": 0.9765625,
|
|
"learning_rate": 7.2e-05,
|
|
"loss": 8.2031,
|
|
"mean_token_accuracy": 0.07261989042162895,
|
|
"num_tokens": 295080.0,
|
|
"step": 145
|
|
},
|
|
{
|
|
"entropy": 8.998386192321778,
|
|
"epoch": 0.00877963125548727,
|
|
"grad_norm": 1.015625,
|
|
"learning_rate": 7.45e-05,
|
|
"loss": 8.1432,
|
|
"mean_token_accuracy": 0.07861845716834068,
|
|
"num_tokens": 304509.0,
|
|
"step": 150
|
|
},
|
|
{
|
|
"entropy": 8.892388534545898,
|
|
"epoch": 0.009072285630670179,
|
|
"grad_norm": 1.078125,
|
|
"learning_rate": 7.7e-05,
|
|
"loss": 8.1203,
|
|
"mean_token_accuracy": 0.07556253299117088,
|
|
"num_tokens": 315214.0,
|
|
"step": 155
|
|
},
|
|
{
|
|
"entropy": 8.732898807525634,
|
|
"epoch": 0.009364940005853088,
|
|
"grad_norm": 1.375,
|
|
"learning_rate": 7.950000000000001e-05,
|
|
"loss": 8.1005,
|
|
"mean_token_accuracy": 0.07594769522547722,
|
|
"num_tokens": 325865.0,
|
|
"step": 160
|
|
},
|
|
{
|
|
"entropy": 8.679925060272216,
|
|
"epoch": 0.009657594381035996,
|
|
"grad_norm": 0.94921875,
|
|
"learning_rate": 8.2e-05,
|
|
"loss": 8.0335,
|
|
"mean_token_accuracy": 0.07580512613058091,
|
|
"num_tokens": 335490.0,
|
|
"step": 165
|
|
},
|
|
{
|
|
"entropy": 8.627653217315673,
|
|
"epoch": 0.009950248756218905,
|
|
"grad_norm": 1.1640625,
|
|
"learning_rate": 8.450000000000001e-05,
|
|
"loss": 8.1191,
|
|
"mean_token_accuracy": 0.06616722382605075,
|
|
"num_tokens": 345507.0,
|
|
"step": 170
|
|
},
|
|
{
|
|
"entropy": 8.511928844451905,
|
|
"epoch": 0.010242903131401814,
|
|
"grad_norm": 1.171875,
|
|
"learning_rate": 8.7e-05,
|
|
"loss": 8.0372,
|
|
"mean_token_accuracy": 0.07108715996146202,
|
|
"num_tokens": 355129.0,
|
|
"step": 175
|
|
},
|
|
{
|
|
"entropy": 8.449845886230468,
|
|
"epoch": 0.010535557506584723,
|
|
"grad_norm": 1.0078125,
|
|
"learning_rate": 8.95e-05,
|
|
"loss": 7.9472,
|
|
"mean_token_accuracy": 0.07913482263684272,
|
|
"num_tokens": 364700.0,
|
|
"step": 180
|
|
},
|
|
{
|
|
"entropy": 8.450096702575683,
|
|
"epoch": 0.010828211881767632,
|
|
"grad_norm": 1.1953125,
|
|
"learning_rate": 9.2e-05,
|
|
"loss": 8.0521,
|
|
"mean_token_accuracy": 0.07427797615528106,
|
|
"num_tokens": 374920.0,
|
|
"step": 185
|
|
},
|
|
{
|
|
"entropy": 8.375695610046387,
|
|
"epoch": 0.01112086625695054,
|
|
"grad_norm": 1.046875,
|
|
"learning_rate": 9.45e-05,
|
|
"loss": 7.8379,
|
|
"mean_token_accuracy": 0.07953086122870445,
|
|
"num_tokens": 384517.0,
|
|
"step": 190
|
|
},
|
|
{
|
|
"entropy": 8.32225456237793,
|
|
"epoch": 0.01141352063213345,
|
|
"grad_norm": 1.0546875,
|
|
"learning_rate": 9.7e-05,
|
|
"loss": 8.0024,
|
|
"mean_token_accuracy": 0.0835827887058258,
