| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 500, | |
| "global_step": 1329, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0007524454477050414, | |
| "grad_norm": 0.006495347712188959, | |
| "learning_rate": 1.999997206043851e-05, | |
| "loss": 0.0004, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0015048908954100827, | |
| "grad_norm": 0.05232545733451843, | |
| "learning_rate": 1.9999888241910165e-05, | |
| "loss": 0.0017, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.002257336343115124, | |
| "grad_norm": 0.016717378050088882, | |
| "learning_rate": 1.999974854488333e-05, | |
| "loss": 0.0009, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.0030097817908201654, | |
| "grad_norm": 0.51008540391922, | |
| "learning_rate": 1.9999552970138628e-05, | |
| "loss": 0.0233, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.003762227238525207, | |
| "grad_norm": 3.963065682910383e-05, | |
| "learning_rate": 1.999930151876891e-05, | |
| "loss": 0.0, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.004514672686230248, | |
| "grad_norm": 8.221525604312774e-06, | |
| "learning_rate": 1.9998994192179256e-05, | |
| "loss": 0.0, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.005267118133935289, | |
| "grad_norm": 0.019136814400553703, | |
| "learning_rate": 1.999863099208699e-05, | |
| "loss": 0.0005, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.006019563581640331, | |
| "grad_norm": 0.0003651017614174634, | |
| "learning_rate": 1.9998211920521646e-05, | |
| "loss": 0.0, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.006772009029345372, | |
| "grad_norm": 0.0008514254004694521, | |
| "learning_rate": 1.9997736979824944e-05, | |
| "loss": 0.0, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.007524454477050414, | |
| "grad_norm": 0.005437616258859634, | |
| "learning_rate": 1.9997206172650826e-05, | |
| "loss": 0.0002, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.008276899924755455, | |
| "grad_norm": 0.5016034245491028, | |
| "learning_rate": 1.9996619501965385e-05, | |
| "loss": 0.0063, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.009029345372460496, | |
| "grad_norm": 0.0040865750052034855, | |
| "learning_rate": 1.9995976971046896e-05, | |
| "loss": 0.0001, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.009781790820165538, | |
| "grad_norm": 0.06221418455243111, | |
| "learning_rate": 1.9995278583485755e-05, | |
| "loss": 0.0013, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.010534236267870579, | |
| "grad_norm": 0.34792619943618774, | |
| "learning_rate": 1.9994524343184494e-05, | |
| "loss": 0.0127, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.011286681715575621, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.999371425435775e-05, | |
| "loss": 0.0, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.012039127163280662, | |
| "grad_norm": 7.092195510864258, | |
| "learning_rate": 1.9992848321532213e-05, | |
| "loss": 0.5353, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.012791572610985704, | |
| "grad_norm": 0.0006019376451149583, | |
| "learning_rate": 1.9991926549546653e-05, | |
| "loss": 0.0, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.013544018058690745, | |
| "grad_norm": 0.026538817211985588, | |
| "learning_rate": 1.9990948943551843e-05, | |
| "loss": 0.0006, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.014296463506395787, | |
| "grad_norm": 0.0013464675284922123, | |
| "learning_rate": 1.998991550901056e-05, | |
| "loss": 0.0, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.015048908954100828, | |
| "grad_norm": 5.049379825592041, | |
| "learning_rate": 1.998882625169755e-05, | |
| "loss": 0.2596, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.01580135440180587, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.9987681177699486e-05, | |
| "loss": 0.0, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.01655379984951091, | |
| "grad_norm": 0.0007531401934102178, | |
| "learning_rate": 1.9986480293414938e-05, | |
| "loss": 0.0, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.01730624529721595, | |
| "grad_norm": 1.847428560256958, | |
| "learning_rate": 1.9985223605554346e-05, | |
| "loss": 0.1395, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.01805869074492099, | |
| "grad_norm": 0.013212837278842926, | |
| "learning_rate": 1.998391112113997e-05, | |
| "loss": 0.0001, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.018811136192626036, | |
| "grad_norm": 0.41415101289749146, | |
| "learning_rate": 1.9982542847505858e-05, | |
| "loss": 0.0105, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.019563581640331076, | |
| "grad_norm": 0.0014141664141789079, | |
| "learning_rate": 1.99811187922978e-05, | |
| "loss": 0.0001, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.020316027088036117, | |
| "grad_norm": 0.002521197311580181, | |
| "learning_rate": 1.9979638963473294e-05, | |
| "loss": 0.0001, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.021068472535741158, | |
| "grad_norm": 4.3378071784973145, | |
| "learning_rate": 1.9978103369301495e-05, | |
| "loss": 0.3246, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.0218209179834462, | |
| "grad_norm": 4.08248852181714e-05, | |
| "learning_rate": 1.997651201836317e-05, | |
| "loss": 0.0, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.022573363431151242, | |
| "grad_norm": 0.006794753950089216, | |
| "learning_rate": 1.9974864919550642e-05, | |
| "loss": 0.0003, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.023325808878856283, | |
| "grad_norm": 0.05969928577542305, | |
| "learning_rate": 1.9973162082067762e-05, | |
| "loss": 0.0011, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.024078254326561323, | |
| "grad_norm": 0.07203614711761475, | |
| "learning_rate": 1.9971403515429833e-05, | |
| "loss": 0.002, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.024830699774266364, | |
| "grad_norm": 0.0011654079426079988, | |
| "learning_rate": 1.996958922946357e-05, | |
| "loss": 0.0001, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.025583145221971408, | |
| "grad_norm": 0.002619029488414526, | |
| "learning_rate": 1.9967719234307044e-05, | |
| "loss": 0.0001, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.02633559066967645, | |
| "grad_norm": 0.48534148931503296, | |
| "learning_rate": 1.9965793540409628e-05, | |
| "loss": 0.0088, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.02708803611738149, | |
| "grad_norm": 0.00325326737947762, | |
| "learning_rate": 1.9963812158531926e-05, | |
| "loss": 0.0001, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.02784048156508653, | |
| "grad_norm": 0.008502104319632053, | |
| "learning_rate": 1.9961775099745727e-05, | |
| "loss": 0.0003, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.028592927012791574, | |
| "grad_norm": 0.9508514404296875, | |
| "learning_rate": 1.995968237543394e-05, | |
| "loss": 0.114, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.029345372460496615, | |
| "grad_norm": 1.4725682735443115, | |
| "learning_rate": 1.9957533997290524e-05, | |
| "loss": 0.096, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.030097817908201655, | |
| "grad_norm": 0.1169065311551094, | |
| "learning_rate": 1.9955329977320422e-05, | |
| "loss": 0.0018, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.030850263355906696, | |
| "grad_norm": 0.054003071039915085, | |
| "learning_rate": 1.9953070327839513e-05, | |
| "loss": 0.0012, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.03160270880361174, | |
| "grad_norm": 0.0011868480360135436, | |
| "learning_rate": 1.9950755061474513e-05, | |
| "loss": 0.0001, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.03235515425131678, | |
| "grad_norm": 0.27812469005584717, | |
| "learning_rate": 1.9948384191162932e-05, | |
| "loss": 0.0074, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.03310759969902182, | |
| "grad_norm": 2.0421926975250244, | |
| "learning_rate": 1.994595773015298e-05, | |
| "loss": 0.2558, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.033860045146726865, | |
| "grad_norm": 0.0036680996417999268, | |
| "learning_rate": 1.9943475692003514e-05, | |
| "loss": 0.0001, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.0346124905944319, | |
| "grad_norm": 0.0018471762305125594, | |
| "learning_rate": 1.994093809058394e-05, | |
| "loss": 0.0001, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.035364936042136946, | |
| "grad_norm": 2.0573832988739014, | |
| "learning_rate": 1.9938344940074162e-05, | |
| "loss": 0.2316, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.03611738148984198, | |
| "grad_norm": 0.02541675977408886, | |
| "learning_rate": 1.9935696254964468e-05, | |
| "loss": 0.0009, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.03686982693754703, | |
| "grad_norm": 0.039695482701063156, | |
| "learning_rate": 1.9932992050055478e-05, | |
| "loss": 0.001, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.03762227238525207, | |
| "grad_norm": 0.0008929629693739116, | |
| "learning_rate": 1.993023234045806e-05, | |
| "loss": 0.0001, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.03837471783295711, | |
| "grad_norm": 0.0004025468078907579, | |
| "learning_rate": 1.992741714159322e-05, | |
| "loss": 0.0, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.03912716328066215, | |
| "grad_norm": 0.012651098892092705, | |
| "learning_rate": 1.992454646919205e-05, | |
| "loss": 0.0006, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.0398796087283672, | |
| "grad_norm": 5.611824035644531, | |
| "learning_rate": 1.9921620339295612e-05, | |
| "loss": 0.0996, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.040632054176072234, | |
| "grad_norm": 0.024261457845568657, | |
| "learning_rate": 1.9918638768254865e-05, | |
| "loss": 0.0008, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.04138449962377728, | |
| "grad_norm": 0.04541151970624924, | |
| "learning_rate": 1.9915601772730562e-05, | |
| "loss": 0.0013, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.042136945071482315, | |
| "grad_norm": 0.0012648458359763026, | |
| "learning_rate": 1.9912509369693172e-05, | |
| "loss": 0.0001, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.04288939051918736, | |
| "grad_norm": 0.009661543183028698, | |
| "learning_rate": 1.990936157642277e-05, | |
| "loss": 0.0004, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.0436418359668924, | |
