1035 lines
27 KiB
JSON
1035 lines
27 KiB
JSON
{
|
|
"best_global_step": null,
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 0.011532163203173651,
|
|
"eval_steps": 500,
|
|
"global_step": 500,
|
|
"is_hyper_param_search": false,
|
|
"is_local_process_zero": true,
|
|
"is_world_process_zero": true,
|
|
"log_history": [
|
|
{
|
|
"entropy": 10.692046642303467,
|
|
"epoch": 0.0001153216320317365,
|
|
"grad_norm": 16.625,
|
|
"learning_rate": 2e-06,
|
|
"loss": 10.4811,
|
|
"mean_token_accuracy": 0.0,
|
|
"num_tokens": 3293.0,
|
|
"step": 5
|
|
},
|
|
{
|
|
"entropy": 10.692021083831786,
|
|
"epoch": 0.000230643264063473,
|
|
"grad_norm": 18.875,
|
|
"learning_rate": 4.5e-06,
|
|
"loss": 10.4344,
|
|
"mean_token_accuracy": 0.0,
|
|
"num_tokens": 6743.0,
|
|
"step": 10
|
|
},
|
|
{
|
|
"entropy": 10.691608142852782,
|
|
"epoch": 0.0003459648960952095,
|
|
"grad_norm": 13.25,
|
|
"learning_rate": 7e-06,
|
|
"loss": 10.184,
|
|
"mean_token_accuracy": 0.01596597593743354,
|
|
"num_tokens": 10261.0,
|
|
"step": 15
|
|
},
|
|
{
|
|
"entropy": 10.686354446411134,
|
|
"epoch": 0.000461286528126946,
|
|
"grad_norm": 9.4375,
|
|
"learning_rate": 9.5e-06,
|
|
"loss": 9.8654,
|
|
"mean_token_accuracy": 0.04747712537646294,
|
|
"num_tokens": 14153.0,
|
|
"step": 20
|
|
},
|
|
{
|
|
"entropy": 10.65447473526001,
|
|
"epoch": 0.0005766081601586826,
|
|
"grad_norm": 6.3125,
|
|
"learning_rate": 1.2e-05,
|
|
"loss": 9.5581,
|
|
"mean_token_accuracy": 0.0560380831360817,
|
|
"num_tokens": 17774.0,
|
|
"step": 25
|
|
},
|
|
{
|
|
"entropy": 10.609045124053955,
|
|
"epoch": 0.000691929792190419,
|
|
"grad_norm": 4.75,
|
|
"learning_rate": 1.4500000000000002e-05,
|
|
"loss": 9.3551,
|
|
"mean_token_accuracy": 0.06091434173285961,
|
|
"num_tokens": 21521.0,
|
|
"step": 30
|
|
},
|
|
{
|
|
"entropy": 10.577239322662354,
|
|
"epoch": 0.0008072514242221556,
|
|
"grad_norm": 4.75,
|
|
"learning_rate": 1.7000000000000003e-05,
|
|
"loss": 9.2179,
|
|
"mean_token_accuracy": 0.06620409153401852,
|
|
"num_tokens": 24880.0,
|
|
"step": 35
|
|
},
|
|
{
|
|
"entropy": 10.546881198883057,
|
|
"epoch": 0.000922573056253892,
|
|
"grad_norm": 4.21875,
|
|
"learning_rate": 1.95e-05,
|
|
"loss": 9.1091,
|
|
"mean_token_accuracy": 0.06204554140567779,
|
|
"num_tokens": 28167.0,
|
|
"step": 40
|
|
},
|
|
{
|
|
"entropy": 10.515935897827148,
|
|
"epoch": 0.0010378946882856287,
|
|
"grad_norm": 3.40625,
|
|
"learning_rate": 2.2e-05,
|
|
"loss": 9.1072,
|
|
"mean_token_accuracy": 0.06435491405427456,
|
|
"num_tokens": 31745.0,
|
|
"step": 45
|
|
},
|
|
{
|
|
"entropy": 10.450439167022704,
|
|
"epoch": 0.0011532163203173652,
|
|
"grad_norm": 4.125,
|
|
"learning_rate": 2.4500000000000003e-05,
|
|
"loss": 9.0469,
|
|
"mean_token_accuracy": 0.06713954657316208,
|
|
"num_tokens": 35270.0,
|
|
"step": 50
|
|
},
|
|
{
|
|
"entropy": 10.369979953765869,
|
|
"epoch": 0.0012685379523491016,
|
|
"grad_norm": 3.328125,
|
|
"learning_rate": 2.7e-05,
|
|
"loss": 9.0259,
|
|
"mean_token_accuracy": 0.07268906272947788,
|
|
"num_tokens": 38943.0,
|
|
"step": 55
|
|
},
|
|
{
|
|
"entropy": 10.2349365234375,
|
|
"epoch": 0.001383859584380838,
|
|
"grad_norm": 4.1875,
|
|
"learning_rate": 2.95e-05,
|
|
"loss": 8.8073,
|
|
"mean_token_accuracy": 0.07592999711632728,
