1035 lines
27 KiB
JSON
1035 lines
27 KiB
JSON
{
|
|
"best_global_step": null,
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 0.03026359592046727,
|
|
"eval_steps": 500,
|
|
"global_step": 500,
|
|
"is_hyper_param_search": false,
|
|
"is_local_process_zero": true,
|
|
"is_world_process_zero": true,
|
|
"log_history": [
|
|
{
|
|
"entropy": 10.74259672164917,
|
|
"epoch": 0.0003026359592046727,
|
|
"grad_norm": 5.3125,
|
|
"learning_rate": 2e-06,
|
|
"loss": 10.7761,
|
|
"mean_token_accuracy": 0.00015748031437397004,
|
|
"num_tokens": 10868.0,
|
|
"step": 5
|
|
},
|
|
{
|
|
"entropy": 10.74251298904419,
|
|
"epoch": 0.0006052719184093454,
|
|
"grad_norm": 4.65625,
|
|
"learning_rate": 4.5e-06,
|
|
"loss": 10.7789,
|
|
"mean_token_accuracy": 0.0,
|
|
"num_tokens": 22589.0,
|
|
"step": 10
|
|
},
|
|
{
|
|
"entropy": 10.74259262084961,
|
|
"epoch": 0.0009079078776140181,
|
|
"grad_norm": 5.0625,
|
|
"learning_rate": 7e-06,
|
|
"loss": 10.7312,
|
|
"mean_token_accuracy": 0.0,
|
|
"num_tokens": 32471.0,
|
|
"step": 15
|
|
},
|
|
{
|
|
"entropy": 10.742576599121094,
|
|
"epoch": 0.0012105438368186909,
|
|
"grad_norm": 4.90625,
|
|
"learning_rate": 9.5e-06,
|
|
"loss": 10.6762,
|
|
"mean_token_accuracy": 0.0004521072842180729,
|
|
"num_tokens": 42231.0,
|
|
"step": 20
|
|
},
|
|
{
|
|
"entropy": 10.742555141448975,
|
|
"epoch": 0.0015131797960233635,
|
|
"grad_norm": 4.71875,
|
|
"learning_rate": 1.2e-05,
|
|
"loss": 10.5709,
|
|
"mean_token_accuracy": 0.0037700447253882887,
|
|
"num_tokens": 52966.0,
|
|
"step": 25
|
|
},
|
|
{
|
|
"entropy": 10.742278385162354,
|
|
"epoch": 0.0018158157552280363,
|
|
"grad_norm": 4.1875,
|
|
"learning_rate": 1.4500000000000002e-05,
|
|
"loss": 10.4476,
|
|
"mean_token_accuracy": 0.03040276560932398,
|
|
"num_tokens": 63711.0,
|
|
"step": 30
|
|
},
|
|
{
|
|
"entropy": 10.741186809539794,
|
|
"epoch": 0.002118451714432709,
|
|
"grad_norm": 3.09375,
|
|
"learning_rate": 1.7000000000000003e-05,
|
|
"loss": 10.3036,
|
|
"mean_token_accuracy": 0.040517190098762514,
|
|
"num_tokens": 74046.0,
|
|
"step": 35
|
|
},
|
|
{
|
|
"entropy": 10.73850612640381,
|
|
"epoch": 0.0024210876736373817,
|
|
"grad_norm": 2.75,
|
|
"learning_rate": 1.95e-05,
|
|
"loss": 10.1339,
|
|
"mean_token_accuracy": 0.043018221482634546,
|
|
"num_tokens": 84275.0,
|
|
"step": 40
|
|
},
|
|
{
|
|
"entropy": 10.73492136001587,
|
|
"epoch": 0.0027237236328420543,
|
|
"grad_norm": 2.234375,
|
|
"learning_rate": 2.2e-05,
|
|
"loss": 10.0376,
|
|
"mean_token_accuracy": 0.04388587549328804,
|
|
"num_tokens": 93302.0,
|
|
"step": 45
|
|
},
|
|
{
|
|
"entropy": 10.731941604614258,
|
|
"epoch": 0.003026359592046727,
|
|
"grad_norm": 2.21875,
|
|
"learning_rate": 2.4500000000000003e-05,
|
|
"loss": 9.9644,
|
|
"mean_token_accuracy": 0.041650505177676676,
|
|
"num_tokens": 103414.0,
|
|
"step": 50
|
|
},
|
|
{
|
|
"entropy": 10.730784225463868,
|
|
"epoch": 0.0033289955512513995,
|
|
"grad_norm": 1.9921875,
|
|
"learning_rate": 2.7e-05,
|
|
"loss": 9.8956,
|
|
"mean_token_accuracy": 0.04557133205235005,
|
|
"num_tokens": 113749.0,
|
|
"step": 55
|
|
},
|
|
{
|
|
"entropy": 10.728888988494873,
|
|
"epoch": 0.0036316315104560726,
|
|
"grad_norm": 1.9765625,
|
|
"learning_rate": 2.95e-05,
|
|
"loss": 9.8525,
|
|