| "grad_norm": 0.00047065879334695637, | |
| "learning_rate": 1.990615841050895e-05, | |
| "loss": 0.0, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.04439428141459744, | |
| "grad_norm": 2.226644992828369, | |
| "learning_rate": 1.990289988985072e-05, | |
| "loss": 0.0725, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.045146726862302484, | |
| "grad_norm": 0.007543186657130718, | |
| "learning_rate": 1.9899586032656407e-05, | |
| "loss": 0.0003, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.04589917231000752, | |
| "grad_norm": 0.3191293478012085, | |
| "learning_rate": 1.9896216857443563e-05, | |
| "loss": 0.0149, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.046651617757712566, | |
| "grad_norm": 0.0007462618523277342, | |
| "learning_rate": 1.989279238303883e-05, | |
| "loss": 0.0, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.04740406320541761, | |
| "grad_norm": 0.1068795919418335, | |
| "learning_rate": 1.9889312628577887e-05, | |
| "loss": 0.0024, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.04815650865312265, | |
| "grad_norm": 0.003478952683508396, | |
| "learning_rate": 1.9885777613505278e-05, | |
| "loss": 0.0002, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.04890895410082769, | |
| "grad_norm": 0.002991155255585909, | |
| "learning_rate": 1.988218735757437e-05, | |
| "loss": 0.0001, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.04966139954853273, | |
| "grad_norm": 0.003038755152374506, | |
| "learning_rate": 1.98785418808472e-05, | |
| "loss": 0.0001, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.05041384499623777, | |
| "grad_norm": 1.9981627464294434, | |
| "learning_rate": 1.987484120369436e-05, | |
| "loss": 0.5126, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.051166290443942816, | |
| "grad_norm": 0.1265110969543457, | |
| "learning_rate": 1.9871085346794922e-05, | |
| "loss": 0.0043, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.05191873589164785, | |
| "grad_norm": 0.5709707736968994, | |
| "learning_rate": 1.9867274331136276e-05, | |
| "loss": 0.1178, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.0526711813393529, | |
| "grad_norm": 0.17010557651519775, | |
| "learning_rate": 1.986340817801405e-05, | |
| "loss": 0.0031, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.05342362678705794, | |
| "grad_norm": 0.1551416665315628, | |
| "learning_rate": 1.985948690903196e-05, | |
| "loss": 0.0035, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.05417607223476298, | |
| "grad_norm": 0.17882609367370605, | |
| "learning_rate": 1.9855510546101725e-05, | |
| "loss": 0.0066, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.05492851768246802, | |
| "grad_norm": 0.014555556699633598, | |
| "learning_rate": 1.9851479111442902e-05, | |
| "loss": 0.0008, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.05568096313017306, | |
| "grad_norm": 0.03164775297045708, | |
| "learning_rate": 1.98473926275828e-05, | |
| "loss": 0.0014, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.056433408577878104, | |
| "grad_norm": 1.0504040718078613, | |
| "learning_rate": 1.984325111735633e-05, | |
| "loss": 0.047, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.05718585402558315, | |
| "grad_norm": 0.06992053985595703, | |
| "learning_rate": 1.9839054603905887e-05, | |
| "loss": 0.0025, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.057938299473288185, | |
| "grad_norm": 3.3065266609191895, | |
| "learning_rate": 1.9834803110681223e-05, | |
| "loss": 0.039, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.05869074492099323, | |
| "grad_norm": 0.04477689415216446, | |
| "learning_rate": 1.983049666143931e-05, | |
| "loss": 0.0017, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.059443190368698266, | |
| "grad_norm": 0.027361512184143066, | |
| "learning_rate": 1.9826135280244204e-05, | |
| "loss": 0.0011, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.06019563581640331, | |
| "grad_norm": 0.2297053188085556, | |
| "learning_rate": 1.9821718991466925e-05, | |
| "loss": 0.0076, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.060948081264108354, | |
| "grad_norm": 0.015010291710495949, | |
| "learning_rate": 1.9817247819785303e-05, | |
| "loss": 0.0007, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.06170052671181339, | |
| "grad_norm": 1.8525534868240356, | |
| "learning_rate": 1.981272179018386e-05, | |
| "loss": 0.3579, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.062452972159518436, | |
| "grad_norm": 0.02152411825954914, | |
| "learning_rate": 1.9808140927953644e-05, | |
| "loss": 0.0006, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.06320541760722348, | |
| "grad_norm": 0.0057389941066503525, | |
| "learning_rate": 1.9803505258692117e-05, | |
| "loss": 0.0004, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.06395786305492852, | |
| "grad_norm": 0.021787824109196663, | |
| "learning_rate": 1.9798814808302992e-05, | |
| "loss": 0.0008, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.06471030850263355, | |
| "grad_norm": 1.793501853942871, | |
| "learning_rate": 1.9794069602996093e-05, | |
| "loss": 0.1386, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.0654627539503386, | |
| "grad_norm": 0.9608948826789856, | |
| "learning_rate": 1.9789269669287212e-05, | |
| "loss": 0.126, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.06621519939804364, | |
| "grad_norm": 0.01728960871696472, | |
| "learning_rate": 1.9784415033997955e-05, | |
| "loss": 0.0008, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.06696764484574869, | |
| "grad_norm": 0.0034939961042255163, | |
| "learning_rate": 1.9779505724255602e-05, | |
| "loss": 0.0002, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.06772009029345373, | |
| "grad_norm": 0.27871963381767273, | |
| "learning_rate": 1.9774541767492942e-05, | |
| "loss": 0.1145, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.06847253574115876, | |
| "grad_norm": 0.020328784361481667, | |
| "learning_rate": 1.9769523191448136e-05, | |
| "loss": 0.0009, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.0692249811888638, | |
| "grad_norm": 0.003004940692335367, | |
| "learning_rate": 1.976445002416454e-05, | |
| "loss": 0.0002, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.06997742663656885, | |
| "grad_norm": 0.051723234355449677, | |
| "learning_rate": 1.975932229399057e-05, | |
| "loss": 0.0021, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.07072987208427389, | |
| "grad_norm": 0.02858145534992218, | |
| "learning_rate": 1.975414002957953e-05, | |
| "loss": 0.0011, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.07148231753197894, | |
| "grad_norm": 0.076414555311203, | |
| "learning_rate": 1.9748903259889466e-05, | |
| "loss": 0.0024, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.07223476297968397, | |
| "grad_norm": 1.1670242547988892, | |
| "learning_rate": 1.9743612014182982e-05, | |
| "loss": 0.0559, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.07298720842738901, | |
| "grad_norm": 0.05960305780172348, | |
| "learning_rate": 1.9738266322027094e-05, | |
| "loss": 0.0026, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.07373965387509406, | |
| "grad_norm": 0.4573524296283722, | |
| "learning_rate": 1.9732866213293066e-05, | |
| "loss": 0.082, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.0744920993227991, | |
| "grad_norm": 0.02990127168595791, | |
| "learning_rate": 1.972741171815623e-05, | |
| "loss": 0.0011, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.07524454477050414, | |
| "grad_norm": 1.6312720775604248, | |
| "learning_rate": 1.9721902867095828e-05, | |
| "loss": 0.1075, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.07599699021820917, | |
| "grad_norm": 0.3260236978530884, | |
| "learning_rate": 1.9716339690894834e-05, | |
| "loss": 0.0788, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.07674943566591422, | |
| "grad_norm": 0.0010308397468179464, | |
| "learning_rate": 1.9710722220639785e-05, | |
| "loss": 0.0001, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.07750188111361926, | |
| "grad_norm": 0.04024788737297058, | |
| "learning_rate": 1.9705050487720618e-05, | |
| "loss": 0.0016, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.0782543265613243, | |
| "grad_norm": 5.523307800292969, | |
| "learning_rate": 1.969932452383048e-05, | |
| "loss": 0.3277, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.07900677200902935, | |
| "grad_norm": 2.359933853149414, | |
| "learning_rate": 1.9693544360965548e-05, | |
| "loss": 0.189, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.0797592174567344, | |
| "grad_norm": 0.6973968744277954, | |
| "learning_rate": 1.9687710031424873e-05, | |
| "loss": 0.0215, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.08051166290443942, | |
| "grad_norm": 0.01043291948735714, | |
| "learning_rate": 1.9681821567810172e-05, | |
| "loss": 0.0004, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.08126410835214447, | |
| "grad_norm": 0.0018412582576274872, | |
| "learning_rate": 1.9675879003025668e-05, | |
| "loss": 0.0001, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.08201655379984951, | |
| "grad_norm": 0.6676415205001831, | |
| "learning_rate": 1.9669882370277885e-05, | |
| "loss": 0.1539, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.08276899924755456, | |
| "grad_norm": 0.006418765056878328, | |
| "learning_rate": 1.9663831703075488e-05, | |
| "loss": 0.0004, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.0835214446952596, | |
| "grad_norm": 0.0075980802066624165, | |
| "learning_rate": 1.9657727035229066e-05, | |
| "loss": 0.0004, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.08427389014296463, | |
| "grad_norm": 0.004072375129908323, | |
| "learning_rate": 1.9651568400850976e-05, | |
| "loss": 0.0001, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.08502633559066967, | |
| "grad_norm": 0.6514503359794617, | |
| "learning_rate": 1.964535583435512e-05, | |
| "loss": 0.0636, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.08577878103837472, | |
| "grad_norm": 0.6835861802101135, | |
| "learning_rate": 1.9639089370456784e-05, | |
| "loss": 0.1035, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.08653122648607976, | |
| "grad_norm": 0.18389393389225006, | |
| "learning_rate": 1.963276904417241e-05, | |
| "loss": 0.0048, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.0872836719337848, | |