|
|
"num_tokens": 42299.0,
|
|
"step": 60
|
|
},
|
|
{
|
|
"entropy": 10.174158954620362,
|
|
"epoch": 0.0014991812164125747,
|
|
"grad_norm": 2.984375,
|
|
"learning_rate": 3.2e-05,
|
|
"loss": 8.8684,
|
|
"mean_token_accuracy": 0.07179933935403823,
|
|
"num_tokens": 45661.0,
|
|
"step": 65
|
|
},
|
|
{
|
|
"entropy": 10.246811008453369,
|
|
"epoch": 0.0016145028484443112,
|
|
"grad_norm": 2.609375,
|
|
"learning_rate": 3.4500000000000005e-05,
|
|
"loss": 8.951,
|
|
"mean_token_accuracy": 0.061866828426718715,
|
|
"num_tokens": 49683.0,
|
|
"step": 70
|
|
},
|
|
{
|
|
"entropy": 10.264655303955077,
|
|
"epoch": 0.0017298244804760477,
|
|
"grad_norm": 2.921875,
|
|
"learning_rate": 3.7e-05,
|
|
"loss": 8.716,
|
|
"mean_token_accuracy": 0.07860686406493186,
|
|
"num_tokens": 52752.0,
|
|
"step": 75
|
|
},
|
|
{
|
|
"entropy": 10.173390865325928,
|
|
"epoch": 0.001845146112507784,
|
|
"grad_norm": 2.546875,
|
|
"learning_rate": 3.95e-05,
|
|
"loss": 8.6778,
|
|
"mean_token_accuracy": 0.07509532310068608,
|
|
"num_tokens": 56367.0,
|
|
"step": 80
|
|
},
|
|
{
|
|
"entropy": 10.120823097229003,
|
|
"epoch": 0.0019604677445395208,
|
|
"grad_norm": 3.46875,
|
|
"learning_rate": 4.2000000000000004e-05,
|
|
"loss": 8.596,
|
|
"mean_token_accuracy": 0.08393591865897179,
|
|
"num_tokens": 59967.0,
|
|
"step": 85
|
|
},
|
|
{
|
|
"entropy": 10.119685459136964,
|
|
"epoch": 0.0020757893765712574,
|
|
"grad_norm": 2.421875,
|
|
"learning_rate": 4.45e-05,
|
|
"loss": 8.5686,
|
|
"mean_token_accuracy": 0.07523884475231171,
|
|
"num_tokens": 63634.0,
|
|
"step": 90
|
|
},
|
|
{
|
|
"entropy": 10.038768196105957,
|
|
"epoch": 0.0021911110086029937,
|
|
"grad_norm": 3.015625,
|
|
"learning_rate": 4.7000000000000004e-05,
|
|
"loss": 8.3464,
|
|
"mean_token_accuracy": 0.0838980995118618,
|
|
"num_tokens": 66883.0,
|
|
"step": 95
|
|
},
|
|
{
|
|
"entropy": 9.92626495361328,
|
|
"epoch": 0.0023064326406347303,
|
|
"grad_norm": 2.671875,
|
|
"learning_rate": 4.9500000000000004e-05,
|
|
"loss": 8.3612,
|
|
"mean_token_accuracy": 0.0789190474897623,
|
|
"num_tokens": 70491.0,
|
|
"step": 100
|
|
},
|
|
{
|
|
"entropy": 9.932885265350341,
|
|
"epoch": 0.0024217542726664666,
|
|
"grad_norm": 2.484375,
|
|
"learning_rate": 5.2e-05,
|
|
"loss": 8.2115,
|
|
"mean_token_accuracy": 0.08061706572771073,
|
|
"num_tokens": 73852.0,
|
|
"step": 105
|
|
},
|
|
{
|
|
"entropy": 9.765224361419678,
|
|
"epoch": 0.0025370759046982033,
|
|
"grad_norm": 2.0,
|
|
"learning_rate": 5.45e-05,
|
|
"loss": 8.164,
|
|
"mean_token_accuracy": 0.0914115659892559,
|
|
"num_tokens": 77442.0,
|
|
"step": 110
|
|
},
|
|
{
|
|
"entropy": 9.612014007568359,
|
|
"epoch": 0.00265239753672994,
|
|
"grad_norm": 1.90625,
|
|
"learning_rate": 5.7e-05,
|
|
"loss": 8.0555,
|
|
"mean_token_accuracy": 0.08430748134851455,
|
|
"num_tokens": 80871.0,
|
|
"step": 115
|
|
},
|
|
{
|
|
"entropy": 9.554380130767822,
|
|
"epoch": 0.002767719168761676,
|
|
"grad_norm": 2.0,
|
|
"learning_rate": 5.9499999999999996e-05,
|
|
"loss": 7.9495,
|
|
"mean_token_accuracy": 0.09184777326881885,
|
|
"num_tokens": 84444.0,
|
|
"step": 120
|
|
},
|
|
{
|
|
"entropy": 9.397439956665039,
|
|
"epoch": 0.002883040800793413,
|
|
"grad_norm": 1.9140625,
|
|
