|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 0.07424456158586383, |
|
"eval_steps": 500, |
|
"global_step": 500, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.0, |
|
"grad_norm": 1.75, |
|
"learning_rate": 2.9673590504451043e-07, |
|
"loss": 1.433, |
|
"step": 1 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"grad_norm": 1.578125, |
|
"learning_rate": 1.483679525222552e-06, |
|
"loss": 1.5006, |
|
"step": 5 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"grad_norm": 1.4921875, |
|
"learning_rate": 2.967359050445104e-06, |
|
"loss": 1.4919, |
|
"step": 10 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"grad_norm": 1.4921875, |
|
"learning_rate": 4.451038575667656e-06, |
|
"loss": 1.4531, |
|
"step": 15 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"grad_norm": 1.0234375, |
|
"learning_rate": 5.934718100890208e-06, |
|
"loss": 1.4242, |
|
"step": 20 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"grad_norm": 0.9453125, |
|
"learning_rate": 7.418397626112759e-06, |
|
"loss": 1.3932, |
|
"step": 25 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"grad_norm": 0.8359375, |
|
"learning_rate": 8.902077151335312e-06, |
|
"loss": 1.3661, |
|
"step": 30 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"grad_norm": 0.828125, |
|
"learning_rate": 1.0385756676557864e-05, |
|
"loss": 1.3677, |
|
"step": 35 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"grad_norm": 0.59375, |
|
"learning_rate": 1.1869436201780416e-05, |
|
"loss": 1.3497, |
|
"step": 40 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"grad_norm": 0.56640625, |
|
"learning_rate": 1.3353115727002968e-05, |
|
"loss": 1.3049, |
|
"step": 45 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"grad_norm": 0.439453125, |
|
"learning_rate": 1.4836795252225518e-05, |
|
"loss": 1.2836, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"grad_norm": 0.46484375, |
|
"learning_rate": 1.6320474777448072e-05, |
|
"loss": 1.3133, |
|
"step": 55 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"grad_norm": 0.40625, |
|
"learning_rate": 1.7804154302670624e-05, |
|
"loss": 1.2596, |
|
"step": 60 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"grad_norm": 0.408203125, |
|
"learning_rate": 1.9287833827893176e-05, |
|
"loss": 1.2612, |
|
"step": 65 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"grad_norm": 0.376953125, |
|
"learning_rate": 2.0771513353115728e-05, |
|
"loss": 1.2584, |
|
"step": 70 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"grad_norm": 0.40625, |
|
"learning_rate": 2.225519287833828e-05, |
|
"loss": 1.2386, |
|
"step": 75 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"grad_norm": 0.353515625, |
|
"learning_rate": 2.3738872403560832e-05, |
|
"loss": 1.1833, |
|
"step": 80 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"grad_norm": 0.349609375, |
|
"learning_rate": 2.5222551928783384e-05, |
|
"loss": 1.2411, |
|
"step": 85 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"grad_norm": 0.36328125, |
|
"learning_rate": 2.6706231454005936e-05, |
|
"loss": 1.1956, |
|
"step": 90 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"grad_norm": 0.369140625, |
|
"learning_rate": 2.818991097922849e-05, |
|
"loss": 1.2327, |
|
"step": 95 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"grad_norm": 0.353515625, |
|
"learning_rate": 2.9673590504451037e-05, |
|
"loss": 1.1896, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"grad_norm": 0.3984375, |
|
"learning_rate": 3.115727002967359e-05, |
|
"loss": 1.1979, |
|
"step": 105 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"grad_norm": 0.388671875, |
|
"learning_rate": 3.2640949554896144e-05, |
|
"loss": 1.2004, |
|
"step": 110 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"grad_norm": 0.392578125, |
|
"learning_rate": 3.41246290801187e-05, |
|
"loss": 1.1624, |
|
"step": 115 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"grad_norm": 0.400390625, |
|
"learning_rate": 3.560830860534125e-05, |
|
"loss": 1.1914, |
|
"step": 120 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"grad_norm": 0.41796875, |