|
|
"num_tokens": 394118.0,
|
|
"step": 195
|
|
},
|
|
{
|
|
"entropy": 8.35567274093628,
|
|
"epoch": 0.011706175007316359,
|
|
"grad_norm": 1.234375,
|
|
"learning_rate": 9.95e-05,
|
|
"loss": 7.9942,
|
|
"mean_token_accuracy": 0.07656502649188042,
|
|
"num_tokens": 404135.0,
|
|
"step": 200
|
|
},
|
|
{
|
|
"entropy": 8.368758583068848,
|
|
"epoch": 0.01199882938249927,
|
|
"grad_norm": 1.03125,
|
|
"learning_rate": 0.000102,
|
|
"loss": 7.9689,
|
|
"mean_token_accuracy": 0.0749567061662674,
|
|
"num_tokens": 414198.0,
|
|
"step": 205
|
|
},
|
|
{
|
|
"entropy": 8.292402458190917,
|
|
"epoch": 0.012291483757682178,
|
|
"grad_norm": 0.953125,
|
|
"learning_rate": 0.00010449999999999999,
|
|
"loss": 7.9303,
|
|
"mean_token_accuracy": 0.0756587840616703,
|
|
"num_tokens": 424747.0,
|
|
"step": 210
|
|
},
|
|
{
|
|
"entropy": 8.321221828460693,
|
|
"epoch": 0.012584138132865087,
|
|
"grad_norm": 0.92578125,
|
|
"learning_rate": 0.000107,
|
|
"loss": 7.9038,
|
|
"mean_token_accuracy": 0.07831631563603877,
|
|
"num_tokens": 435597.0,
|
|
"step": 215
|
|
},
|
|
{
|
|
"entropy": 8.24533395767212,
|
|
"epoch": 0.012876792508047996,
|
|
"grad_norm": 1.0,
|
|
"learning_rate": 0.0001095,
|
|
"loss": 7.9501,
|
|
"mean_token_accuracy": 0.07723408415913582,
|
|
"num_tokens": 445262.0,
|
|
"step": 220
|
|
},
|
|
{
|
|
"entropy": 8.260603141784667,
|
|
"epoch": 0.013169446883230905,
|
|
"grad_norm": 1.296875,
|
|
"learning_rate": 0.000112,
|
|
"loss": 8.0105,
|
|
"mean_token_accuracy": 0.06991274990141391,
|
|
"num_tokens": 456455.0,
|
|
"step": 225
|
|
},
|
|
{
|
|
"entropy": 8.28931884765625,
|
|
"epoch": 0.013462101258413814,
|
|
"grad_norm": 1.1953125,
|
|
"learning_rate": 0.0001145,
|
|
"loss": 7.8175,
|
|
"mean_token_accuracy": 0.07421639449894428,
|
|
"num_tokens": 465696.0,
|
|
"step": 230
|
|
},
|
|
{
|
|
"entropy": 8.135773086547852,
|
|
"epoch": 0.013754755633596722,
|
|
"grad_norm": 1.1875,
|
|
"learning_rate": 0.00011700000000000001,
|
|
"loss": 7.89,
|
|
"mean_token_accuracy": 0.079552548751235,
|
|
"num_tokens": 475171.0,
|
|
"step": 235
|
|
},
|
|
{
|
|
"entropy": 8.152732849121094,
|
|
"epoch": 0.014047410008779631,
|
|
"grad_norm": 1.078125,
|
|
"learning_rate": 0.00011949999999999999,
|
|
"loss": 7.8955,
|
|
"mean_token_accuracy": 0.07634073123335838,
|
|
"num_tokens": 484852.0,
|
|
"step": 240
|
|
},
|
|
{
|
|
"entropy": 8.248970890045166,
|
|
"epoch": 0.01434006438396254,
|
|
"grad_norm": 1.125,
|
|
"learning_rate": 0.000122,
|
|
"loss": 7.8986,
|
|
"mean_token_accuracy": 0.07723180837929249,
|
|
"num_tokens": 494539.0,
|
|
"step": 245
|
|
},
|
|
{
|
|
"entropy": 8.244297409057618,