"learning_rate": 6.2e-05,
|
|
"loss": 7.9236,
|
|
"mean_token_accuracy": 0.0814009454101324,
|
|
"num_tokens": 87939.0,
|
|
"step": 125
|
|
},
|
|
{
|
|
"entropy": 9.200541114807129,
|
|
"epoch": 0.0029983624328251495,
|
|
"grad_norm": 2.046875,
|
|
"learning_rate": 6.450000000000001e-05,
|
|
"loss": 7.6689,
|
|
"mean_token_accuracy": 0.08414206653833389,
|
|
"num_tokens": 91189.0,
|
|
"step": 130
|
|
},
|
|
{
|
|
"entropy": 9.055619144439698,
|
|
"epoch": 0.0031136840648568857,
|
|
"grad_norm": 1.9140625,
|
|
"learning_rate": 6.7e-05,
|
|
"loss": 7.6462,
|
|
"mean_token_accuracy": 0.08812693618237973,
|
|
"num_tokens": 94777.0,
|
|
"step": 135
|
|
},
|
|
{
|
|
"entropy": 9.0249493598938,
|
|
"epoch": 0.0032290056968886224,
|
|
"grad_norm": 2.046875,
|
|
"learning_rate": 6.950000000000001e-05,
|
|
"loss": 7.6653,
|
|
"mean_token_accuracy": 0.08522222451865673,
|
|
"num_tokens": 98581.0,
|
|
"step": 140
|
|
},
|
|
{
|
|
"entropy": 8.796249675750733,
|
|
"epoch": 0.003344327328920359,
|
|
"grad_norm": 2.046875,
|
|
"learning_rate": 7.2e-05,
|
|
"loss": 7.6153,
|
|
"mean_token_accuracy": 0.08283311799168587,
|
|
"num_tokens": 102333.0,
|
|
"step": 145
|
|
},
|
|
{
|
|
"entropy": 8.677543926239014,
|
|
"epoch": 0.0034596489609520953,
|
|
"grad_norm": 1.484375,
|
|
"learning_rate": 7.45e-05,
|
|
"loss": 7.5304,
|
|
"mean_token_accuracy": 0.08452097214758396,
|
|
"num_tokens": 106267.0,
|
|
"step": 150
|
|
},
|
|
{
|
|
"entropy": 8.623601245880128,
|
|
"epoch": 0.003574970592983832,
|
|
"grad_norm": 2.046875,
|
|
"learning_rate": 7.7e-05,
|
|
"loss": 7.4307,
|
|
"mean_token_accuracy": 0.08808897696435451,
|
|
"num_tokens": 109989.0,
|
|
"step": 155
|
|
},
|
|
{
|
|
"entropy": 8.373943901062011,
|
|
"epoch": 0.003690292225015568,
|
|
"grad_norm": 1.8671875,
|
|
"learning_rate": 7.950000000000001e-05,
|
|
"loss": 7.4063,
|
|
"mean_token_accuracy": 0.08707165420055389,
|
|
"num_tokens": 113378.0,
|
|
"step": 160
|
|
},
|
|
{
|
|
"entropy": 8.395255756378173,
|
|
"epoch": 0.003805613857047305,
|
|
"grad_norm": 1.609375,
|
|
"learning_rate": 8.2e-05,
|
|
"loss": 7.3032,
|
|
"mean_token_accuracy": 0.09222758859395981,
|
|
"num_tokens": 116736.0,
|
|
"step": 165
|
|
},
|
|
{
|
|
"entropy": 8.167229080200196,
|
|
"epoch": 0.0039209354890790415,
|
|
"grad_norm": 1.6875,
|
|
"learning_rate": 8.450000000000001e-05,
|
|
"loss": 7.2397,
|
|
"mean_token_accuracy": 0.08980661779642105,
|
|
"num_tokens": 120369.0,
|
|
"step": 170
|
|
},
|
|
{
|
|
"entropy": 8.110415983200074,
|
|
"epoch": 0.004036257121110778,
|
|
"grad_norm": 1.9140625,
|
|
"learning_rate": 8.7e-05,
|
|
"loss": 7.2388,
|
|
"mean_token_accuracy": 0.10482370853424072,
|
|
"num_tokens": 124003.0,
|
|
"step": 175
|
|
},
|
|
{
|
|
"entropy": 8.005911922454834,
|
|
"epoch": 0.004151578753142515,
|
|
"grad_norm": 1.3828125,
|
|
"learning_rate": 8.95e-05,
|
|
"loss": 7.2327,
|
|
"mean_token_accuracy": 0.09916695356369018,
|
|
"num_tokens": 127611.0,
|
|
"step": 180
|
|
},
|
|
{
|
|
"entropy": 8.034901094436645,
|
|
"epoch": 0.004266900385174251,
|
|
"grad_norm": 1.7734375,
|
|
"learning_rate": 9.2e-05,
|
|
"loss": 7.255,
|
|
"mean_token_accuracy": 0.09181497618556023,
|
|
"num_tokens": 130837.0,
|
|
"step": 185
|
|
},
|
|
{
|
|
"entropy": 7.924763011932373,