| "grad_norm": 0.002048420486971736, | |
| "learning_rate": 1.962639489081943e-05, | |
| "loss": 0.0001, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.08803611738148984, | |
| "grad_norm": 0.010052050463855267, | |
| "learning_rate": 1.9619966946016054e-05, | |
| "loss": 0.0005, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.08878856282919488, | |
| "grad_norm": 0.100750632584095, | |
| "learning_rate": 1.9613485245681073e-05, | |
| "loss": 0.0027, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.08954100827689992, | |
| "grad_norm": 0.04654483497142792, | |
| "learning_rate": 1.960694982603366e-05, | |
| "loss": 0.0019, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.09029345372460497, | |
| "grad_norm": 0.01746903918683529, | |
| "learning_rate": 1.960036072359317e-05, | |
| "loss": 0.0006, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.09104589917231001, | |
| "grad_norm": 0.05034410208463669, | |
| "learning_rate": 1.9593717975178924e-05, | |
| "loss": 0.002, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.09179834462001504, | |
| "grad_norm": 3.37355637550354, | |
| "learning_rate": 1.958702161791002e-05, | |
| "loss": 0.5772, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.09255079006772009, | |
| "grad_norm": 0.0004869294061791152, | |
| "learning_rate": 1.958027168920512e-05, | |
| "loss": 0.0, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.09330323551542513, | |
| "grad_norm": 0.013796951621770859, | |
| "learning_rate": 1.9573468226782224e-05, | |
| "loss": 0.0006, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.09405568096313018, | |
| "grad_norm": 0.020256061106920242, | |
| "learning_rate": 1.956661126865849e-05, | |
| "loss": 0.0007, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.09480812641083522, | |
| "grad_norm": 0.4556258022785187, | |
| "learning_rate": 1.9559700853149997e-05, | |
| "loss": 0.007, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.09556057185854025, | |
| "grad_norm": 1.9277138710021973, | |
| "learning_rate": 1.9552737018871543e-05, | |
| "loss": 0.0393, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.0963130173062453, | |
| "grad_norm": 0.0008740609628148377, | |
| "learning_rate": 1.954571980473642e-05, | |
| "loss": 0.0001, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.09706546275395034, | |
| "grad_norm": 0.03682279586791992, | |
| "learning_rate": 1.953864924995621e-05, | |
| "loss": 0.0015, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.09781790820165538, | |
| "grad_norm": 0.03125663846731186, | |
| "learning_rate": 1.9531525394040546e-05, | |
| "loss": 0.0008, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.09857035364936043, | |
| "grad_norm": 0.0102821821346879, | |
| "learning_rate": 1.9524348276796913e-05, | |
| "loss": 0.0004, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.09932279909706546, | |
| "grad_norm": 0.07433848083019257, | |
| "learning_rate": 1.9517117938330414e-05, | |
| "loss": 0.0029, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.1000752445447705, | |
| "grad_norm": 0.31897610425949097, | |
| "learning_rate": 1.9509834419043544e-05, | |
| "loss": 0.0816, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.10082768999247554, | |
| "grad_norm": 0.042768437415361404, | |
| "learning_rate": 1.9502497759635973e-05, | |
| "loss": 0.0017, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.10158013544018059, | |
| "grad_norm": 0.27461105585098267, | |
| "learning_rate": 1.9495108001104312e-05, | |
| "loss": 0.0972, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.10233258088788563, | |
| "grad_norm": 0.012465243227779865, | |
| "learning_rate": 1.9487665184741878e-05, | |
| "loss": 0.0006, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.10308502633559068, | |
| "grad_norm": 0.042917292565107346, | |
| "learning_rate": 1.9480169352138473e-05, | |
| "loss": 0.0014, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.1038374717832957, | |
| "grad_norm": 0.06652813404798508, | |
| "learning_rate": 1.9472620545180165e-05, | |
| "loss": 0.0018, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.10458991723100075, | |
| "grad_norm": 3.4145336151123047, | |
| "learning_rate": 1.9465018806049014e-05, | |
| "loss": 0.0251, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.1053423626787058, | |
| "grad_norm": 0.5462217926979065, | |
| "learning_rate": 1.9457364177222877e-05, | |
| "loss": 0.1197, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.10609480812641084, | |
| "grad_norm": 0.00809057243168354, | |
| "learning_rate": 1.9449656701475147e-05, | |
| "loss": 0.0004, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.10684725357411588, | |
| "grad_norm": 0.03647633641958237, | |
| "learning_rate": 1.944189642187452e-05, | |
| "loss": 0.0008, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.10759969902182091, | |
| "grad_norm": 0.007750968914479017, | |
| "learning_rate": 1.9434083381784764e-05, | |
| "loss": 0.0002, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.10835214446952596, | |
| "grad_norm": 0.002081293612718582, | |
| "learning_rate": 1.9426217624864456e-05, | |
| "loss": 0.0001, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.109104589917231, | |
| "grad_norm": 2.682119131088257, | |
| "learning_rate": 1.9418299195066755e-05, | |
| "loss": 0.1876, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.10985703536493605, | |
| "grad_norm": 0.21694301068782806, | |
| "learning_rate": 1.9410328136639163e-05, | |
| "loss": 0.0035, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.11060948081264109, | |
| "grad_norm": 0.020226623862981796, | |
| "learning_rate": 1.940230449412324e-05, | |
| "loss": 0.0006, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.11136192626034612, | |
| "grad_norm": 0.006655423901975155, | |
| "learning_rate": 1.939422831235441e-05, | |
| "loss": 0.0002, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.11211437170805116, | |
| "grad_norm": 0.011194628663361073, | |
| "learning_rate": 1.938609963646166e-05, | |
| "loss": 0.0005, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.11286681715575621, | |
| "grad_norm": 1.6491260528564453, | |
| "learning_rate": 1.9377918511867318e-05, | |
| "loss": 0.1711, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.11361926260346125, | |
| "grad_norm": 0.20927202701568604, | |
| "learning_rate": 1.9369684984286798e-05, | |
| "loss": 0.0075, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.1143717080511663, | |
| "grad_norm": 0.10048650205135345, | |
| "learning_rate": 1.9361399099728326e-05, | |
| "loss": 0.0033, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.11512415349887133, | |
| "grad_norm": 0.09928741306066513, | |
| "learning_rate": 1.9353060904492694e-05, | |
| "loss": 0.0038, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.11587659894657637, | |
| "grad_norm": 0.3446449935436249, | |
| "learning_rate": 1.934467044517301e-05, | |
| "loss": 0.0654, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.11662904439428141, | |
| "grad_norm": 0.007841469720005989, | |
| "learning_rate": 1.9336227768654424e-05, | |
| "loss": 0.0004, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.11738148984198646, | |
| "grad_norm": 0.24826376140117645, | |
| "learning_rate": 1.9327732922113872e-05, | |
| "loss": 0.006, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.1181339352896915, | |
| "grad_norm": 2.012798547744751, | |
| "learning_rate": 1.9319185953019817e-05, | |
| "loss": 0.1098, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.11888638073739653, | |
| "grad_norm": 0.02779470756649971, | |
| "learning_rate": 1.9310586909131964e-05, | |
| "loss": 0.0009, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.11963882618510158, | |
| "grad_norm": 3.675586462020874, | |
| "learning_rate": 1.930193583850102e-05, | |
| "loss": 0.5443, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.12039127163280662, | |
| "grad_norm": 0.009048126637935638, | |
| "learning_rate": 1.9293232789468403e-05, | |
| "loss": 0.0005, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.12114371708051166, | |
| "grad_norm": 0.00870896689593792, | |
| "learning_rate": 1.9284477810666e-05, | |
| "loss": 0.0004, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.12189616252821671, | |
| "grad_norm": 0.01799067109823227, | |
| "learning_rate": 1.9275670951015854e-05, | |
| "loss": 0.0009, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.12264860797592174, | |
| "grad_norm": 0.07708598673343658, | |
| "learning_rate": 1.9266812259729927e-05, | |
| "loss": 0.0032, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.12340105342362678, | |
| "grad_norm": 0.012997242622077465, | |
| "learning_rate": 1.9257901786309813e-05, | |
| "loss": 0.0003, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.12415349887133183, | |
| "grad_norm": 0.08036967366933823, | |
| "learning_rate": 1.9248939580546453e-05, | |
| "loss": 0.0021, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.12490594431903687, | |
| "grad_norm": 0.08968926221132278, | |
| "learning_rate": 1.9239925692519867e-05, | |
| "loss": 0.0036, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.1256583897667419, | |
| "grad_norm": 0.10857436060905457, | |
| "learning_rate": 1.923086017259887e-05, | |
| "loss": 0.0045, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.12641083521444696, | |
| "grad_norm": 0.33034518361091614, | |
| "learning_rate": 1.9221743071440792e-05, | |
| "loss": 0.012, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.127163280662152, | |
| "grad_norm": 0.020488563925027847, | |
| "learning_rate": 1.92125744399912e-05, | |
| "loss": 0.0011, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.12791572610985705, | |
| "grad_norm": 1.1725200414657593, | |
| "learning_rate": 1.9203354329483593e-05, | |
| "loss": 0.2263, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.12866817155756208, | |
| "grad_norm": 0.009000587277114391, | |
| "learning_rate": 1.9194082791439146e-05, | |
| "loss": 0.0005, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.1294206170052671, | |
| "grad_norm": 0.15425604581832886, | |
| "learning_rate": 1.9184759877666403e-05, | |
| "loss": 0.006, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.13017306245297217, | |
| "grad_norm": 0.04585198312997818, | |
| "learning_rate": 1.917538564026098e-05, | |
| "loss": 0.0011, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.1309255079006772, | |
