1035 lines
27 KiB
JSON
1035 lines
27 KiB
JSON
{
|
|
"best_global_step": null,
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 0.03304583457255213,
|
|
"eval_steps": 500,
|
|
"global_step": 500,
|
|
"is_hyper_param_search": false,
|
|
"is_local_process_zero": true,
|
|
"is_world_process_zero": true,
|
|
"log_history": [
|
|
{
|
|
"entropy": 10.691955471038819,
|
|
"epoch": 0.0003304583457255213,
|
|
"grad_norm": 12.8125,
|
|
"learning_rate": 2e-06,
|
|
"loss": 10.8273,
|
|
"mean_token_accuracy": 0.0,
|
|
"num_tokens": 10288.0,
|
|
"step": 5
|
|
},
|
|
{
|
|
"entropy": 10.691919994354247,
|
|
"epoch": 0.0006609166914510426,
|
|
"grad_norm": 10.875,
|
|
"learning_rate": 4.5e-06,
|
|
"loss": 10.7719,
|
|
"mean_token_accuracy": 0.0,
|
|
"num_tokens": 20647.0,
|
|
"step": 10
|
|
},
|
|
{
|
|
"entropy": 10.6910964012146,
|
|
"epoch": 0.000991375037176564,
|
|
"grad_norm": 8.6875,
|
|
"learning_rate": 7e-06,
|
|
"loss": 10.5372,
|
|
"mean_token_accuracy": 0.016732628224417567,
|
|
"num_tokens": 30934.0,
|
|
"step": 15
|
|
},
|
|
{
|
|
"entropy": 10.682924270629883,
|
|
"epoch": 0.0013218333829020852,
|
|
"grad_norm": 5.9375,
|
|
"learning_rate": 9.5e-06,
|
|
"loss": 10.2268,
|
|
"mean_token_accuracy": 0.039138264022767544,
|
|
"num_tokens": 41985.0,
|
|
"step": 20
|
|
},
|
|
{
|
|
"entropy": 10.630903339385986,
|
|
"epoch": 0.0016522917286276066,
|
|
"grad_norm": 3.859375,
|
|
"learning_rate": 1.2e-05,
|
|
"loss": 9.9501,
|
|
"mean_token_accuracy": 0.04244728535413742,
|
|
"num_tokens": 52249.0,
|
|
"step": 25
|
|
},
|
|
{
|
|
"entropy": 10.59464464187622,
|
|
"epoch": 0.001982750074353128,
|
|
"grad_norm": 3.171875,
|
|
"learning_rate": 1.4500000000000002e-05,
|
|
"loss": 9.8224,
|
|
"mean_token_accuracy": 0.039565733075141905,
|
|
"num_tokens": 63022.0,
|
|
"step": 30
|
|
},
|
|
{
|
|
"entropy": 10.583414649963379,
|
|
"epoch": 0.002313208420078649,
|
|
"grad_norm": 2.8125,
|
|
"learning_rate": 1.7000000000000003e-05,
|
|
"loss": 9.7332,
|
|
"mean_token_accuracy": 0.042595500499010085,
|
|
"num_tokens": 74862.0,
|
|
"step": 35
|
|
},
|
|
{
|
|
"entropy": 10.582724380493165,
|
|
"epoch": 0.0026436667658041703,
|
|
"grad_norm": 2.484375,
|
|
"learning_rate": 1.95e-05,
|
|
"loss": 9.7531,
|
|
"mean_token_accuracy": 0.03622140642255545,
|
|
"num_tokens": 86588.0,
|
|
"step": 40
|
|
},
|
|
{
|
|
"entropy": 10.594015216827392,
|
|
"epoch": 0.0029741251115296915,
|
|
"grad_norm": 2.5625,
|
|
"learning_rate": 2.2e-05,
|
|
"loss": 9.6536,
|
|
"mean_token_accuracy": 0.037497786805033684,
|
|
"num_tokens": 96248.0,
|
|
"step": 45
|
|
},
|
|
{
|
|
"entropy": 10.578313446044922,
|
|
"epoch": 0.003304583457255213,
|
|
"grad_norm": 2.21875,
|
|
"learning_rate": 2.4500000000000003e-05,
|
|
"loss": 9.6594,
|
|
"mean_token_accuracy": 0.03918336480855942,
|
|
"num_tokens": 107167.0,
|
|
"step": 50
|
|
},
|
|
{
|
|
"entropy": 10.555775451660157,
|
|
"epoch": 0.0036350418029807343,
|
|
"grad_norm": 2.265625,
|
|
"learning_rate": 2.7e-05,
|
|
"loss": 9.5383,
|
|
"mean_token_accuracy": 0.043136011064052585,
|
|
"num_tokens": 117207.0,
|
|
"step": 55
|
|
},
|
|
{
|
|
"entropy": 10.549382877349853,
|
|
"epoch": 0.003965500148706256,
|
|
"grad_norm": 2.109375,
|
|
"learning_rate": 2.95e-05,
|
|
"loss": 9.5206,
|
|