|
|
"epoch": 0.004382222017205987,
|
|
"grad_norm": 1.8984375,
|
|
"learning_rate": 9.45e-05,
|
|
"loss": 7.2781,
|
|
"mean_token_accuracy": 0.09120299257338046,
|
|
"num_tokens": 134801.0,
|
|
"step": 190
|
|
},
|
|
{
|
|
"entropy": 8.028991937637329,
|
|
"epoch": 0.004497543649237724,
|
|
"grad_norm": 1.71875,
|
|
"learning_rate": 9.7e-05,
|
|
"loss": 7.3143,
|
|
"mean_token_accuracy": 0.09299080073833466,
|
|
"num_tokens": 138479.0,
|
|
"step": 195
|
|
},
|
|
{
|
|
"entropy": 7.826745653152466,
|
|
"epoch": 0.004612865281269461,
|
|
"grad_norm": 1.5625,
|
|
"learning_rate": 9.95e-05,
|
|
"loss": 7.2351,
|
|
"mean_token_accuracy": 0.09592275992035866,
|
|
"num_tokens": 142536.0,
|
|
"step": 200
|
|
},
|
|
{
|
|
"entropy": 7.951525592803955,
|
|
"epoch": 0.004728186913301197,
|
|
"grad_norm": 1.9609375,
|
|
"learning_rate": 0.000102,
|
|
"loss": 7.2783,
|
|
"mean_token_accuracy": 0.09658894389867782,
|
|
"num_tokens": 146633.0,
|
|
"step": 205
|
|
},
|
|
{
|
|
"entropy": 7.681295728683471,
|
|
"epoch": 0.004843508545332933,
|
|
"grad_norm": 1.84375,
|
|
"learning_rate": 0.00010449999999999999,
|
|
"loss": 7.1036,
|
|
"mean_token_accuracy": 0.09946461021900177,
|
|
"num_tokens": 150070.0,
|
|
"step": 210
|
|
},
|
|
{
|
|
"entropy": 7.870249176025391,
|
|
"epoch": 0.00495883017736467,
|
|
"grad_norm": 1.8515625,
|
|
"learning_rate": 0.000107,
|
|
"loss": 7.2654,
|
|
"mean_token_accuracy": 0.08741186410188675,
|
|
"num_tokens": 153698.0,
|
|
"step": 215
|
|
},
|
|
{
|
|
"entropy": 7.710623836517334,
|
|
"epoch": 0.0050741518093964065,
|
|
"grad_norm": 1.671875,
|
|
"learning_rate": 0.0001095,
|
|
"loss": 7.1792,
|
|
"mean_token_accuracy": 0.09874420538544655,
|
|
"num_tokens": 157381.0,
|
|
"step": 220
|
|
},
|
|
{
|
|
"entropy": 7.73014268875122,
|
|
"epoch": 0.005189473441428143,
|
|
"grad_norm": 1.6015625,
|
|
"learning_rate": 0.000112,
|
|
"loss": 7.1647,
|
|
"mean_token_accuracy": 0.09424602091312409,
|
|
"num_tokens": 160862.0,
|
|
"step": 225
|
|
},
|
|
{
|
|
"entropy": 7.8334917545318605,
|
|
"epoch": 0.00530479507345988,
|
|
"grad_norm": 1.578125,
|
|
"learning_rate": 0.0001145,
|
|
"loss": 7.1663,
|
|
"mean_token_accuracy": 0.0998285748064518,
|
|
"num_tokens": 164231.0,
|
|
"step": 230
|
|
},
|
|
{
|
|
"entropy": 7.634814500808716,
|
|
"epoch": 0.0054201167054916165,
|
|
"grad_norm": 1.7265625,
|
|
"learning_rate": 0.00011700000000000001,
|
|
"loss": 7.2333,
|
|
"mean_token_accuracy": 0.08781378865242004,
|
|
"num_tokens": 168351.0,
|
|
"step": 235
|
|
},
|
|
{
|
|
"entropy": 7.6461118221282955,
|
|
"epoch": 0.005535438337523352,
|
|
"grad_norm": 1.75,
|
|
"learning_rate": 0.00011949999999999999,
|
|
"loss": 7.0404,
|
|
"mean_token_accuracy": 0.10981863886117935,
|
|
"num_tokens": 171645.0,
|
|
"step": 240
|
|
},
|
|
{
|
|
"entropy": 7.518331480026245,
|
|
"epoch": 0.005650759969555089,
|
|
"grad_norm": 1.53125,
|
|
"learning_rate": 0.000122,
|
|
"loss": 7.0655,
|
|
"mean_token_accuracy": 0.10514177232980729,
|
|
"num_tokens": 175279.0,
|
|
"step": 245
|
|
},
|
|
{
|
|
"entropy": 7.588093042373657,
|
|
"epoch": 0.005766081601586826,
|
|
"grad_norm": 1.4765625,
|
|
"learning_rate": 0.0001245,
|
|
"loss": 7.1508,
|
|
"mean_token_accuracy": 0.0997185155749321,
|
|