| "grad_norm": 0.012005002237856388, | |
| "learning_rate": 1.9165960131605304e-05, | |
| "loss": 0.0007, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.13167795334838225, | |
| "grad_norm": 0.04149067774415016, | |
| "learning_rate": 1.915648340436828e-05, | |
| "loss": 0.0017, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.13243039879608728, | |
| "grad_norm": 0.13023491203784943, | |
| "learning_rate": 1.9146955511505035e-05, | |
| "loss": 0.0047, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.13318284424379231, | |
| "grad_norm": 1.0637022256851196, | |
| "learning_rate": 1.91373765062566e-05, | |
| "loss": 0.2339, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.13393528969149737, | |
| "grad_norm": 0.6275419592857361, | |
| "learning_rate": 1.9127746442149612e-05, | |
| "loss": 0.1668, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.1346877351392024, | |
| "grad_norm": 0.009538069367408752, | |
| "learning_rate": 1.9118065372996027e-05, | |
| "loss": 0.0006, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.13544018058690746, | |
| "grad_norm": 0.01008265744894743, | |
| "learning_rate": 1.910833335289281e-05, | |
| "loss": 0.0006, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.1361926260346125, | |
| "grad_norm": 0.030792739242315292, | |
| "learning_rate": 1.9098550436221636e-05, | |
| "loss": 0.0014, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.13694507148231752, | |
| "grad_norm": 0.010570832528173923, | |
| "learning_rate": 1.9088716677648583e-05, | |
| "loss": 0.0005, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.13769751693002258, | |
| "grad_norm": 0.7934188842773438, | |
| "learning_rate": 1.9078832132123833e-05, | |
| "loss": 0.0201, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.1384499623777276, | |
| "grad_norm": 0.004436484072357416, | |
| "learning_rate": 1.9068896854881364e-05, | |
| "loss": 0.0002, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.13920240782543267, | |
| "grad_norm": 0.025362318381667137, | |
| "learning_rate": 1.9058910901438628e-05, | |
| "loss": 0.0017, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.1399548532731377, | |
| "grad_norm": 0.007729747798293829, | |
| "learning_rate": 1.904887432759626e-05, | |
| "loss": 0.0004, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.14070729872084273, | |
| "grad_norm": 2.8755412101745605, | |
| "learning_rate": 1.9038787189437752e-05, | |
| "loss": 0.1614, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.14145974416854779, | |
| "grad_norm": 2.1707630157470703, | |
| "learning_rate": 1.902864954332915e-05, | |
| "loss": 0.3061, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.14221218961625282, | |
| "grad_norm": 0.5352933406829834, | |
| "learning_rate": 1.9018461445918727e-05, | |
| "loss": 0.1411, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.14296463506395787, | |
| "grad_norm": 0.44579315185546875, | |
| "learning_rate": 1.900822295413668e-05, | |
| "loss": 0.1438, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.1437170805116629, | |
| "grad_norm": 2.404839038848877, | |
| "learning_rate": 1.8997934125194806e-05, | |
| "loss": 0.2562, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.14446952595936793, | |
| "grad_norm": 2.7213966846466064, | |
| "learning_rate": 1.898759501658618e-05, | |
| "loss": 0.1053, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.145221971407073, | |
| "grad_norm": 1.2511435747146606, | |
| "learning_rate": 1.8977205686084828e-05, | |
| "loss": 0.1214, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.14597441685477802, | |
| "grad_norm": 0.9504565000534058, | |
| "learning_rate": 1.8966766191745423e-05, | |
| "loss": 0.136, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.14672686230248308, | |
| "grad_norm": 0.053916994482278824, | |
| "learning_rate": 1.895627659190294e-05, | |
| "loss": 0.0035, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.1474793077501881, | |
| "grad_norm": 0.21418577432632446, | |
| "learning_rate": 1.8945736945172345e-05, | |
| "loss": 0.0128, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.14823175319789314, | |
| "grad_norm": 0.22742077708244324, | |
| "learning_rate": 1.8935147310448258e-05, | |
| "loss": 0.011, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.1489841986455982, | |
| "grad_norm": 0.06237643212080002, | |
| "learning_rate": 1.8924507746904628e-05, | |
| "loss": 0.0035, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.14973664409330323, | |
| "grad_norm": 0.027896396815776825, | |
| "learning_rate": 1.8913818313994406e-05, | |
| "loss": 0.0017, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.1504890895410083, | |
| "grad_norm": 0.021885687485337257, | |
| "learning_rate": 1.8903079071449206e-05, | |
| "loss": 0.0013, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.15124153498871332, | |
| "grad_norm": 0.004547674674540758, | |
| "learning_rate": 1.889229007927897e-05, | |
| "loss": 0.0003, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.15199398043641835, | |
| "grad_norm": 0.08865097910165787, | |
| "learning_rate": 1.8881451397771647e-05, | |
| "loss": 0.0046, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.1527464258841234, | |
| "grad_norm": 0.13727599382400513, | |
| "learning_rate": 1.887056308749283e-05, | |
| "loss": 0.0075, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.15349887133182843, | |
| "grad_norm": 0.6231827735900879, | |
| "learning_rate": 1.885962520928545e-05, | |
| "loss": 0.028, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.1542513167795335, | |
| "grad_norm": 0.1673537939786911, | |
| "learning_rate": 1.88486378242694e-05, | |
| "loss": 0.0099, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.15500376222723852, | |
| "grad_norm": 0.008311329409480095, | |
| "learning_rate": 1.8837600993841237e-05, | |
| "loss": 0.0005, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.15575620767494355, | |
| "grad_norm": 0.1961093544960022, | |
| "learning_rate": 1.8826514779673792e-05, | |
| "loss": 0.01, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.1565086531226486, | |
| "grad_norm": 0.3649809658527374, | |
| "learning_rate": 1.881537924371586e-05, | |
| "loss": 0.1412, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.15726109857035364, | |
| "grad_norm": 1.2528727054595947, | |
| "learning_rate": 1.8804194448191843e-05, | |
| "loss": 0.1621, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.1580135440180587, | |
| "grad_norm": 0.0017916648648679256, | |
| "learning_rate": 1.8792960455601396e-05, | |
| "loss": 0.0001, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.15876598946576373, | |
| "grad_norm": 0.2502264976501465, | |
| "learning_rate": 1.8781677328719078e-05, | |
| "loss": 0.01, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.1595184349134688, | |
| "grad_norm": 0.3349985182285309, | |
| "learning_rate": 1.8770345130594017e-05, | |
| "loss": 0.0182, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.16027088036117382, | |
| "grad_norm": 0.007236495614051819, | |
| "learning_rate": 1.875896392454955e-05, | |
| "loss": 0.0005, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.16102332580887885, | |
| "grad_norm": 0.7724468111991882, | |
| "learning_rate": 1.8747533774182845e-05, | |
| "loss": 0.0307, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.1617757712565839, | |
| "grad_norm": 0.0025126487016677856, | |
| "learning_rate": 1.8736054743364587e-05, | |
| "loss": 0.0001, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.16252821670428894, | |
| "grad_norm": 0.1561364233493805, | |
| "learning_rate": 1.872452689623859e-05, | |
| "loss": 0.0034, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.163280662151994, | |
| "grad_norm": 0.004739896394312382, | |
| "learning_rate": 1.8712950297221457e-05, | |
| "loss": 0.0003, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.16403310759969902, | |
| "grad_norm": 0.5904443264007568, | |
| "learning_rate": 1.8701325011002204e-05, | |
| "loss": 0.1231, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.16478555304740405, | |
| "grad_norm": 0.002162502147257328, | |
| "learning_rate": 1.8689651102541915e-05, | |
| "loss": 0.0001, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.1655379984951091, | |
| "grad_norm": 0.022224850952625275, | |
| "learning_rate": 1.8677928637073367e-05, | |
| "loss": 0.0007, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.16629044394281414, | |
| "grad_norm": 0.0014565088786184788, | |
| "learning_rate": 1.8666157680100663e-05, | |
| "loss": 0.0001, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.1670428893905192, | |
| "grad_norm": 0.0017851298907771707, | |
| "learning_rate": 1.865433829739888e-05, | |
| "loss": 0.0001, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.16779533483822423, | |
| "grad_norm": 0.16844072937965393, | |
| "learning_rate": 1.8642470555013695e-05, | |
| "loss": 0.0023, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.16854778028592926, | |
| "grad_norm": 3.2930855751037598, | |
| "learning_rate": 1.8630554519261007e-05, | |
| "loss": 0.0823, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.16930022573363432, | |
| "grad_norm": 0.00976943876594305, | |
| "learning_rate": 1.8618590256726587e-05, | |
| "loss": 0.0006, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.17005267118133935, | |
| "grad_norm": 0.3796117901802063, | |
| "learning_rate": 1.860657783426568e-05, | |
| "loss": 0.1048, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.1708051166290444, | |
| "grad_norm": 2.5069665908813477, | |
| "learning_rate": 1.8594517319002646e-05, | |
| "loss": 0.3186, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.17155756207674944, | |
| "grad_norm": 0.05536516755819321, | |
| "learning_rate": 1.8582408778330588e-05, | |
| "loss": 0.0019, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.17231000752445447, | |
| "grad_norm": 0.013733879663050175, | |
| "learning_rate": 1.8570252279910975e-05, | |
| "loss": 0.0005, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.17306245297215953, | |
| "grad_norm": 0.26580435037612915, | |
| "learning_rate": 1.8558047891673247e-05, | |
| "loss": 0.0078, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.17381489841986456, | |
| "grad_norm": 0.005103581584990025, | |
| "learning_rate": 1.854579568181446e-05, | |