"mean_token_accuracy": 0.04122119285166263,
|
|
"num_tokens": 127335.0,
|
|
"step": 60
|
|
},
|
|
{
|
|
"entropy": 10.5144211769104,
|
|
"epoch": 0.004295958494431777,
|
|
"grad_norm": 2.28125,
|
|
"learning_rate": 3.2e-05,
|
|
"loss": 9.4157,
|
|
"mean_token_accuracy": 0.051365657895803454,
|
|
"num_tokens": 138329.0,
|
|
"step": 65
|
|
},
|
|
{
|
|
"entropy": 10.402010536193847,
|
|
"epoch": 0.004626416840157298,
|
|
"grad_norm": 2.234375,
|
|
"learning_rate": 3.4500000000000005e-05,
|
|
"loss": 9.3531,
|
|
"mean_token_accuracy": 0.04862799309194088,
|
|
"num_tokens": 149993.0,
|
|
"step": 70
|
|
},
|
|
{
|
|
"entropy": 10.443185424804687,
|
|
"epoch": 0.00495687518588282,
|
|
"grad_norm": 2.0625,
|
|
"learning_rate": 3.7e-05,
|
|
"loss": 9.2381,
|
|
"mean_token_accuracy": 0.05048903152346611,
|
|
"num_tokens": 160872.0,
|
|
"step": 75
|
|
},
|
|
{
|
|
"entropy": 10.446312522888183,
|
|
"epoch": 0.005287333531608341,
|
|
"grad_norm": 1.9765625,
|
|
"learning_rate": 3.95e-05,
|
|
"loss": 9.2065,
|
|
"mean_token_accuracy": 0.04706168882548809,
|
|
"num_tokens": 172141.0,
|
|
"step": 80
|
|
},
|
|
{
|
|
"entropy": 10.442552852630616,
|
|
"epoch": 0.005617791877333862,
|
|
"grad_norm": 2.046875,
|
|
"learning_rate": 4.2000000000000004e-05,
|
|
"loss": 9.036,
|
|
"mean_token_accuracy": 0.05399964451789856,
|
|
"num_tokens": 183762.0,
|
|
"step": 85
|
|
},
|
|
{
|
|
"entropy": 10.36781177520752,
|
|
"epoch": 0.005948250223059383,
|
|
"grad_norm": 1.8984375,
|
|
"learning_rate": 4.45e-05,
|
|
"loss": 8.9932,
|
|
"mean_token_accuracy": 0.05456707440316677,
|
|
"num_tokens": 196109.0,
|
|
"step": 90
|
|
},
|
|
{
|
|
"entropy": 10.29437780380249,
|
|
"epoch": 0.006278708568784905,
|
|
"grad_norm": 2.453125,
|
|
"learning_rate": 4.7000000000000004e-05,
|
|
"loss": 8.7929,
|
|
"mean_token_accuracy": 0.061743784323334695,
|
|
"num_tokens": 205671.0,
|
|
"step": 95
|
|
},
|
|
{
|
|
"entropy": 10.216643905639648,
|
|
"epoch": 0.006609166914510426,
|
|
"grad_norm": 1.96875,
|
|
"learning_rate": 4.9500000000000004e-05,
|
|
"loss": 8.7749,
|
|
"mean_token_accuracy": 0.056532309576869014,
|
|
"num_tokens": 217256.0,
|
|
"step": 100
|
|
},
|
|
{
|
|
"entropy": 10.16990032196045,
|
|
"epoch": 0.006939625260235947,
|
|
"grad_norm": 2.0625,
|
|
"learning_rate": 5.2e-05,
|
|
"loss": 8.6032,
|
|
"mean_token_accuracy": 0.06618640683591366,
|
|
"num_tokens": 228066.0,
|
|
"step": 105
|
|
},
|
|
{
|
|
"entropy": 10.083656120300294,
|
|
"epoch": 0.007270083605961469,
|
|
"grad_norm": 1.9609375,
|
|
"learning_rate": 5.45e-05,
|
|
"loss": 8.5398,
|
|
"mean_token_accuracy": 0.06366933546960354,
|
|
"num_tokens": 238643.0,
|
|
"step": 110
|
|
},
|
|
{
|
|
"entropy": 10.01426429748535,
|
|
"epoch": 0.00760054195168699,
|
|
"grad_norm": 1.9453125,
|
|
"learning_rate": 5.7e-05,
|
|
"loss": 8.4763,
|
|
"mean_token_accuracy": 0.0619691651314497,
|
|
"num_tokens": 249812.0,
|
|
"step": 115
|
|
},
|
|
{
|
|
"entropy": 9.912741661071777,
|
|
"epoch": 0.007931000297412512,
|
|
"grad_norm": 1.4921875,
|
|
"learning_rate": 5.9499999999999996e-05,
|
|
"loss": 8.3405,
|
|
"mean_token_accuracy": 0.06653297729790211,
|
|
"num_tokens": 261942.0,
|
|
"step": 120
|
|
},
|
|
{
|
|
"entropy": 9.791018199920654,
|
|
"epoch": 0.008261458643138032,
|
|