"num_tokens": 179111.0,
|
|
"step": 250
|
|
},
|
|
{
|
|
"entropy": 7.750802040100098,
|
|
"epoch": 0.005881403233618562,
|
|
"grad_norm": 1.75,
|
|
"learning_rate": 0.000127,
|
|
"loss": 7.1045,
|
|
"mean_token_accuracy": 0.10787053629755974,
|
|
"num_tokens": 182369.0,
|
|
"step": 255
|
|
},
|
|
{
|
|
"entropy": 7.515914821624756,
|
|
"epoch": 0.005996724865650299,
|
|
"grad_norm": 2.125,
|
|
"learning_rate": 0.0001295,
|
|
"loss": 7.0753,
|
|
"mean_token_accuracy": 0.11265580952167512,
|
|
"num_tokens": 185671.0,
|
|
"step": 260
|
|
},
|
|
{
|
|
"entropy": 7.501624631881714,
|
|
"epoch": 0.006112046497682035,
|
|
"grad_norm": 1.671875,
|
|
"learning_rate": 0.000132,
|
|
"loss": 6.9932,
|
|
"mean_token_accuracy": 0.1023002490401268,
|
|
"num_tokens": 189295.0,
|
|
"step": 265
|
|
},
|
|
{
|
|
"entropy": 7.487598466873169,
|
|
"epoch": 0.0062273681297137715,
|
|
"grad_norm": 1.46875,
|
|
"learning_rate": 0.00013450000000000002,
|
|
"loss": 7.0035,
|
|
"mean_token_accuracy": 0.10346534922719001,
|
|
"num_tokens": 192791.0,
|
|
"step": 270
|
|
},
|
|
{
|
|
"entropy": 7.491527700424195,
|
|
"epoch": 0.006342689761745508,
|
|
"grad_norm": 2.09375,
|
|
"learning_rate": 0.00013700000000000002,
|
|
"loss": 7.0257,
|
|
"mean_token_accuracy": 0.10039645433425903,
|
|
"num_tokens": 196550.0,
|
|
"step": 275
|
|
},
|
|
{
|
|
"entropy": 7.525121784210205,
|
|
"epoch": 0.006458011393777245,
|
|
"grad_norm": 1.53125,
|
|
"learning_rate": 0.0001395,
|
|
"loss": 7.0112,
|
|
"mean_token_accuracy": 0.10743131414055825,
|
|
"num_tokens": 200274.0,
|
|
"step": 280
|
|
},
|
|
{
|
|
"entropy": 7.3943414211273195,
|
|
"epoch": 0.0065733330258089815,
|
|
"grad_norm": 1.4453125,
|
|
"learning_rate": 0.00014199999999999998,
|
|
"loss": 6.9813,
|
|
"mean_token_accuracy": 0.11192047148942948,
|
|
"num_tokens": 203748.0,
|
|
"step": 285
|
|
},
|
|
{
|
|
"entropy": 7.610325717926026,
|
|
"epoch": 0.006688654657840718,
|
|
"grad_norm": 1.8125,
|
|
"learning_rate": 0.0001445,
|
|
"loss": 6.9901,
|
|
"mean_token_accuracy": 0.10699754059314728,
|
|
"num_tokens": 207203.0,
|
|
"step": 290
|
|
},
|
|
{
|
|
"entropy": 7.490588426589966,
|
|
"epoch": 0.006803976289872454,
|
|
"grad_norm": 1.609375,
|
|
"learning_rate": 0.000147,
|
|
"loss": 7.1777,
|
|
"mean_token_accuracy": 0.09906014949083328,
|
|
"num_tokens": 210761.0,
|
|
"step": 295
|
|
},
|
|
{
|
|
"entropy": 7.500645971298217,
|
|
"epoch": 0.006919297921904191,
|
|
"grad_norm": 1.6640625,
|
|
"learning_rate": 0.0001495,
|
|
"loss": 7.0411,
|
|
"mean_token_accuracy": 0.11081741005182266,
|
|
"num_tokens": 214362.0,
|
|
"step": 300
|
|
},
|
|
{
|
|
"entropy": 7.4556807518005375,
|
|
"epoch": 0.007034619553935927,
|
|
"grad_norm": 1.921875,
|
|
"learning_rate": 0.000152,
|
|
"loss": 7.0964,
|
|
"mean_token_accuracy": 0.10465478524565697,
|
|
"num_tokens": 217938.0,
|
|
"step": 305
|
|
},
|
|
{
|
|
"entropy": 7.400477266311645,
|
|
"epoch": 0.007149941185967664,
|
|
"grad_norm": 1.7890625,
|
|
"learning_rate": 0.00015450000000000001,
|
|
"loss": 7.0187,
|
|
"mean_token_accuracy": 0.102249476313591,
|
|
"num_tokens": 221519.0,
|
|
"step": 310
|
|
},
|
|
{
|
|
"entropy": 7.4884284973144535,
|
|
"epoch": 0.007265262817999401,
|
|
"grad_norm": 1.65625,
|
|
"learning_rate": 0.000157,