|
"learning_rate": 3.70919881305638e-05, |
|
"loss": 1.1952, |
|
"step": 125 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"grad_norm": 0.4140625, |
|
"learning_rate": 3.857566765578635e-05, |
|
"loss": 1.1538, |
|
"step": 130 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"grad_norm": 0.4375, |
|
"learning_rate": 4.005934718100891e-05, |
|
"loss": 1.1702, |
|
"step": 135 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"grad_norm": 0.44140625, |
|
"learning_rate": 4.1543026706231456e-05, |
|
"loss": 1.1683, |
|
"step": 140 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"grad_norm": 0.443359375, |
|
"learning_rate": 4.3026706231454005e-05, |
|
"loss": 1.1822, |
|
"step": 145 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"grad_norm": 0.482421875, |
|
"learning_rate": 4.451038575667656e-05, |
|
"loss": 1.1649, |
|
"step": 150 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"grad_norm": 0.46484375, |
|
"learning_rate": 4.5994065281899116e-05, |
|
"loss": 1.1868, |
|
"step": 155 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"grad_norm": 0.5, |
|
"learning_rate": 4.7477744807121664e-05, |
|
"loss": 1.161, |
|
"step": 160 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"grad_norm": 0.451171875, |
|
"learning_rate": 4.896142433234421e-05, |
|
"loss": 1.1454, |
|
"step": 165 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"grad_norm": 0.45703125, |
|
"learning_rate": 5.044510385756677e-05, |
|
"loss": 1.1388, |
|
"step": 170 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"grad_norm": 0.490234375, |
|
"learning_rate": 5.1928783382789324e-05, |
|
"loss": 1.1253, |
|
"step": 175 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"grad_norm": 0.51171875, |
|
"learning_rate": 5.341246290801187e-05, |
|
"loss": 1.1527, |
|
"step": 180 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"grad_norm": 0.486328125, |
|
"learning_rate": 5.489614243323442e-05, |
|
"loss": 1.1234, |
|
"step": 185 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"grad_norm": 0.46875, |
|
"learning_rate": 5.637982195845698e-05, |
|
"loss": 1.113, |
|
"step": 190 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"grad_norm": 0.458984375, |
|
"learning_rate": 5.7863501483679525e-05, |
|
"loss": 1.1368, |
|
"step": 195 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"grad_norm": 0.484375, |
|
"learning_rate": 5.9347181008902074e-05, |
|
"loss": 1.137, |
|
"step": 200 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"grad_norm": 0.4765625, |
|
"learning_rate": 6.0830860534124636e-05, |
|
"loss": 1.1286, |
|
"step": 205 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"grad_norm": 0.484375, |
|
"learning_rate": 6.231454005934718e-05, |
|
"loss": 1.1277, |
|
"step": 210 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"grad_norm": 0.474609375, |
|
"learning_rate": 6.379821958456974e-05, |
|
"loss": 1.1268, |
|
"step": 215 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"grad_norm": 0.47265625, |
|
"learning_rate": 6.528189910979229e-05, |
|
"loss": 1.0993, |
|
"step": 220 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"grad_norm": 0.50390625, |
|
"learning_rate": 6.676557863501484e-05, |
|
"loss": 1.1017, |
|
"step": 225 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"grad_norm": 0.4609375, |
|
"learning_rate": 6.82492581602374e-05, |
|
"loss": 1.1345, |
|
"step": 230 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"grad_norm": 0.48828125, |
|
"learning_rate": 6.973293768545995e-05, |
|
"loss": 1.1086, |
|
"step": 235 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"grad_norm": 0.45703125, |
|
"learning_rate": 7.12166172106825e-05, |
|
"loss": 1.0791, |
|
"step": 240 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"grad_norm": 0.470703125, |
|
"learning_rate": 7.270029673590505e-05, |
|
"loss": 1.1158, |
|
"step": 245 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"grad_norm": 0.46875, |
|
"learning_rate": 7.41839762611276e-05, |
|
"loss": 1.132, |
|
"step": 250 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"grad_norm": 0.4609375, |