"grad_norm": 2.109375,
|
|
"learning_rate": 6.2e-05,
|
|
"loss": 8.179,
|
|
"mean_token_accuracy": 0.07168376296758652,
|
|
"num_tokens": 272537.0,
|
|
"step": 125
|
|
},
|
|
{
|
|
"entropy": 9.592787837982177,
|
|
"epoch": 0.008591916988863553,
|
|
"grad_norm": 1.5078125,
|
|
"learning_rate": 6.450000000000001e-05,
|
|
"loss": 8.0936,
|
|
"mean_token_accuracy": 0.06862132512032985,
|
|
"num_tokens": 283610.0,
|
|
"step": 130
|
|
},
|
|
{
|
|
"entropy": 9.465274620056153,
|
|
"epoch": 0.008922375334589075,
|
|
"grad_norm": 1.2734375,
|
|
"learning_rate": 6.7e-05,
|
|
"loss": 8.0333,
|
|
"mean_token_accuracy": 0.06386486329138279,
|
|
"num_tokens": 294257.0,
|
|
"step": 135
|
|
},
|
|
{
|
|
"entropy": 9.326618099212647,
|
|
"epoch": 0.009252833680314597,
|
|
"grad_norm": 1.4453125,
|
|
"learning_rate": 6.950000000000001e-05,
|
|
"loss": 7.9842,
|
|
"mean_token_accuracy": 0.060935232788324356,
|
|
"num_tokens": 304405.0,
|
|
"step": 140
|
|
},
|
|
{
|
|
"entropy": 9.061574840545655,
|
|
"epoch": 0.009583292026040118,
|
|
"grad_norm": 1.1875,
|
|
"learning_rate": 7.2e-05,
|
|
"loss": 7.7675,
|
|
"mean_token_accuracy": 0.07102037817239762,
|
|
"num_tokens": 314413.0,
|
|
"step": 145
|
|
},
|
|
{
|
|
"entropy": 8.9353346824646,
|
|
"epoch": 0.00991375037176564,
|
|
"grad_norm": 1.3125,
|
|
"learning_rate": 7.45e-05,
|
|
"loss": 7.7801,
|
|
"mean_token_accuracy": 0.06595919616520404,
|
|
"num_tokens": 325010.0,
|
|
"step": 150
|
|
},
|
|
{
|
|
"entropy": 8.705840969085694,
|
|
"epoch": 0.01024420871749116,
|
|
"grad_norm": 1.3359375,
|
|
"learning_rate": 7.7e-05,
|
|
"loss": 7.7226,
|
|
"mean_token_accuracy": 0.06655254885554314,
|
|
"num_tokens": 336421.0,
|
|
"step": 155
|
|
},
|
|
{
|
|
"entropy": 8.54863624572754,
|
|
"epoch": 0.010574667063216681,
|
|
"grad_norm": 1.2890625,
|
|
"learning_rate": 7.950000000000001e-05,
|
|
"loss": 7.5898,
|
|
"mean_token_accuracy": 0.07376853227615357,
|
|
"num_tokens": 346930.0,
|
|
"step": 160
|
|
},
|
|
{
|
|
"entropy": 8.424325942993164,
|
|
"epoch": 0.010905125408942203,
|
|
"grad_norm": 1.3359375,
|
|
"learning_rate": 8.2e-05,
|
|
"loss": 7.6097,
|
|
"mean_token_accuracy": 0.07119092084467411,
|
|
"num_tokens": 357874.0,
|
|
"step": 165
|
|
},
|
|
{
|
|
"entropy": 8.46063985824585,
|
|
"epoch": 0.011235583754667725,
|
|
"grad_norm": 1.203125,
|
|
"learning_rate": 8.450000000000001e-05,
|
|
"loss": 7.6735,
|
|
"mean_token_accuracy": 0.06998510733246803,
|
|
"num_tokens": 370138.0,
|
|
"step": 170
|
|
},
|
|
{
|
|
"entropy": 8.31341323852539,
|
|
"epoch": 0.011566042100393246,
|
|
"grad_norm": 1.2109375,
|
|
"learning_rate": 8.7e-05,
|
|
"loss": 7.6173,
|
|
"mean_token_accuracy": 0.06727928221225739,
|
|
"num_tokens": 379999.0,
|
|
"step": 175
|
|
},
|
|
{
|
|
"entropy": 8.19371862411499,
|
|
"epoch": 0.011896500446118766,
|
|
"grad_norm": 0.96484375,
|
|
"learning_rate": 8.95e-05,
|
|
"loss": 7.4712,
|
|
"mean_token_accuracy": 0.07500850819051266,
|
|
"num_tokens": 392754.0,
|
|
"step": 180
|
|
},
|
|
{
|
|
"entropy": 8.132864665985107,
|
|
"epoch": 0.012226958791844288,
|
|
"grad_norm": 1.0625,
|
|
"learning_rate": 9.2e-05,
|
|
"loss": 7.5206,
|
|
"mean_token_accuracy": 0.06753306314349175,
|
|
"num_tokens": 403771.0,
|
|
"step": 185
|
|
},
|
|
{
|
|