|
|
"epoch": 0.014632718759145449,
|
|
"grad_norm": 0.94140625,
|
|
"learning_rate": 0.0001245,
|
|
"loss": 7.8722,
|
|
"mean_token_accuracy": 0.07467923238873482,
|
|
"num_tokens": 504406.0,
|
|
"step": 250
|
|
},
|
|
{
|
|
"entropy": 8.212084579467774,
|
|
"epoch": 0.014925373134328358,
|
|
"grad_norm": 1.0859375,
|
|
"learning_rate": 0.000127,
|
|
"loss": 7.8875,
|
|
"mean_token_accuracy": 0.07326231747865677,
|
|
"num_tokens": 514905.0,
|
|
"step": 255
|
|
},
|
|
{
|
|
"entropy": 8.192084980010986,
|
|
"epoch": 0.015218027509511267,
|
|
"grad_norm": 1.1171875,
|
|
"learning_rate": 0.0001295,
|
|
"loss": 7.841,
|
|
"mean_token_accuracy": 0.07844860479235649,
|
|
"num_tokens": 524904.0,
|
|
"step": 260
|
|
},
|
|
{
|
|
"entropy": 8.152733612060548,
|
|
"epoch": 0.015510681884694176,
|
|
"grad_norm": 1.8125,
|
|
"learning_rate": 0.000132,
|
|
"loss": 7.8749,
|
|
"mean_token_accuracy": 0.07390310689806938,
|
|
"num_tokens": 534423.0,
|
|
"step": 265
|
|
},
|
|
{
|
|
"entropy": 8.228134441375733,
|
|
"epoch": 0.015803336259877086,
|
|
"grad_norm": 1.09375,
|
|
"learning_rate": 0.00013450000000000002,
|
|
"loss": 7.9388,
|
|
"mean_token_accuracy": 0.06983838640153409,
|
|
"num_tokens": 544468.0,
|
|
"step": 270
|
|
},
|
|
{
|
|
"entropy": 8.17867660522461,
|
|
"epoch": 0.016095990635059995,
|
|
"grad_norm": 0.984375,
|
|
"learning_rate": 0.00013700000000000002,
|
|
"loss": 7.8759,
|
|
"mean_token_accuracy": 0.07778191342949867,
|
|
"num_tokens": 555873.0,
|
|
"step": 275
|
|
},
|
|
{
|
|
"entropy": 8.1208966255188,
|
|
"epoch": 0.016388645010242904,
|
|
"grad_norm": 0.88671875,
|
|
"learning_rate": 0.0001395,
|
|
"loss": 7.8485,
|
|
"mean_token_accuracy": 0.07848807573318481,
|
|
"num_tokens": 566228.0,
|
|
"step": 280
|
|
},
|
|
{
|
|
"entropy": 8.147705793380737,
|
|
"epoch": 0.016681299385425813,
|
|
"grad_norm": 1.0859375,
|
|
"learning_rate": 0.00014199999999999998,
|
|
"loss": 7.7519,
|
|
"mean_token_accuracy": 0.07823162600398063,
|
|
"num_tokens": 576087.0,
|
|
"step": 285
|
|
},
|
|
{
|
|
"entropy": 8.103727769851684,
|
|
"epoch": 0.016973953760608722,
|
|
"grad_norm": 1.8359375,
|
|
"learning_rate": 0.0001445,
|
|
"loss": 7.8782,
|
|
"mean_token_accuracy": 0.07339520268142223,
|
|
"num_tokens": 586768.0,
|
|
"step": 290
|
|
},
|
|
{
|
|
"entropy": 8.19735746383667,
|
|
"epoch": 0.01726660813579163,
|
|
"grad_norm": 1.03125,
|
|
"learning_rate": 0.000147,
|
|
"loss": 7.871,
|
|
"mean_token_accuracy": 0.07267906814813614,
|
|
"num_tokens": 597352.0,
|
|
"step": 295
|
|
},
|
|
{
|
|
"entropy": 8.021362733840942,
|
|
"epoch": 0.01755926251097454,
|
|
"grad_norm": 1.171875,
|
|