| "loss": 0.0003, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.1745673438675696, | |
| "grad_norm": 0.8314829468727112, | |
| "learning_rate": 1.8533495718798882e-05, | |
| "loss": 0.0161, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.17531978931527464, | |
| "grad_norm": 0.04428403079509735, | |
| "learning_rate": 1.8521148071357633e-05, | |
| "loss": 0.0012, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.17607223476297967, | |
| "grad_norm": 0.033717166632413864, | |
| "learning_rate": 1.8508752808488283e-05, | |
| "loss": 0.0014, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.17682468021068473, | |
| "grad_norm": 0.03402457386255264, | |
| "learning_rate": 1.8496309999454475e-05, | |
| "loss": 0.0013, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.17757712565838976, | |
| "grad_norm": 0.062332432717084885, | |
| "learning_rate": 1.848381971378553e-05, | |
| "loss": 0.0019, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.17832957110609482, | |
| "grad_norm": 5.1166229248046875, | |
| "learning_rate": 1.8471282021276073e-05, | |
| "loss": 0.1784, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.17908201655379985, | |
| "grad_norm": 0.0013702790020033717, | |
| "learning_rate": 1.845869699198563e-05, | |
| "loss": 0.0001, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.17983446200150488, | |
| "grad_norm": 0.01880730129778385, | |
| "learning_rate": 1.844606469623824e-05, | |
| "loss": 0.0005, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.18058690744920994, | |
| "grad_norm": 1.5539264678955078, | |
| "learning_rate": 1.8433385204622067e-05, | |
| "loss": 0.1897, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.18133935289691497, | |
| "grad_norm": 4.555483341217041, | |
| "learning_rate": 1.842065858798899e-05, | |
| "loss": 0.327, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.18209179834462003, | |
| "grad_norm": 0.014411378651857376, | |
| "learning_rate": 1.8407884917454233e-05, | |
| "loss": 0.0008, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.18284424379232506, | |
| "grad_norm": 0.3658387362957001, | |
| "learning_rate": 1.8395064264395945e-05, | |
| "loss": 0.0159, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.1835966892400301, | |
| "grad_norm": 0.052855461835861206, | |
| "learning_rate": 1.838219670045481e-05, | |
| "loss": 0.0018, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.18434913468773514, | |
| "grad_norm": 0.03438251465559006, | |
| "learning_rate": 1.836928229753365e-05, | |
| "loss": 0.0012, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.18510158013544017, | |
| "grad_norm": 0.24612069129943848, | |
| "learning_rate": 1.835632112779701e-05, | |
| "loss": 0.1226, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.18585402558314523, | |
| "grad_norm": 0.04806346446275711, | |
| "learning_rate": 1.8343313263670782e-05, | |
| "loss": 0.0021, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 0.18660647103085026, | |
| "grad_norm": 0.6806573867797852, | |
| "learning_rate": 1.8330258777841755e-05, | |
| "loss": 0.0208, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.1873589164785553, | |
| "grad_norm": 0.11799991130828857, | |
| "learning_rate": 1.831715774325726e-05, | |
| "loss": 0.0036, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 0.18811136192626035, | |
| "grad_norm": 0.002671067602932453, | |
| "learning_rate": 1.830401023312472e-05, | |
| "loss": 0.0001, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.18886380737396538, | |
| "grad_norm": 0.2947131097316742, | |
| "learning_rate": 1.8290816320911285e-05, | |
| "loss": 0.1261, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 0.18961625282167044, | |
| "grad_norm": 1.2162272930145264, | |
| "learning_rate": 1.8277576080343362e-05, | |
| "loss": 0.1268, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.19036869826937547, | |
| "grad_norm": 1.2172619104385376, | |
| "learning_rate": 1.8264289585406266e-05, | |
| "loss": 0.0297, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 0.1911211437170805, | |
| "grad_norm": 0.20552274584770203, | |
| "learning_rate": 1.825095691034376e-05, | |
| "loss": 0.0085, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.19187358916478556, | |
| "grad_norm": 0.008679857477545738, | |
| "learning_rate": 1.8237578129657664e-05, | |
| "loss": 0.0002, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.1926260346124906, | |
| "grad_norm": 0.0659710094332695, | |
| "learning_rate": 1.822415331810743e-05, | |
| "loss": 0.002, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.19337848006019565, | |
| "grad_norm": 0.4817584753036499, | |
| "learning_rate": 1.821068255070973e-05, | |
| "loss": 0.1022, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 0.19413092550790068, | |
| "grad_norm": 1.5975401401519775, | |
| "learning_rate": 1.819716590273803e-05, | |
| "loss": 0.1331, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.1948833709556057, | |
| "grad_norm": 0.05313832312822342, | |
| "learning_rate": 1.818360344972217e-05, | |
| "loss": 0.0016, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 0.19563581640331076, | |
| "grad_norm": 0.02122344821691513, | |
| "learning_rate": 1.8169995267447953e-05, | |
| "loss": 0.0008, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.1963882618510158, | |
| "grad_norm": 0.0018245746614411473, | |
| "learning_rate": 1.8156341431956706e-05, | |
| "loss": 0.0001, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 0.19714070729872085, | |
| "grad_norm": 0.07786925882101059, | |
| "learning_rate": 1.814264201954486e-05, | |
| "loss": 0.0018, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.19789315274642588, | |
| "grad_norm": 0.46468350291252136, | |
| "learning_rate": 1.812889710676354e-05, | |
| "loss": 0.013, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 0.1986455981941309, | |
| "grad_norm": 0.4126303493976593, | |
| "learning_rate": 1.811510677041811e-05, | |
| "loss": 0.0055, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.19939804364183597, | |
| "grad_norm": 0.20423245429992676, | |
| "learning_rate": 1.8101271087567753e-05, | |
| "loss": 0.1157, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.200150489089541, | |
| "grad_norm": 0.9413779377937317, | |
| "learning_rate": 1.8087390135525056e-05, | |
| "loss": 0.0718, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.20090293453724606, | |
| "grad_norm": 0.5326927304267883, | |
| "learning_rate": 1.8073463991855562e-05, | |
| "loss": 0.0053, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 0.2016553799849511, | |
| "grad_norm": 0.2141987830400467, | |
| "learning_rate": 1.8059492734377342e-05, | |
| "loss": 0.1024, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.20240782543265612, | |
| "grad_norm": 4.482271194458008, | |
| "learning_rate": 1.8045476441160552e-05, | |
| "loss": 0.3966, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 0.20316027088036118, | |
| "grad_norm": 0.001211398164741695, | |
| "learning_rate": 1.8031415190527016e-05, | |
| "loss": 0.0001, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.2039127163280662, | |
| "grad_norm": 0.06283948570489883, | |
| "learning_rate": 1.8017309061049767e-05, | |
| "loss": 0.0014, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 0.20466516177577126, | |
| "grad_norm": 0.655404806137085, | |
| "learning_rate": 1.8003158131552615e-05, | |
| "loss": 0.0746, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.2054176072234763, | |
| "grad_norm": 0.010463729500770569, | |
| "learning_rate": 1.7988962481109716e-05, | |
| "loss": 0.0003, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 0.20617005267118135, | |
| "grad_norm": 0.0010110132861882448, | |
| "learning_rate": 1.7974722189045126e-05, | |
| "loss": 0.0001, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.20692249811888638, | |
| "grad_norm": 0.1091262623667717, | |
| "learning_rate": 1.7960437334932334e-05, | |
| "loss": 0.0042, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.2076749435665914, | |
| "grad_norm": 5.209902286529541, | |
| "learning_rate": 1.7946107998593867e-05, | |
| "loss": 0.0551, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.20842738901429647, | |
| "grad_norm": 0.00470168748870492, | |
| "learning_rate": 1.7931734260100792e-05, | |
| "loss": 0.0002, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 0.2091798344620015, | |
| "grad_norm": 0.7987887263298035, | |
| "learning_rate": 1.7917316199772296e-05, | |
| "loss": 0.0129, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 0.20993227990970656, | |
| "grad_norm": 0.021809808909893036, | |
| "learning_rate": 1.7902853898175244e-05, | |
| "loss": 0.0011, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 0.2106847253574116, | |
| "grad_norm": 0.3490143120288849, | |
| "learning_rate": 1.7888347436123707e-05, | |
| "loss": 0.1188, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.21143717080511662, | |
| "grad_norm": 0.6932041645050049, | |
| "learning_rate": 1.7873796894678514e-05, | |
| "loss": 0.0158, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 0.21218961625282168, | |
| "grad_norm": 0.006826372817158699, | |
| "learning_rate": 1.7859202355146826e-05, | |
| "loss": 0.0003, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 0.2129420617005267, | |
| "grad_norm": 0.00620373385027051, | |
| "learning_rate": 1.7844563899081642e-05, | |
| "loss": 0.0002, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 0.21369450714823177, | |
| "grad_norm": 0.0037128764670342207, | |
| "learning_rate": 1.782988160828137e-05, | |
| "loss": 0.0002, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.2144469525959368, | |
| "grad_norm": 0.008649413473904133, | |
| "learning_rate": 1.7815155564789374e-05, | |
| "loss": 0.0002, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.21519939804364183, | |
| "grad_norm": 0.0007761880406178534, | |
| "learning_rate": 1.780038585089348e-05, | |
| "loss": 0.0, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 0.21595184349134688, | |
| "grad_norm": 8.137001037597656, | |
| "learning_rate": 1.7785572549125566e-05, | |
| "loss": 0.2602, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 0.21670428893905191, | |
| "grad_norm": 0.011192454025149345, | |
| "learning_rate": 1.7770715742261057e-05, | |
| "loss": 0.0003, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.21745673438675697, | |
| "grad_norm": 0.9992365837097168, | |
| "learning_rate": 1.775581551331848e-05, | |