"entropy": 8.108299064636231,
|
|
"epoch": 0.01255741713756981,
|
|
"grad_norm": 2.125,
|
|
"learning_rate": 9.45e-05,
|
|
"loss": 7.5477,
|
|
"mean_token_accuracy": 0.07148384340107442,
|
|
"num_tokens": 415381.0,
|
|
"step": 190
|
|
},
|
|
{
|
|
"entropy": 8.155412006378175,
|
|
"epoch": 0.01288787548329533,
|
|
"grad_norm": 1.3125,
|
|
"learning_rate": 9.7e-05,
|
|
"loss": 7.4227,
|
|
"mean_token_accuracy": 0.07254996970295906,
|
|
"num_tokens": 425919.0,
|
|
"step": 195
|
|
},
|
|
{
|
|
"entropy": 8.072521829605103,
|
|
"epoch": 0.013218333829020852,
|
|
"grad_norm": 1.296875,
|
|
"learning_rate": 9.95e-05,
|
|
"loss": 7.3718,
|
|
"mean_token_accuracy": 0.06771104149520397,
|
|
"num_tokens": 437753.0,
|
|
"step": 200
|
|
},
|
|
{
|
|
"entropy": 7.959475183486939,
|
|
"epoch": 0.013548792174746372,
|
|
"grad_norm": 1.4453125,
|
|
"learning_rate": 0.000102,
|
|
"loss": 7.3961,
|
|
"mean_token_accuracy": 0.07540397495031356,
|
|
"num_tokens": 449079.0,
|
|
"step": 205
|
|
},
|
|
{
|
|
"entropy": 7.956682586669922,
|
|
"epoch": 0.013879250520471894,
|
|
"grad_norm": 1.3671875,
|
|
"learning_rate": 0.00010449999999999999,
|
|
"loss": 7.3773,
|
|
"mean_token_accuracy": 0.07537127286195755,
|
|
"num_tokens": 459250.0,
|
|
"step": 210
|
|
},
|
|
{
|
|
"entropy": 7.86835880279541,
|
|
"epoch": 0.014209708866197416,
|
|
"grad_norm": 1.2265625,
|
|
"learning_rate": 0.000107,
|
|
"loss": 7.3512,
|
|
"mean_token_accuracy": 0.07468458339571953,
|
|
"num_tokens": 470898.0,
|
|
"step": 215
|
|
},
|
|
{
|
|
"entropy": 7.843323755264282,
|
|
"epoch": 0.014540167211922937,
|
|
"grad_norm": 1.1171875,
|
|
"learning_rate": 0.0001095,
|
|
"loss": 7.3402,
|
|
"mean_token_accuracy": 0.07922964245080948,
|
|
"num_tokens": 481116.0,
|
|
"step": 220
|
|
},
|
|
{
|
|
"entropy": 7.847117757797241,
|
|
"epoch": 0.014870625557648459,
|
|
"grad_norm": 1.0859375,
|
|
"learning_rate": 0.000112,
|
|
"loss": 7.349,
|
|
"mean_token_accuracy": 0.07711787447333336,
|
|
"num_tokens": 492639.0,
|
|
"step": 225
|
|
},
|
|
{
|
|
"entropy": 7.853758859634399,
|
|
"epoch": 0.01520108390337398,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 0.0001145,
|
|
"loss": 7.3957,
|
|
"mean_token_accuracy": 0.07615091502666474,
|
|
"num_tokens": 504838.0,
|
|
"step": 230
|
|
},
|
|
{
|
|
"entropy": 7.945042896270752,
|
|
"epoch": 0.0155315422490995,
|
|
"grad_norm": 1.1796875,
|
|
"learning_rate": 0.00011700000000000001,
|
|
"loss": 7.297,
|
|
"mean_token_accuracy": 0.07279803417623043,
|
|
"num_tokens": 515982.0,
|
|
"step": 235
|
|
},
|
|
{
|
|
"entropy": 7.722148132324219,
|
|
"epoch": 0.015862000594825024,
|
|
"grad_norm": 1.1328125,
|
|
"learning_rate": 0.00011949999999999999,
|
|
"loss": 7.2946,
|
|
"mean_token_accuracy": 0.07706804946064949,
|
|
"num_tokens": 526830.0,
|
|
"step": 240
|
|
},
|
|
{
|
|
"entropy": 7.7506262302398685,
|
|
"epoch": 0.016192458940550544,
|
|
"grad_norm": 1.1171875,
|
|
"learning_rate": 0.000122,
|
|
"loss": 7.342,
|
|
"mean_token_accuracy": 0.07662090063095092,
|
|
"num_tokens": 538111.0,
|
|
"step": 245
|
|
},
|
|
{
|
|
"entropy": 7.804240894317627,
|
|
"epoch": 0.016522917286276063,
|
|
"grad_norm": 1.2421875,
|
|
"learning_rate": 0.0001245,
|
|
"loss": 7.2915,
|
|
"mean_token_accuracy": 0.07277994081377984,