"learning_rate": 0.0001495,
|
|
"loss": 7.7407,
|
|
"mean_token_accuracy": 0.07754108160734177,
|
|
"num_tokens": 607460.0,
|
|
"step": 300
|
|
},
|
|
{
|
|
"entropy": 8.081702327728271,
|
|
"epoch": 0.01785191688615745,
|
|
"grad_norm": 1.0390625,
|
|
"learning_rate": 0.000152,
|
|
"loss": 7.7764,
|
|
"mean_token_accuracy": 0.07714812532067299,
|
|
"num_tokens": 618882.0,
|
|
"step": 305
|
|
},
|
|
{
|
|
"entropy": 8.187482166290284,
|
|
"epoch": 0.018144571261340357,
|
|
"grad_norm": 1.0859375,
|
|
"learning_rate": 0.00015450000000000001,
|
|
"loss": 7.8853,
|
|
"mean_token_accuracy": 0.07371802367269993,
|
|
"num_tokens": 628378.0,
|
|
"step": 310
|
|
},
|
|
{
|
|
"entropy": 8.044809103012085,
|
|
"epoch": 0.018437225636523266,
|
|
"grad_norm": 1.1171875,
|
|
"learning_rate": 0.000157,
|
|
"loss": 7.8403,
|
|
"mean_token_accuracy": 0.07703564129769802,
|
|
"num_tokens": 637339.0,
|
|
"step": 315
|
|
},
|
|
{
|
|
"entropy": 8.112459707260133,
|
|
"epoch": 0.018729880011706175,
|
|
"grad_norm": 0.89453125,
|
|
"learning_rate": 0.0001595,
|
|
"loss": 7.8195,
|
|
"mean_token_accuracy": 0.0773365393280983,
|
|
"num_tokens": 648190.0,
|
|
"step": 320
|
|
},
|
|
{
|
|
"entropy": 8.131685829162597,
|
|
"epoch": 0.019022534386889084,
|
|
"grad_norm": 0.95703125,
|
|
"learning_rate": 0.000162,
|
|
"loss": 7.7489,
|
|
"mean_token_accuracy": 0.07892031595110893,
|
|
"num_tokens": 658704.0,
|
|
"step": 325
|
|
},
|
|
{
|
|
"entropy": 8.087395620346069,
|
|
"epoch": 0.019315188762071993,
|
|
"grad_norm": 1.546875,
|
|
"learning_rate": 0.00016450000000000001,
|
|
"loss": 7.7497,
|
|
"mean_token_accuracy": 0.07515155747532845,
|
|
"num_tokens": 668888.0,
|
|
"step": 330
|
|
},
|
|
{
|
|
"entropy": 8.038054466247559,
|
|
"epoch": 0.0196078431372549,
|
|
"grad_norm": 0.92578125,
|
|
"learning_rate": 0.00016700000000000002,
|
|
"loss": 7.7818,
|
|
"mean_token_accuracy": 0.07353989705443383,
|
|
"num_tokens": 679294.0,
|
|
"step": 335
|
|
},
|
|
{
|
|
"entropy": 8.008176279067992,
|
|
"epoch": 0.01990049751243781,
|
|
"grad_norm": 0.9375,
|
|
"learning_rate": 0.00016950000000000003,
|
|
"loss": 7.7553,
|
|
"mean_token_accuracy": 0.07883304283022881,
|
|
"num_tokens": 689662.0,
|
|
"step": 340
|
|
},
|
|
{
|
|
"entropy": 8.033820152282715,
|
|
"epoch": 0.02019315188762072,
|
|
"grad_norm": 1.0234375,
|
|
"learning_rate": 0.00017199999999999998,
|
|
"loss": 7.7645,
|
|
"mean_token_accuracy": 0.07818084880709648,
|
|
"num_tokens": 700356.0,
|
|
"step": 345
|
|
},
|
|
{
|
|
"entropy": 7.944976902008056,
|
|
"epoch": 0.02048580626280363,
|
|
"grad_norm": 1.1171875,
|
|
"learning_rate": 0.00017449999999999999,
|
|
