| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 500, | |
| "global_step": 527, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "grad_norm": 18.569529999247134, | |
| "learning_rate": 6.25e-06, | |
| "loss": 2.2344, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "grad_norm": 17.774880014109563, | |
| "learning_rate": 1.25e-05, | |
| "loss": 2.2188, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 16.126910180369833, | |
| "learning_rate": 1.8750000000000002e-05, | |
| "loss": 2.0938, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 4.004205399706303, | |
| "learning_rate": 2.5e-05, | |
| "loss": 1.5625, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 4.475863116578534, | |
| "learning_rate": 3.125e-05, | |
| "loss": 1.5547, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 3.9418457877357183, | |
| "learning_rate": 3.7500000000000003e-05, | |
| "loss": 1.5312, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 3.35497476295258, | |
| "learning_rate": 4.375e-05, | |
| "loss": 1.5312, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 2.0238486012287824, | |
| "learning_rate": 5e-05, | |
| "loss": 1.4453, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 1.3025920158121342, | |
| "learning_rate": 5.6250000000000005e-05, | |
| "loss": 1.4062, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 2.1578221590331648, | |
| "learning_rate": 6.25e-05, | |
| "loss": 1.4141, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 1.639048247445056, | |
| "learning_rate": 6.875e-05, | |
| "loss": 1.3906, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 1.0758955455274364, | |
| "learning_rate": 7.500000000000001e-05, | |
| "loss": 1.3672, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 0.9988841624058318, | |
| "learning_rate": 8.125000000000001e-05, | |
| "loss": 1.3516, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 1.036582323851064, | |
| "learning_rate": 8.75e-05, | |
| "loss": 1.3438, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 1.0040586950026587, | |
| "learning_rate": 9.375e-05, | |
| "loss": 1.3438, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 0.8112402128458829, | |
| "learning_rate": 0.0001, | |
| "loss": 1.3047, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 0.806608090114202, | |
| "learning_rate": 9.999905507663936e-05, | |
| "loss": 1.3203, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 0.8164159045230494, | |
| "learning_rate": 9.99962203422726e-05, | |
| "loss": 1.2812, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 0.7335139492736167, | |
| "learning_rate": 9.9991495904044e-05, | |
| "loss": 1.2969, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 0.799772371827826, | |
| "learning_rate": 9.998488194052287e-05, | |
| "loss": 1.2734, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 0.6718755777667486, | |
| "learning_rate": 9.997637870169672e-05, | |
| "loss": 1.2734, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 0.7599864750693303, | |
| "learning_rate": 9.996598650896192e-05, | |
| "loss": 1.2578, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 0.6886726902695142, | |
| "learning_rate": 9.995370575511151e-05, | |
| "loss": 1.2656, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 0.6823417918062775, | |
| "learning_rate": 9.993953690432031e-05, | |
| "loss": 1.2734, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 0.6382222948594751, | |
| "learning_rate": 9.99234804921275e-05, | |
| "loss": 1.2422, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 0.6452753693126838, | |
| "learning_rate": 9.990553712541617e-05, | |
| "loss": 1.2422, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 0.5904936283655213, | |
| "learning_rate": 9.988570748239062e-05, | |
| "loss": 1.2266, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 0.5640365175871925, | |
| "learning_rate": 9.986399231255056e-05, | |
| "loss": 1.2031, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 0.514358456822939, | |
| "learning_rate": 9.984039243666283e-05, | |
| "loss": 1.2344, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 0.636209974460906, | |
| "learning_rate": 9.981490874673039e-05, | |
| "loss": 1.1953, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 0.5479990507502773, | |
| "learning_rate": 9.978754220595861e-05, | |
| "loss": 1.2031, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 0.5561261900343224, | |
| "learning_rate": 9.975829384871884e-05, | |
| "loss": 1.2344, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 0.4948569883144301, | |
| "learning_rate": 9.97271647805093e-05, | |
| "loss": 1.2109, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 0.5167119119909488, | |
| "learning_rate": 9.969415617791336e-05, | |
| "loss": 1.2266, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 0.508354119180698, | |
| "learning_rate": 9.965926928855499e-05, | |
| "loss": 1.2109, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 0.5326133584880987, | |
| "learning_rate": 9.962250543105167e-05, | |
| "loss": 1.1797, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 0.47996354720772333, | |
| "learning_rate": 9.95838659949645e-05, | |
| "loss": 1.1719, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 0.4584403899961059, | |
| "learning_rate": 9.954335244074574e-05, | |
| "loss": 1.1797, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 0.478024134726985, | |
| "learning_rate": 9.950096629968352e-05, | |
| "loss": 1.1953, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 0.4831225145909792, | |