|
|
"num_tokens": 548286.0,
|
|
"step": 250
|
|
},
|
|
{
|
|
"entropy": 7.752009725570678,
|
|
"epoch": 0.016853375632001587,
|
|
"grad_norm": 1.21875,
|
|
"learning_rate": 0.000127,
|
|
"loss": 7.2752,
|
|
"mean_token_accuracy": 0.07613983601331711,
|
|
"num_tokens": 558638.0,
|
|
"step": 255
|
|
},
|
|
{
|
|
"entropy": 7.705926036834716,
|
|
"epoch": 0.017183833977727107,
|
|
"grad_norm": 1.0703125,
|
|
"learning_rate": 0.0001295,
|
|
"loss": 7.2402,
|
|
"mean_token_accuracy": 0.07607613280415534,
|
|
"num_tokens": 568909.0,
|
|
"step": 260
|
|
},
|
|
{
|
|
"entropy": 7.722606134414673,
|
|
"epoch": 0.01751429232345263,
|
|
"grad_norm": 1.484375,
|
|
"learning_rate": 0.000132,
|
|
"loss": 7.2178,
|
|
"mean_token_accuracy": 0.07890873812139035,
|
|
"num_tokens": 580392.0,
|
|
"step": 265
|
|
},
|
|
{
|
|
"entropy": 7.688320446014404,
|
|
"epoch": 0.01784475066917815,
|
|
"grad_norm": 1.1640625,
|
|
"learning_rate": 0.00013450000000000002,
|
|
"loss": 7.2767,
|
|
"mean_token_accuracy": 0.07444539405405522,
|
|
"num_tokens": 590160.0,
|
|
"step": 270
|
|
},
|
|
{
|
|
"entropy": 7.632183170318603,
|
|
"epoch": 0.01817520901490367,
|
|
"grad_norm": 1.1171875,
|
|
"learning_rate": 0.00013700000000000002,
|
|
"loss": 7.1822,
|
|
"mean_token_accuracy": 0.07618632428348064,
|
|
"num_tokens": 601054.0,
|
|
"step": 275
|
|
},
|
|
{
|
|
"entropy": 7.7831042289733885,
|
|
"epoch": 0.018505667360629193,
|
|
"grad_norm": 1.3515625,
|
|
"learning_rate": 0.0001395,
|
|
"loss": 7.3189,
|
|
"mean_token_accuracy": 0.07360220067203045,
|
|
"num_tokens": 611908.0,
|
|
"step": 280
|
|
},
|
|
{
|
|
"entropy": 7.640690326690674,
|
|
"epoch": 0.018836125706354713,
|
|
"grad_norm": 1.1953125,
|
|
"learning_rate": 0.00014199999999999998,
|
|
"loss": 7.258,
|
|
"mean_token_accuracy": 0.07853379994630813,
|
|
"num_tokens": 622259.0,
|
|
"step": 285
|
|
},
|
|
{
|
|
"entropy": 7.618532133102417,
|
|
"epoch": 0.019166584052080236,
|
|
"grad_norm": 1.28125,
|
|
"learning_rate": 0.0001445,
|
|
"loss": 7.1215,
|
|
"mean_token_accuracy": 0.08114137947559356,
|
|
"num_tokens": 632728.0,
|
|
"step": 290
|
|
},
|
|
{
|
|
"entropy": 7.7060990810394285,
|
|
"epoch": 0.019497042397805756,
|
|
"grad_norm": 1.4921875,
|
|
"learning_rate": 0.000147,
|
|
"loss": 7.309,
|
|
"mean_token_accuracy": 0.07869038358330727,
|
|
"num_tokens": 644600.0,
|
|
"step": 295
|
|
},
|
|
{
|
|
"entropy": 7.549208354949951,
|
|
"epoch": 0.01982750074353128,
|
|
"grad_norm": 1.4375,
|
|
"learning_rate": 0.0001495,
|
|
"loss": 7.1684,
|
|
"mean_token_accuracy": 0.08070796430110931,
|
|
"num_tokens": 654004.0,
|
|
"step": 300
|
|
},
|
|
{
|
|
"entropy": 7.680343866348267,
|
|
"epoch": 0.0201579590892568,
|
|
"grad_norm": 1.03125,
|
|
"learning_rate": 0.000152,
|
|
"loss": 7.1505,
|
|
"mean_token_accuracy": 0.08140776604413986,
|
|
"num_tokens": 665275.0,
|
|
"step": 305
|
|
},
|
|
{
|
|
"entropy": 7.586634254455566,
|
|
"epoch": 0.02048841743498232,
|
|
"grad_norm": 1.1875,
|
|
"learning_rate": 0.00015450000000000001,
|
|
"loss": 7.2169,
|
|
"mean_token_accuracy": 0.07984307780861855,
|
|
"num_tokens": 677397.0,
|
|
"step": 310
|
|
},
|
|
{
|
|
"entropy": 7.557528400421143,
|
|
"epoch": 0.020818875780707843,
|
|
"grad_norm": 1.1171875,
|
|
"learning_rate": 0.000157,