"loss": 7.6998,
|
|
"mean_token_accuracy": 0.07956961244344711,
|
|
"num_tokens": 710831.0,
|
|
"step": 350
|
|
},
|
|
{
|
|
"entropy": 8.022411108016968,
|
|
"epoch": 0.020778460637986537,
|
|
"grad_norm": 1.234375,
|
|
"learning_rate": 0.000177,
|
|
"loss": 7.6943,
|
|
"mean_token_accuracy": 0.07844458818435669,
|
|
"num_tokens": 720445.0,
|
|
"step": 355
|
|
},
|
|
{
|
|
"entropy": 7.978969573974609,
|
|
"epoch": 0.021071115013169446,
|
|
"grad_norm": 1.0625,
|
|
"learning_rate": 0.0001795,
|
|
"loss": 7.6882,
|
|
"mean_token_accuracy": 0.08231284841895103,
|
|
"num_tokens": 730355.0,
|
|
"step": 360
|
|
},
|
|
{
|
|
"entropy": 7.976030063629151,
|
|
"epoch": 0.021363769388352355,
|
|
"grad_norm": 1.0078125,
|
|
"learning_rate": 0.000182,
|
|
"loss": 7.714,
|
|
"mean_token_accuracy": 0.08171503022313117,
|
|
"num_tokens": 740397.0,
|
|
"step": 365
|
|
},
|
|
{
|
|
"entropy": 7.986613845825195,
|
|
"epoch": 0.021656423763535264,
|
|
"grad_norm": 1.109375,
|
|
"learning_rate": 0.0001845,
|
|
"loss": 7.718,
|
|
"mean_token_accuracy": 0.07903816923499107,
|
|
"num_tokens": 750277.0,
|
|
"step": 370
|
|
},
|
|
{
|
|
"entropy": 7.9489898681640625,
|
|
"epoch": 0.021949078138718173,
|
|
"grad_norm": 0.9375,
|
|
"learning_rate": 0.000187,
|
|
"loss": 7.6953,
|
|
"mean_token_accuracy": 0.07957587093114853,
|
|
"num_tokens": 759764.0,
|
|
"step": 375
|
|
},
|
|
{
|
|
"entropy": 7.996847534179688,
|
|
"epoch": 0.02224173251390108,
|
|
"grad_norm": 0.91796875,
|
|
"learning_rate": 0.0001895,
|
|
"loss": 7.7437,
|
|
"mean_token_accuracy": 0.07961214706301689,
|
|
"num_tokens": 769709.0,
|
|
"step": 380
|
|
},
|
|
{
|
|
"entropy": 7.960434198379517,
|
|
"epoch": 0.02253438688908399,
|
|
"grad_norm": 1.0390625,
|
|
"learning_rate": 0.000192,
|
|
"loss": 7.6888,
|
|
"mean_token_accuracy": 0.08052227124571801,
|
|
"num_tokens": 780267.0,
|
|
"step": 385
|
|
},
|
|
{
|
|
"entropy": 7.887183332443238,
|
|
"epoch": 0.0228270412642669,
|
|
"grad_norm": 1.078125,
|
|
"learning_rate": 0.0001945,
|
|
"loss": 7.6611,
|
|
"mean_token_accuracy": 0.0823482058942318,
|
|
"num_tokens": 790184.0,
|
|
"step": 390
|
|
},
|
|
{
|
|
"entropy": 8.074115419387818,
|
|
"epoch": 0.023119695639449808,
|
|
"grad_norm": 1.0078125,
|
|
"learning_rate": 0.00019700000000000002,
|
|
"loss": 7.7323,
|
|
"mean_token_accuracy": 0.07797951698303222,
|
|
"num_tokens": 800645.0,
|
|
"step": 395
|
|
},
|
|
{
|
|
"entropy": 7.919103336334229,
|
|
"epoch": 0.023412350014632717,
|
|
"grad_norm": 0.9140625,
|
|
"learning_rate": 0.00019950000000000002,
|
|
"loss": 7.6645,
|
|
"mean_token_accuracy": 0.08313405811786652,
|
|
"num_tokens": 810873.0,