"mean_token_accuracy": 0.04352857358753681,
|
|
"num_tokens": 123922.0,
|
|
"step": 60
|
|
},
|
|
{
|
|
"entropy": 10.72574872970581,
|
|
"epoch": 0.003934267469660745,
|
|
"grad_norm": 2.078125,
|
|
"learning_rate": 3.2e-05,
|
|
"loss": 9.8144,
|
|
"mean_token_accuracy": 0.04175829254090786,
|
|
"num_tokens": 134720.0,
|
|
"step": 65
|
|
},
|
|
{
|
|
"entropy": 10.723058605194092,
|
|
"epoch": 0.004236903428865418,
|
|
"grad_norm": 2.015625,
|
|
"learning_rate": 3.4500000000000005e-05,
|
|
"loss": 9.7116,
|
|
"mean_token_accuracy": 0.04281677231192589,
|
|
"num_tokens": 145025.0,
|
|
"step": 70
|
|
},
|
|
{
|
|
"entropy": 10.72093152999878,
|
|
"epoch": 0.00453953938807009,
|
|
"grad_norm": 2.078125,
|
|
"learning_rate": 3.7e-05,
|
|
"loss": 9.6537,
|
|
"mean_token_accuracy": 0.04347668327391148,
|
|
"num_tokens": 155182.0,
|
|
"step": 75
|
|
},
|
|
{
|
|
"entropy": 10.716725921630859,
|
|
"epoch": 0.004842175347274763,
|
|
"grad_norm": 1.9140625,
|
|
"learning_rate": 3.95e-05,
|
|
"loss": 9.6032,
|
|
"mean_token_accuracy": 0.04583056308329105,
|
|
"num_tokens": 166231.0,
|
|
"step": 80
|
|
},
|
|
{
|
|
"entropy": 10.70965871810913,
|
|
"epoch": 0.005144811306479436,
|
|
"grad_norm": 1.9609375,
|
|
"learning_rate": 4.2000000000000004e-05,
|
|
"loss": 9.4954,
|
|
"mean_token_accuracy": 0.04315895065665245,
|
|
"num_tokens": 175799.0,
|
|
"step": 85
|
|
},
|
|
{
|
|
"entropy": 10.702201461791992,
|
|
"epoch": 0.005447447265684109,
|
|
"grad_norm": 2.0,
|
|
"learning_rate": 4.45e-05,
|
|
"loss": 9.4271,
|
|
"mean_token_accuracy": 0.04461735654622316,
|
|
"num_tokens": 185455.0,
|
|
"step": 90
|
|
},
|
|
{
|
|
"entropy": 10.692846012115478,
|
|
"epoch": 0.005750083224888782,
|
|
"grad_norm": 1.796875,
|
|
"learning_rate": 4.7000000000000004e-05,
|
|
"loss": 9.4192,
|
|
"mean_token_accuracy": 0.04293463826179504,
|
|
"num_tokens": 196210.0,
|
|
"step": 95
|
|
},
|
|
{
|
|
"entropy": 10.680758380889893,
|
|
"epoch": 0.006052719184093454,
|
|
"grad_norm": 1.7890625,
|
|
"learning_rate": 4.9500000000000004e-05,
|
|
"loss": 9.3191,
|
|
"mean_token_accuracy": 0.047341904044151305,
|
|
"num_tokens": 206618.0,
|
|
"step": 100
|
|
},
|
|
{
|
|
"entropy": 10.663753604888916,
|
|
"epoch": 0.006355355143298127,
|
|
"grad_norm": 1.890625,
|
|
"learning_rate": 5.2e-05,
|
|
"loss": 9.1872,
|
|
"mean_token_accuracy": 0.05284549556672573,
|
|
"num_tokens": 217020.0,
|
|
"step": 105
|
|
},
|
|
{
|
|
"entropy": 10.640412425994873,
|
|
"epoch": 0.006657991102502799,
|
|
"grad_norm": 1.78125,
|
|
"learning_rate": 5.45e-05,
|
|
"loss": 9.06,
|
|
"mean_token_accuracy": 0.057008244842290876,
|
|
"num_tokens": 227195.0,
|
|
"step": 110
|
|
},
|
|
{
|
|
"entropy": 10.60520315170288,
|
|
"epoch": 0.006960627061707472,
|
|
"grad_norm": 1.8515625,
|
|
"learning_rate": 5.7e-05,
|
|
"loss": 8.9862,
|
|
"mean_token_accuracy": 0.057433782145380974,
|
|
"num_tokens": 238473.0,
|
|
"step": 115
|
|
},
|
|
{
|
|
"entropy": 10.55847454071045,
|
|
"epoch": 0.007263263020912145,
|
|
"grad_norm": 1.78125,
|
|
"learning_rate": 5.9499999999999996e-05,
|
|
"loss": 8.8089,
|
|
"mean_token_accuracy": 0.06153528727591038,
|
|
"num_tokens": 247778.0,
|
|
"step": 120
|
|
},
|
|
{
|
|
"entropy": 10.495486640930176,
|
|
"epoch": 0.007565898980116817,
|
|