|
|
"loss": 7.0796,
|
|
"mean_token_accuracy": 0.08578637950122356,
|
|
"num_tokens": 687975.0,
|
|
"step": 315
|
|
},
|
|
{
|
|
"entropy": 7.519621467590332,
|
|
"epoch": 0.021149334126433363,
|
|
"grad_norm": 1.234375,
|
|
"learning_rate": 0.0001595,
|
|
"loss": 7.0386,
|
|
"mean_token_accuracy": 0.09029009193181992,
|
|
"num_tokens": 699102.0,
|
|
"step": 320
|
|
},
|
|
{
|
|
"entropy": 7.4558220386505125,
|
|
"epoch": 0.021479792472158886,
|
|
"grad_norm": 1.1171875,
|
|
"learning_rate": 0.000162,
|
|
"loss": 7.1014,
|
|
"mean_token_accuracy": 0.07989090159535409,
|
|
"num_tokens": 712007.0,
|
|
"step": 325
|
|
},
|
|
{
|
|
"entropy": 7.526741886138916,
|
|
"epoch": 0.021810250817884406,
|
|
"grad_norm": 1.2578125,
|
|
"learning_rate": 0.00016450000000000001,
|
|
"loss": 7.0863,
|
|
"mean_token_accuracy": 0.08748645186424256,
|
|
"num_tokens": 723757.0,
|
|
"step": 330
|
|
},
|
|
{
|
|
"entropy": 7.52968955039978,
|
|
"epoch": 0.022140709163609926,
|
|
"grad_norm": 1.3359375,
|
|
"learning_rate": 0.00016700000000000002,
|
|
"loss": 7.0994,
|
|
"mean_token_accuracy": 0.08689600750803947,
|
|
"num_tokens": 734353.0,
|
|
"step": 335
|
|
},
|
|
{
|
|
"entropy": 7.439078617095947,
|
|
"epoch": 0.02247116750933545,
|
|
"grad_norm": 1.1875,
|
|
"learning_rate": 0.00016950000000000003,
|
|
"loss": 6.9892,
|
|
"mean_token_accuracy": 0.0902569204568863,
|
|
"num_tokens": 744673.0,
|
|
"step": 340
|
|
},
|
|
{
|
|
"entropy": 7.432306814193725,
|
|
"epoch": 0.02280162585506097,
|
|
"grad_norm": 1.390625,
|
|
"learning_rate": 0.00017199999999999998,
|
|
"loss": 7.0258,
|
|
"mean_token_accuracy": 0.0794426791369915,
|
|
"num_tokens": 755181.0,
|
|
"step": 345
|
|
},
|
|
{
|
|
"entropy": 7.557726144790649,
|
|
"epoch": 0.023132084200786492,
|
|
"grad_norm": 1.1328125,
|
|
"learning_rate": 0.00017449999999999999,
|
|
"loss": 7.0575,
|
|
"mean_token_accuracy": 0.08544816374778748,
|
|
"num_tokens": 765558.0,
|
|
"step": 350
|
|
},
|
|
{
|
|
"entropy": 7.414818954467774,
|
|
"epoch": 0.023462542546512012,
|
|
"grad_norm": 1.1328125,
|
|
"learning_rate": 0.000177,
|
|
"loss": 7.0836,
|
|
"mean_token_accuracy": 0.07729014120995999,
|
|
"num_tokens": 778224.0,
|
|
"step": 355
|
|
},
|
|
{
|
|
"entropy": 7.490838050842285,
|
|
"epoch": 0.023793000892237532,
|
|
"grad_norm": 1.21875,
|
|
"learning_rate": 0.0001795,
|
|
"loss": 6.991,
|
|
"mean_token_accuracy": 0.08522340208292008,
|
|
"num_tokens": 789586.0,
|
|
"step": 360
|
|
},
|
|
{
|
|
"entropy": 7.417488718032837,
|
|
"epoch": 0.024123459237963055,
|
|
"grad_norm": 1.015625,
|
|
"learning_rate": 0.000182,
|
|
"loss": 7.0535,
|
|
"mean_token_accuracy": 0.08176884576678276,
|
|
"num_tokens": 801595.0,
|
|
"step": 365
|
|
},
|
|
{
|
|
"entropy": 7.4084100246429445,
|
|
"epoch": 0.024453917583688575,
|
|
"grad_norm": 1.015625,
|
|
"learning_rate": 0.0001845,
|
|
"loss": 7.0797,
|
|
"mean_token_accuracy": 0.08523592539131641,
|
|
"num_tokens": 812850.0,
|
|
"step": 370
|
|
},
|
|
{
|
|
"entropy": 7.462594747543335,
|
|
"epoch": 0.0247843759294141,
|
|
"grad_norm": 1.140625,
|
|
"learning_rate": 0.000187,
|
|
"loss": 7.078,
|
|
"mean_token_accuracy": 0.08119027838110923,
|
|
"num_tokens": 823125.0,
|
|
"step": 375
|
|
},
|
|
{
|
|
"entropy": 7.436292457580566,