|
|
"step": 400
|
|
},
|
|
{
|
|
"entropy": 7.905709600448608,
|
|
"epoch": 0.02370500438981563,
|
|
"grad_norm": 1.015625,
|
|
"learning_rate": 0.000202,
|
|
"loss": 7.6646,
|
|
"mean_token_accuracy": 0.08137974441051483,
|
|
"num_tokens": 821661.0,
|
|
"step": 405
|
|
},
|
|
{
|
|
"entropy": 7.993849468231201,
|
|
"epoch": 0.02399765876499854,
|
|
"grad_norm": 1.0625,
|
|
"learning_rate": 0.00020449999999999998,
|
|
"loss": 7.6417,
|
|
"mean_token_accuracy": 0.08281447514891624,
|
|
"num_tokens": 831818.0,
|
|
"step": 410
|
|
},
|
|
{
|
|
"entropy": 7.86600136756897,
|
|
"epoch": 0.024290313140181447,
|
|
"grad_norm": 1.078125,
|
|
"learning_rate": 0.000207,
|
|
"loss": 7.6784,
|
|
"mean_token_accuracy": 0.08450354561209679,
|
|
"num_tokens": 843030.0,
|
|
"step": 415
|
|
},
|
|
{
|
|
"entropy": 7.940770626068115,
|
|
"epoch": 0.024582967515364356,
|
|
"grad_norm": 1.046875,
|
|
"learning_rate": 0.0002095,
|
|
"loss": 7.6337,
|
|
"mean_token_accuracy": 0.09171406775712967,
|
|
"num_tokens": 852477.0,
|
|
"step": 420
|
|
},
|
|
{
|
|
"entropy": 7.925455904006958,
|
|
"epoch": 0.024875621890547265,
|
|
"grad_norm": 1.234375,
|
|
"learning_rate": 0.000212,
|
|
"loss": 7.6457,
|
|
"mean_token_accuracy": 0.08530597761273384,
|
|
"num_tokens": 863018.0,
|
|
"step": 425
|
|
},
|
|
{
|
|
"entropy": 7.887920331954956,
|
|
"epoch": 0.025168276265730174,
|
|
"grad_norm": 1.203125,
|
|
"learning_rate": 0.0002145,
|
|
"loss": 7.5837,
|
|
"mean_token_accuracy": 0.07957525178790092,
|
|
"num_tokens": 872732.0,
|
|
"step": 430
|
|
},
|
|
{
|
|
"entropy": 7.84278564453125,
|
|
"epoch": 0.025460930640913083,
|
|
"grad_norm": 1.171875,
|
|
"learning_rate": 0.00021700000000000002,
|
|
"loss": 7.6344,
|
|
"mean_token_accuracy": 0.0791689209640026,
|
|
"num_tokens": 883189.0,
|
|
"step": 435
|
|
},
|
|
{
|
|
"entropy": 7.75295729637146,
|
|
"epoch": 0.02575358501609599,
|
|
"grad_norm": 1.1015625,
|
|
"learning_rate": 0.0002195,
|
|
"loss": 7.6724,
|
|
"mean_token_accuracy": 0.08212153613567352,
|
|
"num_tokens": 893874.0,
|
|
"step": 440
|
|
},
|
|
{
|
|
"entropy": 7.893164730072021,
|
|
"epoch": 0.0260462393912789,
|
|
"grad_norm": 1.1875,
|
|
"learning_rate": 0.000222,
|
|
"loss": 7.5997,
|
|
"mean_token_accuracy": 0.08359748423099518,
|
|
"num_tokens": 904424.0,
|
|
"step": 445
|
|
},
|
|
{
|
|
"entropy": 7.8800774097442625,
|
|
"epoch": 0.02633889376646181,
|
|
"grad_norm": 1.28125,
|
|
"learning_rate": 0.0002245,
|
|
"loss": 7.6555,
|
|
"mean_token_accuracy": 0.07997112423181534,
|
|
"num_tokens": 914601.0,
|
|
"step": 450
|
|
},
|
|
{
|
|
"entropy": 7.880753421783448,
|
|
"epoch": 0.026631548141644718,
|