| "loss": 0.013, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 0.218209179834462, | |
| "grad_norm": 3.0127367973327637, | |
| "learning_rate": 1.7740871945559022e-05, | |
| "loss": 0.2808, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.21896162528216703, | |
| "grad_norm": 6.692657188978046e-05, | |
| "learning_rate": 1.772588512248602e-05, | |
| "loss": 0.0, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.2197140707298721, | |
| "grad_norm": 9.9885743111372e-05, | |
| "learning_rate": 1.771085512784453e-05, | |
| "loss": 0.0, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 0.22046651617757712, | |
| "grad_norm": 0.002337940037250519, | |
| "learning_rate": 1.769578204562084e-05, | |
| "loss": 0.0001, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 0.22121896162528218, | |
| "grad_norm": 4.053681373596191, | |
| "learning_rate": 1.7680665960042016e-05, | |
| "loss": 0.2921, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 0.2219714070729872, | |
| "grad_norm": 0.004839350003749132, | |
| "learning_rate": 1.7665506955575417e-05, | |
| "loss": 0.0003, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.22272385252069224, | |
| "grad_norm": 3.5069193840026855, | |
| "learning_rate": 1.765030511692823e-05, | |
| "loss": 0.3916, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 0.2234762979683973, | |
| "grad_norm": 0.10265739262104034, | |
| "learning_rate": 1.7635060529046994e-05, | |
| "loss": 0.0018, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 0.22422874341610233, | |
| "grad_norm": 8.340572357177734, | |
| "learning_rate": 1.7619773277117135e-05, | |
| "loss": 0.2325, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 0.22498118886380739, | |
| "grad_norm": 0.004961786326020956, | |
| "learning_rate": 1.760444344656247e-05, | |
| "loss": 0.0003, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 0.22573363431151242, | |
| "grad_norm": 0.2516036033630371, | |
| "learning_rate": 1.758907112304475e-05, | |
| "loss": 0.0051, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.22648607975921745, | |
| "grad_norm": 0.01247998047620058, | |
| "learning_rate": 1.757365639246317e-05, | |
| "loss": 0.0006, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 0.2272385252069225, | |
| "grad_norm": 0.0020404146052896976, | |
| "learning_rate": 1.7558199340953893e-05, | |
| "loss": 0.0001, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 0.22799097065462753, | |
| "grad_norm": 0.3927852213382721, | |
| "learning_rate": 1.7542700054889572e-05, | |
| "loss": 0.0126, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 0.2287434161023326, | |
| "grad_norm": 0.012273562140762806, | |
| "learning_rate": 1.752715862087885e-05, | |
| "loss": 0.0003, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 0.22949586155003762, | |
| "grad_norm": 0.0031097414903342724, | |
| "learning_rate": 1.7511575125765902e-05, | |
| "loss": 0.0002, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.23024830699774265, | |
| "grad_norm": 0.36811593174934387, | |
| "learning_rate": 1.7495949656629933e-05, | |
| "loss": 0.0076, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 0.2310007524454477, | |
| "grad_norm": 0.008364018052816391, | |
| "learning_rate": 1.748028230078469e-05, | |
| "loss": 0.0002, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 0.23175319789315274, | |
| "grad_norm": 0.47602003812789917, | |
| "learning_rate": 1.7464573145777987e-05, | |
| "loss": 0.1192, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 0.2325056433408578, | |
| "grad_norm": 0.858063280582428, | |
| "learning_rate": 1.7448822279391204e-05, | |
| "loss": 0.0553, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 0.23325808878856283, | |
| "grad_norm": 4.250735759735107, | |
| "learning_rate": 1.7433029789638794e-05, | |
| "loss": 0.0937, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.23401053423626786, | |
| "grad_norm": 0.029880870133638382, | |
| "learning_rate": 1.7417195764767816e-05, | |
| "loss": 0.0013, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 0.23476297968397292, | |
| "grad_norm": 0.04074908420443535, | |
| "learning_rate": 1.7401320293257403e-05, | |
| "loss": 0.0021, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 0.23551542513167795, | |
| "grad_norm": 0.03922910988330841, | |
| "learning_rate": 1.7385403463818308e-05, | |
| "loss": 0.0012, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 0.236267870579383, | |
| "grad_norm": 0.06185416132211685, | |
| "learning_rate": 1.7369445365392365e-05, | |
| "loss": 0.0031, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.23702031602708803, | |
| "grad_norm": 6.229604721069336, | |
| "learning_rate": 1.7353446087152038e-05, | |
| "loss": 0.0754, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.23777276147479307, | |
| "grad_norm": 0.4178987443447113, | |
| "learning_rate": 1.733740571849989e-05, | |
| "loss": 0.1242, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 0.23852520692249812, | |
| "grad_norm": 0.2695218324661255, | |
| "learning_rate": 1.732132434906809e-05, | |
| "loss": 0.0052, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 0.23927765237020315, | |
| "grad_norm": 0.011477818712592125, | |
| "learning_rate": 1.730520206871792e-05, | |
| "loss": 0.0004, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 0.2400300978179082, | |
| "grad_norm": 0.01223902590572834, | |
| "learning_rate": 1.728903896753927e-05, | |
| "loss": 0.0006, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 0.24078254326561324, | |
| "grad_norm": 0.02941061556339264, | |
| "learning_rate": 1.7272835135850133e-05, | |
| "loss": 0.0012, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.24153498871331827, | |
| "grad_norm": 0.45937439799308777, | |
| "learning_rate": 1.72565906641961e-05, | |
| "loss": 0.0131, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 0.24228743416102333, | |
| "grad_norm": 0.06633768230676651, | |
| "learning_rate": 1.7240305643349854e-05, | |
| "loss": 0.0012, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 0.24303987960872836, | |
| "grad_norm": 1.1109821796417236, | |
| "learning_rate": 1.7223980164310658e-05, | |
| "loss": 0.0369, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 0.24379232505643342, | |
| "grad_norm": 5.301788330078125, | |
| "learning_rate": 1.720761431830386e-05, | |
| "loss": 0.2436, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 0.24454477050413845, | |
| "grad_norm": 3.4349093437194824, | |
| "learning_rate": 1.719120819678038e-05, | |
| "loss": 0.3296, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.24529721595184348, | |
| "grad_norm": 0.13575129210948944, | |
| "learning_rate": 1.7174761891416176e-05, | |
| "loss": 0.0041, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 0.24604966139954854, | |
| "grad_norm": 0.7291696667671204, | |
| "learning_rate": 1.7158275494111763e-05, | |
| "loss": 0.1419, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 0.24680210684725357, | |
| "grad_norm": 0.011178013868629932, | |
| "learning_rate": 1.7141749096991686e-05, | |
| "loss": 0.0004, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 0.24755455229495862, | |
| "grad_norm": 0.17662179470062256, | |
| "learning_rate": 1.7125182792403995e-05, | |
| "loss": 0.0028, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 0.24830699774266365, | |
| "grad_norm": 0.8367990851402283, | |
| "learning_rate": 1.7108576672919757e-05, | |
| "loss": 0.0191, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.24905944319036868, | |
| "grad_norm": 3.8350296020507812, | |
| "learning_rate": 1.7091930831332507e-05, | |
| "loss": 0.1464, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 0.24981188863807374, | |
| "grad_norm": 0.8443143963813782, | |
| "learning_rate": 1.7075245360657744e-05, | |
| "loss": 0.0053, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 0.2505643340857788, | |
| "grad_norm": 0.04546971619129181, | |
| "learning_rate": 1.705852035413242e-05, | |
| "loss": 0.0014, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 0.2513167795334838, | |
| "grad_norm": 0.025590356439352036, | |
| "learning_rate": 1.7041755905214404e-05, | |
| "loss": 0.0013, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 0.2520692249811889, | |
| "grad_norm": 0.013746229000389576, | |
| "learning_rate": 1.7024952107581965e-05, | |
| "loss": 0.0004, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.2528216704288939, | |
| "grad_norm": 0.12485899776220322, | |
| "learning_rate": 1.700810905513325e-05, | |
| "loss": 0.0049, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 0.25357411587659895, | |
| "grad_norm": 0.20258508622646332, | |
| "learning_rate": 1.699122684198576e-05, | |
| "loss": 0.1096, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 0.254326561324304, | |
| "grad_norm": 0.026910100132226944, | |
| "learning_rate": 1.6974305562475825e-05, | |
| "loss": 0.0011, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 0.255079006772009, | |
| "grad_norm": 0.8987321853637695, | |
| "learning_rate": 1.6957345311158066e-05, | |
| "loss": 0.0763, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 0.2558314522197141, | |
| "grad_norm": 0.0026313799899071455, | |
| "learning_rate": 1.6940346182804876e-05, | |
| "loss": 0.0001, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.2565838976674191, | |
| "grad_norm": 6.234673976898193, | |
| "learning_rate": 1.69233082724059e-05, | |
| "loss": 0.1531, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 0.25733634311512416, | |
| "grad_norm": 0.0024953181855380535, | |
| "learning_rate": 1.6906231675167488e-05, | |
| "loss": 0.0002, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 0.2580887885628292, | |
| "grad_norm": 0.3546563684940338, | |
| "learning_rate": 1.6889116486512165e-05, | |
| "loss": 0.0036, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 0.2588412340105342, | |
| "grad_norm": 0.0036818115040659904, | |
| "learning_rate": 1.6871962802078103e-05, | |
| "loss": 0.0002, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 0.2595936794582393, | |
| "grad_norm": 0.7731168270111084, | |
| "learning_rate": 1.6854770717718587e-05, | |
| "loss": 0.0134, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.26034612490594433, | |
| "grad_norm": 0.006896435748785734, | |
| "learning_rate": 1.683754032950148e-05, | |
| "loss": 0.0004, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 0.26109857035364936, | |