|
|
"epoch": 0.02511483427513962,
|
|
"grad_norm": 1.1875,
|
|
"learning_rate": 0.0001895,
|
|
"loss": 6.9898,
|
|
"mean_token_accuracy": 0.08521195650100707,
|
|
"num_tokens": 834838.0,
|
|
"step": 380
|
|
},
|
|
{
|
|
"entropy": 7.3805746078491214,
|
|
"epoch": 0.02544529262086514,
|
|
"grad_norm": 1.359375,
|
|
"learning_rate": 0.000192,
|
|
"loss": 7.0507,
|
|
"mean_token_accuracy": 0.0866511344909668,
|
|
"num_tokens": 845976.0,
|
|
"step": 385
|
|
},
|
|
{
|
|
"entropy": 7.368652486801148,
|
|
"epoch": 0.02577575096659066,
|
|
"grad_norm": 1.15625,
|
|
"learning_rate": 0.0001945,
|
|
"loss": 7.0367,
|
|
"mean_token_accuracy": 0.08296765461564064,
|
|
"num_tokens": 857193.0,
|
|
"step": 390
|
|
},
|
|
{
|
|
"entropy": 7.454722309112549,
|
|
"epoch": 0.02610620931231618,
|
|
"grad_norm": 1.0546875,
|
|
"learning_rate": 0.00019700000000000002,
|
|
"loss": 7.0047,
|
|
"mean_token_accuracy": 0.08462346903979778,
|
|
"num_tokens": 868471.0,
|
|
"step": 395
|
|
},
|
|
{
|
|
"entropy": 7.42730746269226,
|
|
"epoch": 0.026436667658041705,
|
|
"grad_norm": 1.2578125,
|
|
"learning_rate": 0.00019950000000000002,
|
|
"loss": 6.9478,
|
|
"mean_token_accuracy": 0.08773355185985565,
|
|
"num_tokens": 879339.0,
|
|
"step": 400
|
|
},
|
|
{
|
|
"entropy": 7.206215667724609,
|
|
"epoch": 0.026767126003767225,
|
|
"grad_norm": 1.078125,
|
|
"learning_rate": 0.000202,
|
|
"loss": 6.8858,
|
|
"mean_token_accuracy": 0.08911687806248665,
|
|
"num_tokens": 890081.0,
|
|
"step": 405
|
|
},
|
|
{
|
|
"entropy": 7.307197189331054,
|
|
"epoch": 0.027097584349492745,
|
|
"grad_norm": 1.2421875,
|
|
"learning_rate": 0.00020449999999999998,
|
|
"loss": 6.909,
|
|
"mean_token_accuracy": 0.08829740136861801,
|
|
"num_tokens": 899967.0,
|
|
"step": 410
|
|
},
|
|
{
|
|
"entropy": 7.394805383682251,
|
|
"epoch": 0.027428042695218268,
|
|
"grad_norm": 1.171875,
|
|
"learning_rate": 0.000207,
|
|
"loss": 6.9474,
|
|
"mean_token_accuracy": 0.08440508171916009,
|
|
"num_tokens": 910697.0,
|
|
"step": 415
|
|
},
|
|
{
|
|
"entropy": 7.307165718078613,
|
|
"epoch": 0.027758501040943788,
|
|
"grad_norm": 1.21875,
|
|
"learning_rate": 0.0002095,
|
|
"loss": 7.0184,
|
|
"mean_token_accuracy": 0.08860758394002914,
|
|
"num_tokens": 920879.0,
|
|
"step": 420
|
|
},
|
|
{
|
|
"entropy": 7.332070398330688,
|
|
"epoch": 0.02808895938666931,
|
|
"grad_norm": 1.1796875,
|
|
"learning_rate": 0.000212,
|
|
"loss": 6.93,
|
|
"mean_token_accuracy": 0.0954623855650425,
|
|
"num_tokens": 931566.0,
|
|
"step": 425
|
|
},
|
|
{
|
|
"entropy": 7.345228767395019,
|
|
"epoch": 0.02841941773239483,
|
|
"grad_norm": 1.0078125,
|
|
"learning_rate": 0.0002145,
|
|
"loss": 6.868,
|
|
"mean_token_accuracy": 0.08827793225646019,
|
|
"num_tokens": 941736.0,
|
|
"step": 430
|
|
},
|
|
{
|
|
"entropy": 7.245734167098999,
|
|
"epoch": 0.028749876078120355,
|
|
"grad_norm": 1.234375,
|
|
"learning_rate": 0.00021700000000000002,
|
|
"loss": 6.9063,
|
|
"mean_token_accuracy": 0.08495340272784233,
|
|
"num_tokens": 951652.0,
|
|
"step": 435
|
|
},
|
|
{
|
|
"entropy": 7.267821741104126,
|
|
"epoch": 0.029080334423845874,
|
|
"grad_norm": 1.109375,
|
|
"learning_rate": 0.0002195,
|
|
"loss": 6.8298,
|
|
"mean_token_accuracy": 0.08889664858579635,
|
|