| "learning_rate": 9.945670917384403e-05, | |
| "loss": 1.1797, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 0.47054504799219404, | |
| "learning_rate": 9.941058273601096e-05, | |
| "loss": 1.1719, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 0.4940788339187542, | |
| "learning_rate": 9.936258872962228e-05, | |
| "loss": 1.1719, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 0.44655071198353546, | |
| "learning_rate": 9.931272896870426e-05, | |
| "loss": 1.1953, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 0.4620978351861828, | |
| "learning_rate": 9.926100533780303e-05, | |
| "loss": 1.1797, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 0.4777738095085538, | |
| "learning_rate": 9.920741979191331e-05, | |
| "loss": 1.1641, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 0.46687195896903444, | |
| "learning_rate": 9.91519743564044e-05, | |
| "loss": 1.1562, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 0.4572029233269296, | |
| "learning_rate": 9.909467112694384e-05, | |
| "loss": 1.1719, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 0.4379619610466461, | |
| "learning_rate": 9.903551226941801e-05, | |
| "loss": 1.1562, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 0.4625672515990539, | |
| "learning_rate": 9.897450001985039e-05, | |
| "loss": 1.1406, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 0.45763509639124433, | |
| "learning_rate": 9.891163668431695e-05, | |
| "loss": 1.1719, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 0.4759593297993926, | |
| "learning_rate": 9.88469246388591e-05, | |
| "loss": 1.1484, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 0.47690674058291993, | |
| "learning_rate": 9.878036632939374e-05, | |
| "loss": 1.2031, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 0.44532349751989236, | |
| "learning_rate": 9.871196427162092e-05, | |
| "loss": 1.1172, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 0.452106382176349, | |
| "learning_rate": 9.86417210509288e-05, | |
| "loss": 1.1094, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 0.452753191760464, | |
| "learning_rate": 9.85696393222957e-05, | |
| "loss": 1.1953, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 0.45441530517528445, | |
| "learning_rate": 9.849572181019007e-05, | |
| "loss": 1.1406, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 0.5007896403646297, | |
| "learning_rate": 9.84199713084672e-05, | |
| "loss": 1.1406, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 0.4608644722591853, | |
| "learning_rate": 9.834239068026387e-05, | |
| "loss": 1.1328, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 0.4968111745587068, | |
| "learning_rate": 9.826298285789002e-05, | |
| "loss": 1.1328, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 0.44573050581792, | |
| "learning_rate": 9.818175084271786e-05, | |
| "loss": 1.125, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "grad_norm": 0.46589557608902915, | |
| "learning_rate": 9.809869770506856e-05, | |
| "loss": 1.1328, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "grad_norm": 0.45678602237538535, | |
| "learning_rate": 9.80138265840961e-05, | |
| "loss": 1.1328, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "grad_norm": 0.43959375518292026, | |
| "learning_rate": 9.792714068766872e-05, | |
| "loss": 1.125, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "grad_norm": 0.41984631355693813, | |
| "learning_rate": 9.783864329224752e-05, | |
| "loss": 1.1406, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "grad_norm": 0.43389373514903795, | |
| "learning_rate": 9.774833774276278e-05, | |
| "loss": 1.1172, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "grad_norm": 0.4336892341975021, | |
| "learning_rate": 9.765622745248739e-05, | |
| "loss": 1.1406, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "grad_norm": 0.4410380396458464, | |
| "learning_rate": 9.7562315902908e-05, | |
| "loss": 1.1172, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "grad_norm": 0.5141753129119535, | |
| "learning_rate": 9.746660664359326e-05, | |
| "loss": 1.1172, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "grad_norm": 0.46159702650365997, | |
| "learning_rate": 9.736910329205978e-05, | |
| "loss": 1.1328, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "grad_norm": 0.45575998013176866, | |
| "learning_rate": 9.726980953363536e-05, | |
| "loss": 1.1328, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "grad_norm": 0.4458400134106401, | |
| "learning_rate": 9.716872912131964e-05, | |
| "loss": 1.1172, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "grad_norm": 0.4403442516042718, | |
| "learning_rate": 9.706586587564237e-05, | |
| "loss": 1.1406, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "grad_norm": 0.4513799624319814, | |
| "learning_rate": 9.696122368451886e-05, | |
| "loss": 1.1406, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "grad_norm": 0.4348698465314392, | |
| "learning_rate": 9.685480650310318e-05, | |
| "loss": 1.1641, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "grad_norm": 0.4498345537005005, | |
| "learning_rate": 9.674661835363858e-05, | |
| "loss": 1.125, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "grad_norm": 0.43144309981302603, | |
| "learning_rate": 9.663666332530541e-05, | |
| "loss": 1.1172, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "grad_norm": 0.4399921408069473, | |
| "learning_rate": 9.652494557406666e-05, |