|
|
"loss": 6.9382,
|
|
"mean_token_accuracy": 0.11059758365154267,
|
|
"num_tokens": 224918.0,
|
|
"step": 315
|
|
},
|
|
{
|
|
"entropy": 7.453373765945434,
|
|
"epoch": 0.007380584450031136,
|
|
"grad_norm": 1.7578125,
|
|
"learning_rate": 0.0001595,
|
|
"loss": 6.97,
|
|
"mean_token_accuracy": 0.11169943362474441,
|
|
"num_tokens": 228334.0,
|
|
"step": 320
|
|
},
|
|
{
|
|
"entropy": 7.396696710586548,
|
|
"epoch": 0.007495906082062873,
|
|
"grad_norm": 1.4765625,
|
|
"learning_rate": 0.000162,
|
|
"loss": 7.0012,
|
|
"mean_token_accuracy": 0.11026087254285813,
|
|
"num_tokens": 231826.0,
|
|
"step": 325
|
|
},
|
|
{
|
|
"entropy": 7.417343282699585,
|
|
"epoch": 0.00761122771409461,
|
|
"grad_norm": 1.625,
|
|
"learning_rate": 0.00016450000000000001,
|
|
"loss": 6.9298,
|
|
"mean_token_accuracy": 0.11452125087380409,
|
|
"num_tokens": 235519.0,
|
|
"step": 330
|
|
},
|
|
{
|
|
"entropy": 7.386319303512574,
|
|
"epoch": 0.007726549346126346,
|
|
"grad_norm": 1.8828125,
|
|
"learning_rate": 0.00016700000000000002,
|
|
"loss": 7.1777,
|
|
"mean_token_accuracy": 0.10753077790141105,
|
|
"num_tokens": 239451.0,
|
|
"step": 335
|
|
},
|
|
{
|
|
"entropy": 7.28514552116394,
|
|
"epoch": 0.007841870978158083,
|
|
"grad_norm": 1.5078125,
|
|
"learning_rate": 0.00016950000000000003,
|
|
"loss": 7.0215,
|
|
"mean_token_accuracy": 0.1078951895236969,
|
|
"num_tokens": 243193.0,
|
|
"step": 340
|
|
},
|
|
{
|
|
"entropy": 7.671026420593262,
|
|
"epoch": 0.007957192610189819,
|
|
"grad_norm": 1.7578125,
|
|
"learning_rate": 0.00017199999999999998,
|
|
"loss": 7.0501,
|
|
"mean_token_accuracy": 0.11457417979836464,
|
|
"num_tokens": 246859.0,
|
|
"step": 345
|
|
},
|
|
{
|
|
"entropy": 7.1775932788848875,
|
|
"epoch": 0.008072514242221556,
|
|
"grad_norm": 1.8359375,
|
|
"learning_rate": 0.00017449999999999999,
|
|
"loss": 6.9368,
|
|
"mean_token_accuracy": 0.11339911893010139,
|
|
"num_tokens": 250345.0,
|
|
"step": 350
|
|
},
|
|
{
|
|
"entropy": 7.443285751342773,
|
|
"epoch": 0.008187835874253292,
|
|
"grad_norm": 1.78125,
|
|
"learning_rate": 0.000177,
|
|
"loss": 6.9842,
|
|
"mean_token_accuracy": 0.10223566293716431,
|
|
"num_tokens": 254201.0,
|
|
"step": 355
|
|
},
|
|
{
|
|
"entropy": 7.363960456848145,
|
|
"epoch": 0.00830315750628503,
|
|
"grad_norm": 1.640625,
|
|
"learning_rate": 0.0001795,
|
|
"loss": 6.9875,
|
|
"mean_token_accuracy": 0.11335918605327606,
|
|
"num_tokens": 258073.0,
|
|
"step": 360
|
|
},
|
|
{
|
|
"entropy": 7.3993799686431885,
|
|
"epoch": 0.008418479138316766,
|
|
"grad_norm": 1.484375,
|
|
"learning_rate": 0.000182,
|
|
"loss": 6.9985,
|
|
"mean_token_accuracy": 0.1107051931321621,
|
|
"num_tokens": 262221.0,
|
|
"step": 365
|
|
},
|
|
{
|
|
"entropy": 7.2681444644927975,
|
|
"epoch": 0.008533800770348501,
|
|
"grad_norm": 1.609375,
|
|
"learning_rate": 0.0001845,
|
|
"loss": 6.8422,
|
|
"mean_token_accuracy": 0.11727887839078903,
|
|
"num_tokens": 265920.0,
|
|
"step": 370
|
|
},
|
|
{
|
|
"entropy": 7.339032983779907,
|
|
"epoch": 0.008649122402380239,
|
|
"grad_norm": 1.5625,
|
|
"learning_rate": 0.000187,
|
|
"loss": 6.8632,
|
|
"mean_token_accuracy": 0.1098533809185028,
|
|
"num_tokens": 269617.0,
|
|
"step": 375
|
|
},
|
|
{
|
|
"entropy": 7.235262298583985,