| "grad_norm": 0.08711463212966919, | |
| "learning_rate": 1.6820271733708676e-05, | |
| "loss": 0.0019, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 0.2618510158013544, | |
| "grad_norm": 0.31446611881256104, | |
| "learning_rate": 1.6802965026835575e-05, | |
| "loss": 0.0066, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 0.2626034612490594, | |
| "grad_norm": 0.038149867206811905, | |
| "learning_rate": 1.6785620305590536e-05, | |
| "loss": 0.0012, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 0.2633559066967645, | |
| "grad_norm": 0.003243704093620181, | |
| "learning_rate": 1.676823766689434e-05, | |
| "loss": 0.0002, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.26410835214446954, | |
| "grad_norm": 0.0029318176675587893, | |
| "learning_rate": 1.6750817207879655e-05, | |
| "loss": 0.0001, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 0.26486079759217457, | |
| "grad_norm": 1.8430068492889404, | |
| "learning_rate": 1.673335902589047e-05, | |
| "loss": 0.0449, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 0.2656132430398796, | |
| "grad_norm": 0.00958191603422165, | |
| "learning_rate": 1.6715863218481573e-05, | |
| "loss": 0.0004, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 0.26636568848758463, | |
| "grad_norm": 2.806882619857788, | |
| "learning_rate": 1.6698329883418008e-05, | |
| "loss": 0.1586, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 0.2671181339352897, | |
| "grad_norm": 0.0034162893425673246, | |
| "learning_rate": 1.6680759118674512e-05, | |
| "loss": 0.0002, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.26787057938299474, | |
| "grad_norm": 0.008835663087666035, | |
| "learning_rate": 1.6663151022434978e-05, | |
| "loss": 0.0003, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 0.2686230248306998, | |
| "grad_norm": 0.0020498326048254967, | |
| "learning_rate": 1.6645505693091897e-05, | |
| "loss": 0.0001, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 0.2693754702784048, | |
| "grad_norm": 0.0011977710528299212, | |
| "learning_rate": 1.662782322924583e-05, | |
| "loss": 0.0001, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.27012791572610984, | |
| "grad_norm": 0.013555724173784256, | |
| "learning_rate": 1.661010372970483e-05, | |
| "loss": 0.0004, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 0.2708803611738149, | |
| "grad_norm": 0.0006222277879714966, | |
| "learning_rate": 1.6592347293483908e-05, | |
| "loss": 0.0, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.27163280662151995, | |
| "grad_norm": 0.008171453140676022, | |
| "learning_rate": 1.6574554019804474e-05, | |
| "loss": 0.0003, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 0.272385252069225, | |
| "grad_norm": 0.004504789598286152, | |
| "learning_rate": 1.655672400809378e-05, | |
| "loss": 0.0001, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 0.27313769751693, | |
| "grad_norm": 0.00137099449057132, | |
| "learning_rate": 1.6538857357984358e-05, | |
| "loss": 0.0001, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 0.27389014296463504, | |
| "grad_norm": 0.06337831914424896, | |
| "learning_rate": 1.6520954169313498e-05, | |
| "loss": 0.0016, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 0.2746425884123401, | |
| "grad_norm": 0.0002586379705462605, | |
| "learning_rate": 1.6503014542122637e-05, | |
| "loss": 0.0, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.27539503386004516, | |
| "grad_norm": 0.004504516255110502, | |
| "learning_rate": 1.6485038576656842e-05, | |
| "loss": 0.0001, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 0.2761474793077502, | |
| "grad_norm": 0.31715670228004456, | |
| "learning_rate": 1.646702637336423e-05, | |
| "loss": 0.0032, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 0.2768999247554552, | |
| "grad_norm": 0.0008062532288022339, | |
| "learning_rate": 1.6448978032895417e-05, | |
| "loss": 0.0, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 0.27765237020316025, | |
| "grad_norm": 0.0017397012561559677, | |
| "learning_rate": 1.6430893656102942e-05, | |
| "loss": 0.0001, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 0.27840481565086533, | |
| "grad_norm": 0.3345002830028534, | |
| "learning_rate": 1.6412773344040717e-05, | |
| "loss": 0.0042, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.27915726109857036, | |
| "grad_norm": 9.196252358378842e-05, | |
| "learning_rate": 1.6394617197963462e-05, | |
| "loss": 0.0, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 0.2799097065462754, | |
| "grad_norm": 0.2911829352378845, | |
| "learning_rate": 1.6376425319326125e-05, | |
| "loss": 0.1043, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 0.2806621519939804, | |
| "grad_norm": 0.0008261505281552672, | |
| "learning_rate": 1.635819780978333e-05, | |
| "loss": 0.0, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 0.28141459744168545, | |
| "grad_norm": 0.013018965721130371, | |
| "learning_rate": 1.6339934771188796e-05, | |
| "loss": 0.0003, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 0.28216704288939054, | |
| "grad_norm": 0.5947379469871521, | |
| "learning_rate": 1.6321636305594784e-05, | |
| "loss": 0.0068, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.28291948833709557, | |
| "grad_norm": 0.005715230479836464, | |
| "learning_rate": 1.630330251525152e-05, | |
| "loss": 0.0002, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 0.2836719337848006, | |
| "grad_norm": 0.07143538445234299, | |
| "learning_rate": 1.6284933502606614e-05, | |
| "loss": 0.0009, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 0.28442437923250563, | |
| "grad_norm": 4.414242267608643, | |
| "learning_rate": 1.6266529370304492e-05, | |
| "loss": 0.0671, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 0.28517682468021066, | |
| "grad_norm": 0.0007347184582613409, | |
| "learning_rate": 1.624809022118584e-05, | |
| "loss": 0.0, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 0.28592927012791575, | |
| "grad_norm": 0.00025847938377410173, | |
| "learning_rate": 1.6229616158286997e-05, | |
| "loss": 0.0, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.2866817155756208, | |
| "grad_norm": 12.279823303222656, | |
| "learning_rate": 1.6211107284839417e-05, | |
| "loss": 0.1936, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 0.2874341610233258, | |
| "grad_norm": 0.0016370975645259023, | |
| "learning_rate": 1.6192563704269048e-05, | |
| "loss": 0.0001, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 0.28818660647103084, | |
| "grad_norm": 5.603551835520193e-05, | |
| "learning_rate": 1.6173985520195805e-05, | |
| "loss": 0.0, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 0.28893905191873587, | |
| "grad_norm": 0.017844753339886665, | |
| "learning_rate": 1.6155372836432944e-05, | |
| "loss": 0.0007, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 0.28969149736644095, | |
| "grad_norm": 0.0008964858716353774, | |
| "learning_rate": 1.6136725756986514e-05, | |
| "loss": 0.0, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.290443942814146, | |
| "grad_norm": 0.12251462787389755, | |
| "learning_rate": 1.6118044386054755e-05, | |
| "loss": 0.0021, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 0.291196388261851, | |
| "grad_norm": 8.942232131958008, | |
| "learning_rate": 1.609932882802753e-05, | |
| "loss": 0.1369, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 0.29194883370955604, | |
| "grad_norm": 0.005296614021062851, | |
| "learning_rate": 1.6080579187485738e-05, | |
| "loss": 0.0001, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 0.2927012791572611, | |
| "grad_norm": 0.00021417946845758706, | |
| "learning_rate": 1.6061795569200725e-05, | |
| "loss": 0.0, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 0.29345372460496616, | |
| "grad_norm": 0.0008177552372217178, | |
| "learning_rate": 1.60429780781337e-05, | |
| "loss": 0.0, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.2942061700526712, | |
| "grad_norm": 2.832350015640259, | |
| "learning_rate": 1.6024126819435156e-05, | |
| "loss": 0.0632, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 0.2949586155003762, | |
| "grad_norm": 0.007019749376922846, | |
| "learning_rate": 1.6005241898444275e-05, | |
| "loss": 0.0002, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 0.29571106094808125, | |
| "grad_norm": 0.49707838892936707, | |
| "learning_rate": 1.5986323420688335e-05, | |
| "loss": 0.093, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 0.2964635063957863, | |
| "grad_norm": 5.089029788970947, | |
| "learning_rate": 1.5967371491882136e-05, | |
| "loss": 0.0233, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 0.29721595184349137, | |
| "grad_norm": 0.0013898630859330297, | |
| "learning_rate": 1.5948386217927384e-05, | |
| "loss": 0.0001, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.2979683972911964, | |
| "grad_norm": 0.0012057372368872166, | |
| "learning_rate": 1.592936770491214e-05, | |
| "loss": 0.0001, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 0.2987208427389014, | |
| "grad_norm": 0.0038578074891120195, | |
| "learning_rate": 1.591031605911017e-05, | |
| "loss": 0.0001, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 0.29947328818660646, | |
| "grad_norm": 8.180018630810082e-05, | |
| "learning_rate": 1.5891231386980415e-05, | |
| "loss": 0.0, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 0.3002257336343115, | |
| "grad_norm": 0.0012178582837805152, | |
| "learning_rate": 1.5872113795166337e-05, | |
| "loss": 0.0, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 0.3009781790820166, | |
| "grad_norm": 7.556064520031214e-05, | |
| "learning_rate": 1.585296339049537e-05, | |
| "loss": 0.0, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.3017306245297216, | |
| "grad_norm": 0.0032481662929058075, | |
| "learning_rate": 1.5833780279978293e-05, | |
| "loss": 0.0001, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 0.30248306997742663, | |
| "grad_norm": 4.7872754294076e-06, | |
| "learning_rate": 1.5814564570808643e-05, | |
| "loss": 0.0, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 0.30323551542513166, | |
| "grad_norm": 0.014581550844013691, | |
| "learning_rate": 1.5795316370362122e-05, | |
| "loss": 0.0003, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 0.3039879608728367, | |
| "grad_norm": 0.3884536623954773, | |
| "learning_rate": 1.5776035786195983e-05, | |