|
"learning_rate": 7.566765578635016e-05, |
|
"loss": 1.1438, |
|
"step": 255 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"grad_norm": 0.470703125, |
|
"learning_rate": 7.71513353115727e-05, |
|
"loss": 1.1405, |
|
"step": 260 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"grad_norm": 0.447265625, |
|
"learning_rate": 7.863501483679525e-05, |
|
"loss": 1.1124, |
|
"step": 265 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"grad_norm": 0.486328125, |
|
"learning_rate": 8.011869436201782e-05, |
|
"loss": 1.0813, |
|
"step": 270 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"grad_norm": 0.48046875, |
|
"learning_rate": 8.160237388724036e-05, |
|
"loss": 1.1194, |
|
"step": 275 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"grad_norm": 0.462890625, |
|
"learning_rate": 8.308605341246291e-05, |
|
"loss": 1.0927, |
|
"step": 280 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"grad_norm": 0.48046875, |
|
"learning_rate": 8.456973293768546e-05, |
|
"loss": 1.1277, |
|
"step": 285 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"grad_norm": 0.4453125, |
|
"learning_rate": 8.605341246290801e-05, |
|
"loss": 1.1271, |
|
"step": 290 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"grad_norm": 0.435546875, |
|
"learning_rate": 8.753709198813057e-05, |
|
"loss": 1.1162, |
|
"step": 295 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"grad_norm": 0.44921875, |
|
"learning_rate": 8.902077151335312e-05, |
|
"loss": 1.0857, |
|
"step": 300 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"grad_norm": 0.4375, |
|
"learning_rate": 9.050445103857568e-05, |
|
"loss": 1.0869, |
|
"step": 305 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"grad_norm": 0.4453125, |
|
"learning_rate": 9.198813056379823e-05, |
|
"loss": 1.0655, |
|
"step": 310 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"grad_norm": 0.4296875, |
|
"learning_rate": 9.347181008902077e-05, |
|
"loss": 1.0585, |
|
"step": 315 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"grad_norm": 0.41796875, |
|
"learning_rate": 9.495548961424333e-05, |
|
"loss": 1.1144, |
|
"step": 320 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"grad_norm": 0.43359375, |
|
"learning_rate": 9.643916913946588e-05, |
|
"loss": 1.0719, |
|
"step": 325 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"grad_norm": 0.416015625, |
|
"learning_rate": 9.792284866468843e-05, |
|
"loss": 1.0919, |
|
"step": 330 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"grad_norm": 0.423828125, |
|
"learning_rate": 9.940652818991099e-05, |
|
"loss": 1.1223, |
|
"step": 335 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"grad_norm": 0.431640625, |
|
"learning_rate": 0.00010089020771513354, |
|
"loss": 1.0565, |
|
"step": 340 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"grad_norm": 0.431640625, |
|
"learning_rate": 0.00010237388724035609, |
|
"loss": 1.0962, |
|
"step": 345 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"grad_norm": 0.44921875, |
|
"learning_rate": 0.00010385756676557865, |
|
"loss": 1.0959, |
|
"step": 350 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"grad_norm": 0.43359375, |
|
"learning_rate": 0.0001053412462908012, |
|
"loss": 1.0628, |
|
"step": 355 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"grad_norm": 0.431640625, |
|
"learning_rate": 0.00010682492581602374, |
|
"loss": 1.0975, |
|
"step": 360 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"grad_norm": 0.42578125, |
|
"learning_rate": 0.0001083086053412463, |
|
"loss": 1.0727, |
|
"step": 365 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"grad_norm": 0.416015625, |
|
"learning_rate": 0.00010979228486646884, |
|
"loss": 1.0649, |
|
"step": 370 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"grad_norm": 0.4296875, |
|
"learning_rate": 0.00011127596439169139, |
|
"loss": 1.0904, |
|
"step": 375 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"grad_norm": 0.3984375, |
|
"learning_rate": 0.00011275964391691397, |
|
"loss": 1.079, |
|