"num_tokens": 963242.0,
|
|
"step": 440
|
|
},
|
|
{
|
|
"entropy": 7.126494026184082,
|
|
"epoch": 0.029410792769571394,
|
|
"grad_norm": 1.1875,
|
|
"learning_rate": 0.000222,
|
|
"loss": 6.8716,
|
|
"mean_token_accuracy": 0.08861989453434944,
|
|
"num_tokens": 974078.0,
|
|
"step": 445
|
|
},
|
|
{
|
|
"entropy": 7.203296804428101,
|
|
"epoch": 0.029741251115296918,
|
|
"grad_norm": 0.94140625,
|
|
"learning_rate": 0.0002245,
|
|
"loss": 6.8335,
|
|
"mean_token_accuracy": 0.0920184887945652,
|
|
"num_tokens": 986324.0,
|
|
"step": 450
|
|
},
|
|
{
|
|
"entropy": 7.216691255569458,
|
|
"epoch": 0.030071709461022438,
|
|
"grad_norm": 1.0546875,
|
|
"learning_rate": 0.00022700000000000002,
|
|
"loss": 6.8409,
|
|
"mean_token_accuracy": 0.09037284776568413,
|
|
"num_tokens": 997625.0,
|
|
"step": 455
|
|
},
|
|
{
|
|
"entropy": 7.202143144607544,
|
|
"epoch": 0.03040216780674796,
|
|
"grad_norm": 1.1171875,
|
|
"learning_rate": 0.00022950000000000002,
|
|
"loss": 6.8316,
|
|
"mean_token_accuracy": 0.09061756655573845,
|
|
"num_tokens": 1007914.0,
|
|
"step": 460
|
|
},
|
|
{
|
|
"entropy": 7.096453285217285,
|
|
"epoch": 0.03073262615247348,
|
|
"grad_norm": 1.0,
|
|
"learning_rate": 0.00023200000000000003,
|
|
"loss": 6.792,
|
|
"mean_token_accuracy": 0.0925156094133854,
|
|
"num_tokens": 1018510.0,
|
|
"step": 465
|
|
},
|
|
{
|
|
"entropy": 7.169570064544677,
|
|
"epoch": 0.031063084498199,
|
|
"grad_norm": 1.2578125,
|
|
"learning_rate": 0.00023449999999999998,
|
|
"loss": 6.7677,
|
|
"mean_token_accuracy": 0.09652838632464408,
|
|
"num_tokens": 1028532.0,
|
|
"step": 470
|
|
},
|
|
{
|
|
"entropy": 7.168693256378174,
|
|
"epoch": 0.031393542843924524,
|
|
"grad_norm": 1.09375,
|
|
"learning_rate": 0.000237,
|
|
"loss": 6.9262,
|
|
"mean_token_accuracy": 0.08653468191623688,
|
|
"num_tokens": 1040252.0,
|
|
"step": 475
|
|
},
|
|
{
|
|
"entropy": 7.152310705184936,
|
|
"epoch": 0.03172400118965005,
|
|
"grad_norm": 1.078125,
|
|
"learning_rate": 0.0002395,
|
|
"loss": 6.7222,
|
|
"mean_token_accuracy": 0.09424662217497826,
|
|
"num_tokens": 1050486.0,
|
|
"step": 480
|
|
},
|
|
{
|
|
"entropy": 7.076179075241089,
|
|
"epoch": 0.032054459535375564,
|
|
"grad_norm": 1.2109375,
|
|
"learning_rate": 0.000242,
|
|
"loss": 6.766,
|
|
"mean_token_accuracy": 0.09703566282987594,
|
|
"num_tokens": 1061218.0,
|
|
"step": 485
|
|
},
|
|
{
|
|
"entropy": 7.191670942306518,
|
|
"epoch": 0.03238491788110109,
|
|
"grad_norm": 1.265625,
|
|
"learning_rate": 0.0002445,
|
|
"loss": 6.888,
|
|
"mean_token_accuracy": 0.0887337252497673,
|
|
"num_tokens": 1071077.0,
|
|
"step": 490
|
|
},
|
|
{
|
|
"entropy": 7.052482748031617,
|
|
"epoch": 0.03271537622682661,
|
|
"grad_norm": 1.6328125,
|
|
"learning_rate": 0.000247,
|
|
"loss": 6.8699,
|
|
"mean_token_accuracy": 0.08658004626631736,
|
|
"num_tokens": 1081434.0,
|
|
"step": 495
|
|
},
|
|
{
|
|
"entropy": 7.08971209526062,
|
|
"epoch": 0.03304583457255213,
|
|
"grad_norm": 1.21875,
|
|
"learning_rate": 0.0002495,
|
|
"loss": 6.811,
|
|
"mean_token_accuracy": 0.09111799448728561,
|
|
"num_tokens": 1092237.0,
|
|
"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": 1792332693504000.0,
|
|
"train_batch_size": 16,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|