"grad_norm": 1.671875,
|
|
"learning_rate": 6.2e-05,
|
|
"loss": 8.7828,
|
|
"mean_token_accuracy": 0.06823135390877724,
|
|
"num_tokens": 258424.0,
|
|
"step": 125
|
|
},
|
|
{
|
|
"entropy": 10.432032680511474,
|
|
"epoch": 0.00786853493932149,
|
|
"grad_norm": 1.7578125,
|
|
"learning_rate": 6.450000000000001e-05,
|
|
"loss": 8.6115,
|
|
"mean_token_accuracy": 0.06918121092021465,
|
|
"num_tokens": 267715.0,
|
|
"step": 130
|
|
},
|
|
{
|
|
"entropy": 10.359144115447998,
|
|
"epoch": 0.008171170898526163,
|
|
"grad_norm": 1.5859375,
|
|
"learning_rate": 6.7e-05,
|
|
"loss": 8.4638,
|
|
"mean_token_accuracy": 0.07366116344928741,
|
|
"num_tokens": 277127.0,
|
|
"step": 135
|
|
},
|
|
{
|
|
"entropy": 10.267242240905762,
|
|
"epoch": 0.008473806857730836,
|
|
"grad_norm": 1.5546875,
|
|
"learning_rate": 6.950000000000001e-05,
|
|
"loss": 8.3343,
|
|
"mean_token_accuracy": 0.0729695912450552,
|
|
"num_tokens": 287275.0,
|
|
"step": 140
|
|
},
|
|
{
|
|
"entropy": 10.183104038238525,
|
|
"epoch": 0.008776442816935508,
|
|
"grad_norm": 1.4296875,
|
|
"learning_rate": 7.2e-05,
|
|
"loss": 8.2075,
|
|
"mean_token_accuracy": 0.07483375370502472,
|
|
"num_tokens": 297131.0,
|
|
"step": 145
|
|
},
|
|
{
|
|
"entropy": 10.032635116577149,
|
|
"epoch": 0.00907907877614018,
|
|
"grad_norm": 1.4609375,
|
|
"learning_rate": 7.45e-05,
|
|
"loss": 8.0837,
|
|
"mean_token_accuracy": 0.0766814447939396,
|
|
"num_tokens": 307649.0,
|
|
"step": 150
|
|
},
|
|
{
|
|
"entropy": 9.881086444854736,
|
|
"epoch": 0.009381714735344854,
|
|
"grad_norm": 1.3359375,
|
|
"learning_rate": 7.7e-05,
|
|
"loss": 7.9254,
|
|
"mean_token_accuracy": 0.0791646283119917,
|
|
"num_tokens": 317655.0,
|
|
"step": 155
|
|
},
|
|
{
|
|
"entropy": 9.67042417526245,
|
|
"epoch": 0.009684350694549527,
|
|
"grad_norm": 1.15625,
|
|
"learning_rate": 7.950000000000001e-05,
|
|
"loss": 7.9643,
|
|
"mean_token_accuracy": 0.07545723952353,
|
|
"num_tokens": 329286.0,
|
|
"step": 160
|
|
},
|
|
{
|
|
"entropy": 9.513041305541993,
|
|
"epoch": 0.0099869866537542,
|
|
"grad_norm": 1.265625,
|
|
"learning_rate": 8.2e-05,
|
|
"loss": 7.7586,
|
|
"mean_token_accuracy": 0.0771209493279457,
|
|
"num_tokens": 339212.0,
|
|
"step": 165
|
|
},
|
|
{
|
|
"entropy": 9.245701503753661,
|
|
"epoch": 0.010289622612958871,
|
|
"grad_norm": 1.1171875,
|
|
"learning_rate": 8.450000000000001e-05,
|
|
"loss": 7.7092,
|
|
"mean_token_accuracy": 0.07468102127313614,
|
|
"num_tokens": 349205.0,
|
|
"step": 170
|
|
},
|
|
{
|
|
"entropy": 9.026909351348877,
|
|
"epoch": 0.010592258572163544,
|
|
"grad_norm": 0.9765625,
|
|
"learning_rate": 8.7e-05,
|
|
"loss": 7.6824,
|
|
"mean_token_accuracy": 0.08052874132990837,
|
|
"num_tokens": 360548.0,
|
|
"step": 175
|
|
},
|
|
{
|
|
"entropy": 8.78289623260498,
|
|
"epoch": 0.010894894531368217,
|
|
"grad_norm": 0.890625,
|
|
"learning_rate": 8.95e-05,
|
|
"loss": 7.5607,
|
|
"mean_token_accuracy": 0.07696975283324718,
|
|
"num_tokens": 371049.0,
|
|
"step": 180
|
|
},
|
|
{
|
|
"entropy": 8.596883869171142,
|
|
"epoch": 0.01119753049057289,
|
|
"grad_norm": 0.7890625,
|
|
"learning_rate": 9.2e-05,
|
|
"loss": 7.6138,
|
|
"mean_token_accuracy": 0.07861833088099957,
|
|
"num_tokens": 382334.0,
|
|
"step": 185
|
|
},
|
|
{
|
|