|
|
"epoch": 0.008764444034411975,
|
|
"grad_norm": 1.5859375,
|
|
"learning_rate": 0.0001895,
|
|
"loss": 7.0343,
|
|
"mean_token_accuracy": 0.11555488631129265,
|
|
"num_tokens": 273482.0,
|
|
"step": 380
|
|
},
|
|
{
|
|
"entropy": 7.33788275718689,
|
|
"epoch": 0.008879765666443712,
|
|
"grad_norm": 1.8515625,
|
|
"learning_rate": 0.000192,
|
|
"loss": 6.8574,
|
|
"mean_token_accuracy": 0.12360220775008202,
|
|
"num_tokens": 276877.0,
|
|
"step": 385
|
|
},
|
|
{
|
|
"entropy": 7.284343814849853,
|
|
"epoch": 0.008995087298475448,
|
|
"grad_norm": 1.78125,
|
|
"learning_rate": 0.0001945,
|
|
"loss": 6.8065,
|
|
"mean_token_accuracy": 0.1250680521130562,
|
|
"num_tokens": 280342.0,
|
|
"step": 390
|
|
},
|
|
{
|
|
"entropy": 7.164630126953125,
|
|
"epoch": 0.009110408930507184,
|
|
"grad_norm": 1.6484375,
|
|
"learning_rate": 0.00019700000000000002,
|
|
"loss": 6.8672,
|
|
"mean_token_accuracy": 0.11904213950037956,
|
|
"num_tokens": 283740.0,
|
|
"step": 395
|
|
},
|
|
{
|
|
"entropy": 7.220073223114014,
|
|
"epoch": 0.009225730562538921,
|
|
"grad_norm": 1.828125,
|
|
"learning_rate": 0.00019950000000000002,
|
|
"loss": 6.8448,
|
|
"mean_token_accuracy": 0.114653430134058,
|
|
"num_tokens": 287534.0,
|
|
"step": 400
|
|
},
|
|
{
|
|
"entropy": 7.279120349884034,
|
|
"epoch": 0.009341052194570657,
|
|
"grad_norm": 1.75,
|
|
"learning_rate": 0.000202,
|
|
"loss": 6.9441,
|
|
"mean_token_accuracy": 0.11235265359282494,
|
|
"num_tokens": 290847.0,
|
|
"step": 405
|
|
},
|
|
{
|
|
"entropy": 7.284884881973267,
|
|
"epoch": 0.009456373826602395,
|
|
"grad_norm": 1.8125,
|
|
"learning_rate": 0.00020449999999999998,
|
|
"loss": 6.8057,
|
|
"mean_token_accuracy": 0.11482253968715668,
|
|
"num_tokens": 294317.0,
|
|
"step": 410
|
|
},
|
|
{
|
|
"entropy": 7.207005405426026,
|
|
"epoch": 0.00957169545863413,
|
|
"grad_norm": 1.734375,
|
|
"learning_rate": 0.000207,
|
|
"loss": 6.7533,
|
|
"mean_token_accuracy": 0.11797683015465736,
|
|
"num_tokens": 297899.0,
|
|
"step": 415
|
|
},
|
|
{
|
|
"entropy": 7.00657844543457,
|
|
"epoch": 0.009687017090665866,
|
|
"grad_norm": 1.5,
|
|
"learning_rate": 0.0002095,
|
|
"loss": 6.789,
|
|
"mean_token_accuracy": 0.1228546530008316,
|
|
"num_tokens": 301076.0,
|
|
"step": 420
|
|
},
|
|
{
|
|
"entropy": 7.3213354587554935,
|
|
"epoch": 0.009802338722697604,
|
|
"grad_norm": 1.625,
|
|
"learning_rate": 0.000212,
|
|
"loss": 6.7636,
|
|
"mean_token_accuracy": 0.12574946507811546,
|
|
"num_tokens": 304848.0,
|
|
"step": 425
|
|
},
|
|
{
|
|
"entropy": 7.14543981552124,
|
|
"epoch": 0.00991766035472934,
|
|
"grad_norm": 1.703125,
|
|
"learning_rate": 0.0002145,
|
|
"loss": 6.8416,
|
|
"mean_token_accuracy": 0.11870098114013672,
|
|
"num_tokens": 308223.0,
|
|
"step": 430
|
|
},
|
|
{
|
|
"entropy": 7.262237453460694,
|
|
"epoch": 0.010032981986761077,
|
|
"grad_norm": 1.578125,
|
|
"learning_rate": 0.00021700000000000002,
|
|
"loss": 6.9775,
|
|
"mean_token_accuracy": 0.11694321408867836,
|
|
"num_tokens": 311755.0,
|
|
"step": 435
|
|
},
|
|
{
|
|
"entropy": 7.160612010955811,
|
|
"epoch": 0.010148303618792813,
|
|
"grad_norm": 1.4609375,
|
|
"learning_rate": 0.0002195,
|
|
"loss": 6.8402,
|
|
"mean_token_accuracy": 0.12193377017974853,
|
|