|
"grad_norm": 1.1015625,
|
|
"learning_rate": 0.00022700000000000002,
|
|
"loss": 7.6627,
|
|
"mean_token_accuracy": 0.08024522922933101,
|
|
"num_tokens": 925162.0,
|
|
"step": 455
|
|
},
|
|
{
|
|
"entropy": 7.87500581741333,
|
|
"epoch": 0.026924202516827627,
|
|
"grad_norm": 1.2734375,
|
|
"learning_rate": 0.00022950000000000002,
|
|
"loss": 7.5647,
|
|
"mean_token_accuracy": 0.08628017753362656,
|
|
"num_tokens": 935182.0,
|
|
"step": 460
|
|
},
|
|
{
|
|
"entropy": 7.85167350769043,
|
|
"epoch": 0.027216856892010536,
|
|
"grad_norm": 1.0546875,
|
|
"learning_rate": 0.00023200000000000003,
|
|
"loss": 7.6007,
|
|
"mean_token_accuracy": 0.07952202931046486,
|
|
"num_tokens": 944812.0,
|
|
"step": 465
|
|
},
|
|
{
|
|
"entropy": 7.8027184963226315,
|
|
"epoch": 0.027509511267193445,
|
|
"grad_norm": 1.203125,
|
|
"learning_rate": 0.00023449999999999998,
|
|
"loss": 7.6031,
|
|
"mean_token_accuracy": 0.0839798703789711,
|
|
"num_tokens": 954532.0,
|
|
"step": 470
|
|
},
|
|
{
|
|
"entropy": 7.882350301742553,
|
|
"epoch": 0.027802165642376354,
|
|
"grad_norm": 1.125,
|
|
"learning_rate": 0.000237,
|
|
"loss": 7.557,
|
|
"mean_token_accuracy": 0.08790333345532417,
|
|
"num_tokens": 963611.0,
|
|
"step": 475
|
|
},
|
|
{
|
|
"entropy": 7.754200458526611,
|
|
"epoch": 0.028094820017559263,
|
|
"grad_norm": 1.078125,
|
|
"learning_rate": 0.0002395,
|
|
"loss": 7.6005,
|
|
"mean_token_accuracy": 0.08356534838676452,
|
|
"num_tokens": 973177.0,
|
|
"step": 480
|
|
},
|
|
{
|
|
"entropy": 7.877309036254883,
|
|
"epoch": 0.02838747439274217,
|
|
"grad_norm": 1.0234375,
|
|
"learning_rate": 0.000242,
|
|
"loss": 7.5999,
|
|
"mean_token_accuracy": 0.08032175153493881,
|
|
"num_tokens": 983623.0,
|
|
"step": 485
|
|
},
|
|
{
|
|
"entropy": 7.859593296051026,
|
|
"epoch": 0.02868012876792508,
|
|
"grad_norm": 0.99609375,
|
|
"learning_rate": 0.0002445,
|
|
"loss": 7.6515,
|
|
"mean_token_accuracy": 0.0777646329253912,
|
|
"num_tokens": 993810.0,
|
|
"step": 490
|
|
},
|
|
{
|
|
"entropy": 7.860140466690064,
|
|
"epoch": 0.02897278314310799,
|
|
"grad_norm": 1.015625,
|
|
"learning_rate": 0.000247,
|
|
"loss": 7.5529,
|
|
"mean_token_accuracy": 0.08074892833828926,
|
|
"num_tokens": 1004155.0,
|
|
"step": 495
|
|
},
|
|
{
|
|
"entropy": 7.809819650650025,
|
|
"epoch": 0.029265437518290898,
|
|
"grad_norm": 1.078125,
|
|
"learning_rate": 0.0002495,
|
|
"loss": 7.5186,
|
|
"mean_token_accuracy": 0.08852879330515862,
|
|
"num_tokens": 1014896.0,
|
|
"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": 1518547636224000.0,
|
|
"train_batch_size": 16,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|