"entropy": 8.436706256866454,
|
|
"epoch": 0.011500166449777563,
|
|
"grad_norm": 0.765625,
|
|
"learning_rate": 9.45e-05,
|
|
"loss": 7.5283,
|
|
"mean_token_accuracy": 0.07383836694061756,
|
|
"num_tokens": 393169.0,
|
|
"step": 190
|
|
},
|
|
{
|
|
"entropy": 8.288833332061767,
|
|
"epoch": 0.011802802408982235,
|
|
"grad_norm": 0.9140625,
|
|
"learning_rate": 9.7e-05,
|
|
"loss": 7.4185,
|
|
"mean_token_accuracy": 0.0791880264878273,
|
|
"num_tokens": 403826.0,
|
|
"step": 195
|
|
},
|
|
{
|
|
"entropy": 8.156514549255371,
|
|
"epoch": 0.012105438368186908,
|
|
"grad_norm": 1.0390625,
|
|
"learning_rate": 9.95e-05,
|
|
"loss": 7.3629,
|
|
"mean_token_accuracy": 0.08756700083613396,
|
|
"num_tokens": 414231.0,
|
|
"step": 200
|
|
},
|
|
{
|
|
"entropy": 8.109231567382812,
|
|
"epoch": 0.01240807432739158,
|
|
"grad_norm": 0.87890625,
|
|
"learning_rate": 0.000102,
|
|
"loss": 7.3613,
|
|
"mean_token_accuracy": 0.07943096160888671,
|
|
"num_tokens": 424532.0,
|
|
"step": 205
|
|
},
|
|
{
|
|
"entropy": 8.120400428771973,
|
|
"epoch": 0.012710710286596254,
|
|
"grad_norm": 0.96875,
|
|
"learning_rate": 0.00010449999999999999,
|
|
"loss": 7.4376,
|
|
"mean_token_accuracy": 0.0809888195246458,
|
|
"num_tokens": 435002.0,
|
|
"step": 210
|
|
},
|
|
{
|
|
"entropy": 8.078549480438232,
|
|
"epoch": 0.013013346245800927,
|
|
"grad_norm": 1.0546875,
|
|
"learning_rate": 0.000107,
|
|
"loss": 7.3373,
|
|
"mean_token_accuracy": 0.088172397762537,
|
|
"num_tokens": 444954.0,
|
|
"step": 215
|
|
},
|
|
{
|
|
"entropy": 8.021297836303711,
|
|
"epoch": 0.013315982205005598,
|
|
"grad_norm": 0.87890625,
|
|
"learning_rate": 0.0001095,
|
|
"loss": 7.3151,
|
|
"mean_token_accuracy": 0.08277993872761727,
|
|
"num_tokens": 455078.0,
|
|
"step": 220
|
|
},
|
|
{
|
|
"entropy": 7.960602712631226,
|
|
"epoch": 0.013618618164210271,
|
|
"grad_norm": 0.90625,
|
|
"learning_rate": 0.000112,
|
|
"loss": 7.3156,
|
|
"mean_token_accuracy": 0.08474647626280785,
|
|
"num_tokens": 464688.0,
|
|
"step": 225
|
|
},
|
|
{
|
|
"entropy": 7.901602554321289,
|
|
"epoch": 0.013921254123414944,
|
|
"grad_norm": 0.9296875,
|
|
"learning_rate": 0.0001145,
|
|
"loss": 7.2243,
|
|
"mean_token_accuracy": 0.08522045239806175,
|
|
"num_tokens": 475365.0,
|
|
"step": 230
|
|
},
|
|
{
|
|
"entropy": 7.995245122909546,
|
|
"epoch": 0.014223890082619617,
|
|
"grad_norm": 1.546875,
|
|
"learning_rate": 0.00011700000000000001,
|
|
"loss": 7.3312,
|
|
"mean_token_accuracy": 0.08323194459080696,
|
|
"num_tokens": 487238.0,
|
|
"step": 235
|
|
},
|
|
{
|
|
"entropy": 7.879671859741211,
|
|
"epoch": 0.01452652604182429,
|
|
"grad_norm": 1.328125,
|
|
"learning_rate": 0.00011949999999999999,
|
|
"loss": 7.2215,
|
|
"mean_token_accuracy": 0.08390360698103905,
|
|
"num_tokens": 497485.0,
|
|
"step": 240
|
|
},
|
|
{
|
|
"entropy": 7.803989362716675,
|
|
"epoch": 0.014829162001028962,
|
|
"grad_norm": 1.109375,
|
|
"learning_rate": 0.000122,
|
|
"loss": 7.2663,
|
|
"mean_token_accuracy": 0.09161997810006142,
|
|
"num_tokens": 508228.0,
|
|
"step": 245
|
|
},
|
|
{
|
|
"entropy": 7.834836626052857,
|
|
"epoch": 0.015131797960233635,
|
|
"grad_norm": 1.2421875,
|
|
"learning_rate": 0.0001245,
|
|
"loss": 7.1851,
|
|
"mean_token_accuracy": 0.08978703171014786,
|
|