"step": 380 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"grad_norm": 0.40234375, |
|
"learning_rate": 0.0001142433234421365, |
|
"loss": 1.0522, |
|
"step": 385 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"grad_norm": 0.431640625, |
|
"learning_rate": 0.00011572700296735905, |
|
"loss": 1.0579, |
|
"step": 390 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"grad_norm": 0.396484375, |
|
"learning_rate": 0.0001172106824925816, |
|
"loss": 1.0871, |
|
"step": 395 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"grad_norm": 0.41015625, |
|
"learning_rate": 0.00011869436201780415, |
|
"loss": 1.0936, |
|
"step": 400 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"grad_norm": 0.412109375, |
|
"learning_rate": 0.00012017804154302672, |
|
"loss": 1.0734, |
|
"step": 405 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"grad_norm": 0.392578125, |
|
"learning_rate": 0.00012166172106824927, |
|
"loss": 1.0657, |
|
"step": 410 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"grad_norm": 0.4140625, |
|
"learning_rate": 0.00012314540059347182, |
|
"loss": 1.0884, |
|
"step": 415 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"grad_norm": 0.408203125, |
|
"learning_rate": 0.00012462908011869436, |
|
"loss": 1.0683, |
|
"step": 420 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"grad_norm": 0.3984375, |
|
"learning_rate": 0.00012611275964391692, |
|
"loss": 1.1073, |
|
"step": 425 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"grad_norm": 0.412109375, |
|
"learning_rate": 0.00012759643916913948, |
|
"loss": 1.0849, |
|
"step": 430 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"grad_norm": 0.40625, |
|
"learning_rate": 0.00012908011869436204, |
|
"loss": 1.0798, |
|
"step": 435 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"grad_norm": 0.404296875, |
|
"learning_rate": 0.00013056379821958458, |
|
"loss": 1.1029, |
|
"step": 440 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"grad_norm": 0.3828125, |
|
"learning_rate": 0.0001320474777448071, |
|
"loss": 1.0664, |
|
"step": 445 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"grad_norm": 0.38671875, |
|
"learning_rate": 0.00013353115727002967, |
|
"loss": 1.0998, |
|
"step": 450 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"grad_norm": 0.40234375, |
|
"learning_rate": 0.00013501483679525224, |
|
"loss": 1.0834, |
|
"step": 455 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"grad_norm": 0.400390625, |
|
"learning_rate": 0.0001364985163204748, |
|
"loss": 1.062, |
|
"step": 460 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"grad_norm": 0.3984375, |
|
"learning_rate": 0.00013798219584569733, |
|
"loss": 1.0825, |
|
"step": 465 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"grad_norm": 0.388671875, |
|
"learning_rate": 0.0001394658753709199, |
|
"loss": 1.0689, |
|
"step": 470 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"grad_norm": 0.392578125, |
|
"learning_rate": 0.00014094955489614243, |
|
"loss": 1.0557, |
|
"step": 475 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"grad_norm": 0.380859375, |
|
"learning_rate": 0.000142433234421365, |
|
"loss": 1.0582, |
|
"step": 480 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"grad_norm": 0.380859375, |
|
"learning_rate": 0.00014391691394658756, |
|
"loss": 1.0921, |
|
"step": 485 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"grad_norm": 0.384765625, |
|
"learning_rate": 0.0001454005934718101, |
|
"loss": 1.0544, |
|
"step": 490 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"grad_norm": 0.39453125, |
|
"learning_rate": 0.00014688427299703265, |
|
"loss": 1.0333, |
|
"step": 495 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"grad_norm": 0.384765625, |
|
"learning_rate": 0.0001483679525222552, |
|
"loss": 1.0454, |
|
"step": 500 |
|
} |
|
], |
|
"logging_steps": 5, |
|
"max_steps": 6734, |
|
"num_input_tokens_seen": 0, |
|
"num_train_epochs": 1, |
|
"save_steps": 100, |
|
"total_flos": 7.031295252564541e+17, |
|
"train_batch_size": 2, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|