"num_tokens": 315362.0,
|
|
"step": 440
|
|
},
|
|
{
|
|
"entropy": 7.3824732303619385,
|
|
"epoch": 0.010263625250824549,
|
|
"grad_norm": 1.40625,
|
|
"learning_rate": 0.000222,
|
|
"loss": 7.0482,
|
|
"mean_token_accuracy": 0.10882357284426689,
|
|
"num_tokens": 319705.0,
|
|
"step": 445
|
|
},
|
|
{
|
|
"entropy": 7.185869121551514,
|
|
"epoch": 0.010378946882856286,
|
|
"grad_norm": 1.6875,
|
|
"learning_rate": 0.0002245,
|
|
"loss": 6.8825,
|
|
"mean_token_accuracy": 0.11569961458444596,
|
|
"num_tokens": 323371.0,
|
|
"step": 450
|
|
},
|
|
{
|
|
"entropy": 7.384743690490723,
|
|
"epoch": 0.010494268514888022,
|
|
"grad_norm": 1.578125,
|
|
"learning_rate": 0.00022700000000000002,
|
|
"loss": 6.981,
|
|
"mean_token_accuracy": 0.11269342377781869,
|
|
"num_tokens": 326846.0,
|
|
"step": 455
|
|
},
|
|
{
|
|
"entropy": 7.022752046585083,
|
|
"epoch": 0.01060959014691976,
|
|
"grad_norm": 1.5,
|
|
"learning_rate": 0.00022950000000000002,
|
|
"loss": 6.7201,
|
|
"mean_token_accuracy": 0.13362340852618218,
|
|
"num_tokens": 330305.0,
|
|
"step": 460
|
|
},
|
|
{
|
|
"entropy": 7.039274597167969,
|
|
"epoch": 0.010724911778951495,
|
|
"grad_norm": 1.6796875,
|
|
"learning_rate": 0.00023200000000000003,
|
|
"loss": 6.7251,
|
|
"mean_token_accuracy": 0.1254519484937191,
|
|
"num_tokens": 333532.0,
|
|
"step": 465
|
|
},
|
|
{
|
|
"entropy": 7.0497644424438475,
|
|
"epoch": 0.010840233410983233,
|
|
"grad_norm": 1.5625,
|
|
"learning_rate": 0.00023449999999999998,
|
|
"loss": 6.7665,
|
|
"mean_token_accuracy": 0.12782322838902474,
|
|
"num_tokens": 336892.0,
|
|
"step": 470
|
|
},
|
|
{
|
|
"entropy": 7.166635799407959,
|
|
"epoch": 0.010955555043014969,
|
|
"grad_norm": 1.5625,
|
|
"learning_rate": 0.000237,
|
|
"loss": 6.7096,
|
|
"mean_token_accuracy": 0.1333428256213665,
|
|
"num_tokens": 340372.0,
|
|
"step": 475
|
|
},
|
|
{
|
|
"entropy": 7.2982118129730225,
|
|
"epoch": 0.011070876675046705,
|
|
"grad_norm": 1.75,
|
|
"learning_rate": 0.0002395,
|
|
"loss": 6.9193,
|
|
"mean_token_accuracy": 0.1203345313668251,
|
|
"num_tokens": 344204.0,
|
|
"step": 480
|
|
},
|
|
{
|
|
"entropy": 6.994412517547607,
|
|
"epoch": 0.011186198307078442,
|
|
"grad_norm": 1.6484375,
|
|
"learning_rate": 0.000242,
|
|
"loss": 6.7874,
|
|
"mean_token_accuracy": 0.1303581975400448,
|
|
"num_tokens": 347962.0,
|
|
"step": 485
|
|
},
|
|
{
|
|
"entropy": 7.16865119934082,
|
|
"epoch": 0.011301519939110178,
|
|
"grad_norm": 1.59375,
|
|
"learning_rate": 0.0002445,
|
|
"loss": 6.7901,
|
|
"mean_token_accuracy": 0.12705896645784379,
|
|
"num_tokens": 351559.0,
|
|
"step": 490
|
|
},
|
|
{
|
|
"entropy": 7.113827466964722,
|
|
"epoch": 0.011416841571141916,
|
|
"grad_norm": 1.6796875,
|
|
"learning_rate": 0.000247,
|
|
"loss": 6.8509,
|
|
"mean_token_accuracy": 0.11782361045479775,
|
|
"num_tokens": 354878.0,
|
|
"step": 495
|
|
},
|
|
{
|
|
"entropy": 7.1837897300720215,
|
|
"epoch": 0.011532163203173651,
|
|
"grad_norm": 1.4609375,
|
|
"learning_rate": 0.0002495,
|
|
"loss": 6.8408,
|
|
"mean_token_accuracy": 0.11947782784700393,
|
|
"num_tokens": 358998.0,
|
|
"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": 559663036416000.0,
|
|
"train_batch_size": 16,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|