"num_tokens": 518645.0,
|
|
"step": 250
|
|
},
|
|
{
|
|
"entropy": 7.774400329589843,
|
|
"epoch": 0.015434433919438308,
|
|
"grad_norm": 1.1484375,
|
|
"learning_rate": 0.000127,
|
|
"loss": 7.1447,
|
|
"mean_token_accuracy": 0.09164771437644958,
|
|
"num_tokens": 528151.0,
|
|
"step": 255
|
|
},
|
|
{
|
|
"entropy": 7.811225557327271,
|
|
"epoch": 0.01573706987864298,
|
|
"grad_norm": 0.94921875,
|
|
"learning_rate": 0.0001295,
|
|
"loss": 7.2129,
|
|
"mean_token_accuracy": 0.08425109386444092,
|
|
"num_tokens": 539160.0,
|
|
"step": 260
|
|
},
|
|
{
|
|
"entropy": 7.784347248077393,
|
|
"epoch": 0.016039705837847654,
|
|
"grad_norm": 0.94140625,
|
|
"learning_rate": 0.000132,
|
|
"loss": 7.2983,
|
|
"mean_token_accuracy": 0.08336320072412491,
|
|
"num_tokens": 550053.0,
|
|
"step": 265
|
|
},
|
|
{
|
|
"entropy": 7.730723190307617,
|
|
"epoch": 0.016342341797052327,
|
|
"grad_norm": 1.0546875,
|
|
"learning_rate": 0.00013450000000000002,
|
|
"loss": 7.0826,
|
|
"mean_token_accuracy": 0.09093882888555527,
|
|
"num_tokens": 560229.0,
|
|
"step": 270
|
|
},
|
|
{
|
|
"entropy": 7.7058757781982425,
|
|
"epoch": 0.016644977756257,
|
|
"grad_norm": 0.8828125,
|
|
"learning_rate": 0.00013700000000000002,
|
|
"loss": 7.1653,
|
|
"mean_token_accuracy": 0.08866406977176666,
|
|
"num_tokens": 570630.0,
|
|
"step": 275
|
|
},
|
|
{
|
|
"entropy": 7.813187074661255,
|
|
"epoch": 0.016947613715461673,
|
|
"grad_norm": 0.9140625,
|
|
"learning_rate": 0.0001395,
|
|
"loss": 7.2263,
|
|
"mean_token_accuracy": 0.08829710632562637,
|
|
"num_tokens": 581208.0,
|
|
"step": 280
|
|
},
|
|
{
|
|
"entropy": 7.697104406356812,
|
|
"epoch": 0.017250249674666342,
|
|
"grad_norm": 0.90625,
|
|
"learning_rate": 0.00014199999999999998,
|
|
"loss": 7.0955,
|
|
"mean_token_accuracy": 0.09898408353328705,
|
|
"num_tokens": 591352.0,
|
|
"step": 285
|
|
},
|
|
{
|
|
"entropy": 7.6714756965637205,
|
|
"epoch": 0.017552885633871015,
|
|
"grad_norm": 0.94921875,
|
|
"learning_rate": 0.0001445,
|
|
"loss": 7.1736,
|
|
"mean_token_accuracy": 0.09172456562519074,
|
|
"num_tokens": 600670.0,
|
|
"step": 290
|
|
},
|
|
{
|
|
"entropy": 7.675610876083374,
|
|
"epoch": 0.01785552159307569,
|
|
"grad_norm": 1.046875,
|
|
"learning_rate": 0.000147,
|
|
"loss": 7.1785,
|
|
"mean_token_accuracy": 0.09298023506999016,
|
|
"num_tokens": 610625.0,
|
|
"step": 295
|
|
},
|
|
{
|
|
"entropy": 7.745320653915405,
|
|
"epoch": 0.01815815755228036,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 0.0001495,
|
|
"loss": 7.1448,
|
|
"mean_token_accuracy": 0.09545003101229668,
|
|
"num_tokens": 621096.0,
|
|
"step": 300
|
|
},
|
|
{
|
|
"entropy": 7.587613916397094,
|
|
"epoch": 0.018460793511485035,
|
|
"grad_norm": 0.88671875,
|
|
"learning_rate": 0.000152,
|
|
"loss": 7.1062,
|
|
"mean_token_accuracy": 0.09240609779953957,
|
|
"num_tokens": 632168.0,
|
|
"step": 305
|
|
},
|
|
{
|
|
"entropy": 7.6709612846374515,
|
|
"epoch": 0.018763429470689708,
|
|
"grad_norm": 0.90234375,
|
|
"learning_rate": 0.00015450000000000001,
|
|
"loss": 7.2109,
|
|
"mean_token_accuracy": 0.09153169840574264,
|
|
"num_tokens": 642277.0,
|
|
"step": 310
|
|
},
|
|
{
|
|
"entropy": 7.587620401382447,
|
|
"epoch": 0.01906606542989438,
|
|
"grad_norm": 1.03125,
|
|
"learning_rate": 0.000157,
|
|
"loss": 7.0974,
|
|
"mean_token_accuracy": 0.09367592856287957,
|
|
"num_tokens": 652429.0,
|
|
"step": 315
|
|
},
|
|
{
|
|
"entropy": 7.541977453231811,
|
|
"epoch": 0.019368701389099054,
|
|
"grad_norm": 1.0859375,
|
|
"learning_rate": 0.0001595,
|
|
"loss": 6.989,
|
|
"mean_token_accuracy": 0.09826880320906639,
|
|
"num_tokens": 662805.0,
|
|
"step": 320
|
|
},
|
|
{
|
|
"entropy": 7.580408334732056,
|
|
"epoch": 0.019671337348303727,
|
|
"grad_norm": 1.0546875,
|
|
"learning_rate": 0.000162,
|
|
"loss": 6.9876,
|
|
"mean_token_accuracy": 0.10082038789987564,
|
|
"num_tokens": 673015.0,
|
|
"step": 325
|
|
},
|
|
{
|
|
"entropy": 7.452456760406494,
|
|
"epoch": 0.0199739733075084,
|
|
"grad_norm": 1.046875,
|
|
"learning_rate": 0.00016450000000000001,
|
|
"loss": 6.9221,
|
|
"mean_token_accuracy": 0.09537633135914803,
|
|
"num_tokens": 683357.0,
|
|
"step": 330
|
|
},
|
|
{
|
|
"entropy": 7.544377756118775,
|
|
"epoch": 0.02027660926671307,
|
|
"grad_norm": 0.95703125,
|
|
"learning_rate": 0.00016700000000000002,
|
|
"loss": 7.0466,
|
|
"mean_token_accuracy": 0.09294784143567085,
|
|
"num_tokens": 693310.0,
|
|
"step": 335
|
|
},
|
|
{
|
|
"entropy": 7.5124753475189205,
|
|
"epoch": 0.020579245225917742,
|
|
"grad_norm": 0.94140625,
|
|
"learning_rate": 0.00016950000000000003,
|
|
"loss": 7.0218,
|
|
"mean_token_accuracy": 0.09687028974294662,
|
|
"num_tokens": 703189.0,
|
|
"step": 340
|
|
},
|
|
{
|
|
"entropy": 7.457939147949219,
|
|
"epoch": 0.020881881185122415,
|
|
"grad_norm": 1.015625,
|
|
"learning_rate": 0.00017199999999999998,
|
|
"loss": 6.9878,
|
|
"mean_token_accuracy": 0.09731099009513855,
|
|
"num_tokens": 714359.0,
|
|
"step": 345
|
|
},
|
|
{
|
|
"entropy": 7.501646852493286,
|
|
"epoch": 0.02118451714432709,
|
|
"grad_norm": 0.9296875,
|
|
"learning_rate": 0.00017449999999999999,
|
|
"loss": 7.0421,
|
|
"mean_token_accuracy": 0.0931004598736763,
|
|
"num_tokens": 724991.0,
|
|
"step": 350
|
|
},
|
|
{
|
|
"entropy": 7.500136518478394,
|
|
"epoch": 0.02148715310353176,
|
|
"grad_norm": 0.91796875,
|
|
"learning_rate": 0.000177,
|
|
"loss": 6.8986,
|
|
"mean_token_accuracy": 0.09156099259853363,
|
|
"num_tokens": 735879.0,
|
|
"step": 355
|
|
},
|
|
{
|
|
"entropy": 7.451840782165528,
|
|
"epoch": 0.021789789062736434,
|
|
"grad_norm": 1.1484375,
|
|
"learning_rate": 0.0001795,
|
|
"loss": 6.9577,
|
|
"mean_token_accuracy": 0.10013179183006286,
|
|
"num_tokens": 745887.0,
|
|
"step": 360
|
|
},
|
|
{
|
|
"entropy": 7.5548046112060545,
|
|
"epoch": 0.022092425021941108,
|
|
"grad_norm": 1.1875,
|
|
"learning_rate": 0.000182,
|
|
"loss": 6.9763,
|
|
"mean_token_accuracy": 0.1025528259575367,
|
|
"num_tokens": 755822.0,
|
|
"step": 365
|
|
},
|
|
{
|
|
"entropy": 7.366734838485717,
|
|
"epoch": 0.02239506098114578,
|
|
"grad_norm": 0.9921875,
|
|
"learning_rate": 0.0001845,
|
|
"loss": 6.9315,
|
|
"mean_token_accuracy": 0.10159048363566399,
|
|
"num_tokens": 765330.0,
|
|
"step": 370
|
|
},
|
|
{
|
|
"entropy": 7.413720750808716,
|
|
"epoch": 0.022697696940350454,
|
|
"grad_norm": 0.82421875,
|
|
"learning_rate": 0.000187,
|
|
"loss": 6.8919,
|
|
"mean_token_accuracy": 0.10146145895123482,
|
|
"num_tokens": 775577.0,
|
|
"step": 375
|
|
},
|
|
{
|
|
"entropy": 7.391197729110718,
|
|
"epoch": 0.023000332899555127,
|
|
"grad_norm": 1.046875,
|
|
"learning_rate": 0.0001895,
|
|
"loss": 6.9459,
|
|
"mean_token_accuracy": 0.0984920509159565,
|
|
"num_tokens": 784999.0,
|
|
"step": 380
|
|
},
|
|
{
|
|
"entropy": 7.430153036117554,
|
|
"epoch": 0.023302968858759796,
|
|
"grad_norm": 0.94921875,
|
|
"learning_rate": 0.000192,
|
|
"loss": 6.8889,
|
|
"mean_token_accuracy": 0.10047663822770118,
|
|
"num_tokens": 794917.0,
|
|
"step": 385
|
|
},
|
|
{
|
|
"entropy": 7.323439264297486,
|
|
"epoch": 0.02360560481796447,
|
|
"grad_norm": 1.0078125,
|
|
"learning_rate": 0.0001945,
|
|
"loss": 6.8756,
|
|
"mean_token_accuracy": 0.09725546389818192,
|
|
"num_tokens": 805772.0,
|
|
"step": 390
|
|
},
|
|
{
|
|
"entropy": 7.467493295669556,
|
|
"epoch": 0.023908240777169142,
|
|
"grad_norm": 1.09375,
|
|
"learning_rate": 0.00019700000000000002,
|
|
"loss": 7.0954,
|
|
"mean_token_accuracy": 0.0907193399965763,
|
|
"num_tokens": 816326.0,
|
|
"step": 395
|
|
},
|
|
{
|
|
"entropy": 7.409446811676025,
|
|
"epoch": 0.024210876736373815,
|
|
"grad_norm": 0.984375,
|
|
"learning_rate": 0.00019950000000000002,
|
|
"loss": 6.8854,
|
|
"mean_token_accuracy": 0.0965262807905674,
|
|
"num_tokens": 825723.0,
|
|
"step": 400
|
|
},
|
|
{
|
|
"entropy": 7.347925472259521,
|
|
"epoch": 0.02451351269557849,
|
|
"grad_norm": 1.125,
|
|
"learning_rate": 0.000202,
|
|
"loss": 6.8529,
|
|
"mean_token_accuracy": 0.10136102885007858,
|
|
"num_tokens": 834939.0,
|
|
"step": 405
|
|
},
|
|
{
|
|
"entropy": 7.287171125411987,
|
|
"epoch": 0.02481614865478316,
|
|
"grad_norm": 1.046875,
|
|
"learning_rate": 0.00020449999999999998,
|
|
"loss": 6.7984,
|
|
"mean_token_accuracy": 0.10428313314914703,
|
|
"num_tokens": 844093.0,
|
|
"step": 410
|
|
},
|
|
{
|
|
"entropy": 7.37103533744812,
|
|
"epoch": 0.025118784613987834,
|
|
"grad_norm": 1.0546875,
|
|
"learning_rate": 0.000207,
|
|
"loss": 6.8791,
|
|
"mean_token_accuracy": 0.09733516573905945,
|
|
"num_tokens": 853879.0,
|
|
"step": 415
|
|
},
|
|
{
|
|
"entropy": 7.328188419342041,
|
|
"epoch": 0.025421420573192507,
|
|
"grad_norm": 1.0390625,
|
|
"learning_rate": 0.0002095,
|
|
"loss": 6.8858,
|
|
"mean_token_accuracy": 0.09786264151334763,
|
|
"num_tokens": 864013.0,
|
|
"step": 420
|
|
},
|
|
{
|
|
"entropy": 7.331731557846069,
|
|
"epoch": 0.02572405653239718,
|
|
"grad_norm": 1.15625,
|
|
"learning_rate": 0.000212,
|
|
"loss": 6.8476,
|
|
"mean_token_accuracy": 0.10083968937397003,
|
|
"num_tokens": 874295.0,
|
|
"step": 425
|
|
},
|
|
{
|
|
"entropy": 7.2822911739349365,
|
|
"epoch": 0.026026692491601854,
|
|
"grad_norm": 1.0390625,
|
|
"learning_rate": 0.0002145,
|
|
"loss": 6.8964,
|
|
"mean_token_accuracy": 0.09731603786349297,
|
|
"num_tokens": 884748.0,
|
|
"step": 430
|
|
},
|
|
{
|
|
"entropy": 7.363057661056518,
|
|
"epoch": 0.026329328450806523,
|
|
"grad_norm": 1.0859375,
|
|
"learning_rate": 0.00021700000000000002,
|
|
"loss": 6.9178,
|
|
"mean_token_accuracy": 0.09977400824427604,
|
|
"num_tokens": 895809.0,
|
|
"step": 435
|
|
},
|
|
{
|
|
"entropy": 7.325506687164307,
|
|
"epoch": 0.026631964410011196,
|
|
"grad_norm": 1.015625,
|
|
"learning_rate": 0.0002195,
|
|
"loss": 6.8938,
|
|
"mean_token_accuracy": 0.10050061941146851,
|
|
"num_tokens": 905252.0,
|
|
"step": 440
|
|
},
|
|
{
|
|
"entropy": 7.357236623764038,
|
|
"epoch": 0.02693460036921587,
|
|
"grad_norm": 0.94921875,
|
|
"learning_rate": 0.000222,
|
|
"loss": 6.8265,
|
|
"mean_token_accuracy": 0.09752592518925667,
|
|
"num_tokens": 913876.0,
|
|
"step": 445
|
|
},
|
|
{
|
|
"entropy": 7.245391464233398,
|
|
"epoch": 0.027237236328420542,
|
|
"grad_norm": 1.046875,
|
|
"learning_rate": 0.0002245,
|
|
"loss": 6.8026,
|
|
"mean_token_accuracy": 0.09813540726900101,
|
|
"num_tokens": 924255.0,
|
|
"step": 450
|
|
},
|
|
{
|
|
"entropy": 7.230363607406616,
|
|
"epoch": 0.027539872287625215,
|
|
"grad_norm": 0.9375,
|
|
"learning_rate": 0.00022700000000000002,
|
|
"loss": 6.7995,
|
|
"mean_token_accuracy": 0.10496273562312126,
|
|
"num_tokens": 934799.0,
|
|
"step": 455
|
|
},
|
|
{
|
|
"entropy": 7.288649463653565,
|
|
"epoch": 0.02784250824682989,
|
|
"grad_norm": 1.0390625,
|
|
"learning_rate": 0.00022950000000000002,
|
|
"loss": 6.8847,
|
|
"mean_token_accuracy": 0.10213120430707931,
|
|
"num_tokens": 945757.0,
|
|
"step": 460
|
|
},
|
|
{
|
|
"entropy": 7.276965951919555,
|
|
"epoch": 0.02814514420603456,
|
|
"grad_norm": 1.0,
|
|
"learning_rate": 0.00023200000000000003,
|
|
"loss": 6.8831,
|
|
"mean_token_accuracy": 0.0962677277624607,
|
|
"num_tokens": 956143.0,
|
|
"step": 465
|
|
},
|
|
{
|
|
"entropy": 7.307726716995239,
|
|
"epoch": 0.028447780165239234,
|
|
"grad_norm": 1.0859375,
|
|
"learning_rate": 0.00023449999999999998,
|
|
"loss": 6.7977,
|
|
"mean_token_accuracy": 0.10646740570664406,
|
|
"num_tokens": 966437.0,
|
|
"step": 470
|
|
},
|
|
{
|
|
"entropy": 7.251181745529175,
|
|
"epoch": 0.028750416124443907,
|
|
"grad_norm": 1.0390625,
|
|
"learning_rate": 0.000237,
|
|
"loss": 6.8266,
|
|
"mean_token_accuracy": 0.09941198974847794,
|
|
"num_tokens": 977350.0,
|
|
"step": 475
|
|
},
|
|
{
|
|
"entropy": 7.3098362445831295,
|
|
"epoch": 0.02905305208364858,
|
|
"grad_norm": 0.890625,
|
|
"learning_rate": 0.0002395,
|
|
"loss": 6.9111,
|
|
"mean_token_accuracy": 0.09969582483172416,
|
|
"num_tokens": 988340.0,
|
|
"step": 480
|
|
},
|
|
{
|
|
"entropy": 7.254582738876342,
|
|
"epoch": 0.029355688042853254,
|
|
"grad_norm": 0.94921875,
|
|
"learning_rate": 0.000242,
|
|
"loss": 6.7656,
|
|
"mean_token_accuracy": 0.10246146842837334,
|
|
"num_tokens": 998323.0,
|
|
"step": 485
|
|
},
|
|
{
|
|
"entropy": 7.18165078163147,
|
|
"epoch": 0.029658324002057923,
|
|
"grad_norm": 0.984375,
|
|
"learning_rate": 0.0002445,
|
|
"loss": 6.6808,
|
|
"mean_token_accuracy": 0.10682199671864509,
|
|
"num_tokens": 1007731.0,
|
|
"step": 490
|
|
},
|
|
{
|
|
"entropy": 7.063388872146606,
|
|
"epoch": 0.029960959961262596,
|
|
"grad_norm": 1.0703125,
|
|
"learning_rate": 0.000247,
|
|
"loss": 6.6377,
|
|
"mean_token_accuracy": 0.10866327434778214,
|
|
"num_tokens": 1017543.0,
|
|
"step": 495
|
|
},
|
|
{
|
|
"entropy": 7.027464437484741,
|
|
"epoch": 0.03026359592046727,
|
|
"grad_norm": 1.109375,
|
|
"learning_rate": 0.0002495,
|
|
"loss": 6.7361,
|
|
"mean_token_accuracy": 0.10657615810632706,
|
|
"num_tokens": 1028558.0,
|
|
"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": 222273528791040.0,
|
|
"train_batch_size": 16,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|