| "loss": 0.1174, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 0.3047404063205418, | |
| "grad_norm": 0.061029549688100815, | |
| "learning_rate": 1.575672292604844e-05, | |
| "loss": 0.0002, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.3054928517682468, | |
| "grad_norm": 0.0007864750223234296, | |
| "learning_rate": 1.5737377897838065e-05, | |
| "loss": 0.0, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 0.30624529721595184, | |
| "grad_norm": 0.0008468715823255479, | |
| "learning_rate": 1.5718000809663173e-05, | |
| "loss": 0.0, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 0.30699774266365687, | |
| "grad_norm": 6.305576243903488e-05, | |
| "learning_rate": 1.569859176980124e-05, | |
| "loss": 0.0, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 0.3077501881113619, | |
| "grad_norm": 0.6824759840965271, | |
| "learning_rate": 1.5679150886708273e-05, | |
| "loss": 0.1164, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 0.308502633559067, | |
| "grad_norm": 0.01927897147834301, | |
| "learning_rate": 1.565967826901822e-05, | |
| "loss": 0.0002, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.309255079006772, | |
| "grad_norm": 0.001680307206697762, | |
| "learning_rate": 1.564017402554237e-05, | |
| "loss": 0.0001, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 0.31000752445447705, | |
| "grad_norm": 0.036916520446538925, | |
| "learning_rate": 1.5620638265268718e-05, | |
| "loss": 0.0005, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 0.3107599699021821, | |
| "grad_norm": 4.316399097442627, | |
| "learning_rate": 1.560107109736138e-05, | |
| "loss": 0.4106, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 0.3115124153498871, | |
| "grad_norm": 0.0003387642209418118, | |
| "learning_rate": 1.5581472631159977e-05, | |
| "loss": 0.0, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 0.3122648607975922, | |
| "grad_norm": 0.5591127872467041, | |
| "learning_rate": 1.5561842976179013e-05, | |
| "loss": 0.1023, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.3130173062452972, | |
| "grad_norm": 6.071132659912109, | |
| "learning_rate": 1.5542182242107284e-05, | |
| "loss": 0.0244, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 0.31376975169300225, | |
| "grad_norm": 0.0023193645756691694, | |
| "learning_rate": 1.5522490538807248e-05, | |
| "loss": 0.0001, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 0.3145221971407073, | |
| "grad_norm": 0.0017903498373925686, | |
| "learning_rate": 1.5502767976314413e-05, | |
| "loss": 0.0001, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 0.3152746425884123, | |
| "grad_norm": 10.95837116241455, | |
| "learning_rate": 1.5483014664836732e-05, | |
| "loss": 0.6184, | |
| "step": 419 | |
| }, | |
| { | |
| "epoch": 0.3160270880361174, | |
| "grad_norm": 0.002409271663054824, | |
| "learning_rate": 1.546323071475397e-05, | |
| "loss": 0.0001, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.31677953348382243, | |
| "grad_norm": 7.29747480363585e-05, | |
| "learning_rate": 1.544341623661711e-05, | |
| "loss": 0.0, | |
| "step": 421 | |
| }, | |
| { | |
| "epoch": 0.31753197893152746, | |
| "grad_norm": 2.6204562187194824, | |
| "learning_rate": 1.5423571341147724e-05, | |
| "loss": 0.1352, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 0.3182844243792325, | |
| "grad_norm": 0.0007865215302444994, | |
| "learning_rate": 1.5403696139237338e-05, | |
| "loss": 0.0, | |
| "step": 423 | |
| }, | |
| { | |
| "epoch": 0.3190368698269376, | |
| "grad_norm": 3.1011784076690674, | |
| "learning_rate": 1.538379074194684e-05, | |
| "loss": 0.4246, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 0.3197893152746426, | |
| "grad_norm": 0.000731561507564038, | |
| "learning_rate": 1.5363855260505848e-05, | |
| "loss": 0.0, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.32054176072234764, | |
| "grad_norm": 6.430411338806152, | |
| "learning_rate": 1.534388980631208e-05, | |
| "loss": 0.2966, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 0.32129420617005267, | |
| "grad_norm": 0.010142396204173565, | |
| "learning_rate": 1.5323894490930743e-05, | |
| "loss": 0.0004, | |
| "step": 427 | |
| }, | |
| { | |
| "epoch": 0.3220466516177577, | |
| "grad_norm": 0.5542206764221191, | |
| "learning_rate": 1.5303869426093906e-05, | |
| "loss": 0.0149, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 0.3227990970654628, | |
| "grad_norm": 0.35877907276153564, | |
| "learning_rate": 1.5283814723699877e-05, | |
| "loss": 0.0831, | |
| "step": 429 | |
| }, | |
| { | |
| "epoch": 0.3235515425131678, | |
| "grad_norm": 0.082036592066288, | |
| "learning_rate": 1.5263730495812568e-05, | |
| "loss": 0.0032, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.32430398796087284, | |
| "grad_norm": 0.017158055678009987, | |
| "learning_rate": 1.5243616854660894e-05, | |
| "loss": 0.001, | |
| "step": 431 | |
| }, | |
| { | |
| "epoch": 0.32505643340857787, | |
| "grad_norm": 0.00297352927736938, | |
| "learning_rate": 1.522347391263811e-05, | |
| "loss": 0.0001, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 0.3258088788562829, | |
| "grad_norm": 0.003190380521118641, | |
| "learning_rate": 1.5203301782301212e-05, | |
| "loss": 0.0001, | |
| "step": 433 | |
| }, | |
| { | |
| "epoch": 0.326561324303988, | |
| "grad_norm": 0.0062565989792346954, | |
| "learning_rate": 1.5183100576370291e-05, | |
| "loss": 0.0003, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 0.327313769751693, | |
| "grad_norm": 2.2504420280456543, | |
| "learning_rate": 1.5162870407727922e-05, | |
| "loss": 0.1951, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.32806621519939805, | |
| "grad_norm": 1.0565892457962036, | |
| "learning_rate": 1.5142611389418505e-05, | |
| "loss": 0.0199, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 0.3288186606471031, | |
| "grad_norm": 1.0833739042282104, | |
| "learning_rate": 1.512232363464766e-05, | |
| "loss": 0.0284, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 0.3295711060948081, | |
| "grad_norm": 0.32872238755226135, | |
| "learning_rate": 1.5102007256781583e-05, | |
| "loss": 0.0861, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 0.3303235515425132, | |
| "grad_norm": 0.08302167803049088, | |
| "learning_rate": 1.5081662369346412e-05, | |
| "loss": 0.0027, | |
| "step": 439 | |
| }, | |
| { | |
| "epoch": 0.3310759969902182, | |
| "grad_norm": 0.01665945164859295, | |
| "learning_rate": 1.5061289086027593e-05, | |
| "loss": 0.0009, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.33182844243792325, | |
| "grad_norm": 0.00258431863039732, | |
| "learning_rate": 1.5040887520669245e-05, | |
| "loss": 0.0001, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 0.3325808878856283, | |
| "grad_norm": 0.11993154883384705, | |
| "learning_rate": 1.502045778727353e-05, | |
| "loss": 0.0023, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 0.3333333333333333, | |
| "grad_norm": 10.732891082763672, | |
| "learning_rate": 1.5000000000000002e-05, | |
| "loss": 0.0681, | |
| "step": 443 | |
| }, | |
| { | |
| "epoch": 0.3340857787810384, | |
| "grad_norm": 0.0034522817004472017, | |
| "learning_rate": 1.497951427316498e-05, | |
| "loss": 0.0002, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 0.33483822422874343, | |
| "grad_norm": 0.0015809908509254456, | |
| "learning_rate": 1.495900072124092e-05, | |
| "loss": 0.0001, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.33559066967644846, | |
| "grad_norm": 0.006714050658047199, | |
| "learning_rate": 1.4938459458855739e-05, | |
| "loss": 0.0004, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 0.3363431151241535, | |
| "grad_norm": 0.010480109602212906, | |
| "learning_rate": 1.4917890600792215e-05, | |
| "loss": 0.0006, | |
| "step": 447 | |
| }, | |
| { | |
| "epoch": 0.3370955605718585, | |
| "grad_norm": 0.06001771241426468, | |
| "learning_rate": 1.4897294261987325e-05, | |
| "loss": 0.0021, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 0.3378480060195636, | |
| "grad_norm": 0.0026993490755558014, | |
| "learning_rate": 1.4876670557531598e-05, | |
| "loss": 0.0001, | |
| "step": 449 | |
| }, | |
| { | |
| "epoch": 0.33860045146726864, | |
| "grad_norm": 1.938813328742981, | |
| "learning_rate": 1.485601960266849e-05, | |
| "loss": 0.0378, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.33935289691497367, | |
| "grad_norm": 0.004080132115632296, | |
| "learning_rate": 1.4835341512793727e-05, | |
| "loss": 0.0002, | |
| "step": 451 | |
| }, | |
| { | |
| "epoch": 0.3401053423626787, | |
| "grad_norm": 3.724914073944092, | |
| "learning_rate": 1.4814636403454656e-05, | |
| "loss": 0.1663, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 0.34085778781038373, | |
| "grad_norm": 0.6385138034820557, | |
| "learning_rate": 1.4793904390349618e-05, | |
| "loss": 0.1188, | |
| "step": 453 | |
| }, | |
| { | |
| "epoch": 0.3416102332580888, | |
| "grad_norm": 4.093441963195801, | |
| "learning_rate": 1.477314558932728e-05, | |
| "loss": 0.0387, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 0.34236267870579384, | |
| "grad_norm": 0.028323644772171974, | |
| "learning_rate": 1.4752360116386002e-05, | |
| "loss": 0.0012, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.3431151241534989, | |
| "grad_norm": 0.14701040089130402, | |
| "learning_rate": 1.4731548087673186e-05, | |
| "loss": 0.0046, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 0.3438675696012039, | |
| "grad_norm": 0.22901006042957306, | |
| "learning_rate": 1.4710709619484623e-05, | |
| "loss": 0.0061, | |
| "step": 457 | |
| }, | |
| { | |
| "epoch": 0.34462001504890893, | |
| "grad_norm": 0.007301859557628632, | |
| "learning_rate": 1.4689844828263846e-05, | |
| "loss": 0.0003, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 0.345372460496614, | |
| "grad_norm": 0.005133870989084244, | |
| "learning_rate": 1.4668953830601473e-05, | |
| "loss": 0.0002, | |
| "step": 459 | |
| }, | |
| { | |
| "epoch": 0.34612490594431905, | |
| "grad_norm": 0.003975029569119215, | |
| "learning_rate": 1.4648036743234573e-05, | |
| "loss": 0.0002, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.3468773513920241, | |
| "grad_norm": 0.012133643962442875, | |
| "learning_rate": 1.4627093683045997e-05, | |
| "loss": 0.0006, | |
| "step": 461 | |
| }, | |
| { | |
| "epoch": 0.3476297968397291, | |
| "grad_norm": 0.00555196451023221, | |
| "learning_rate": 1.4606124767063721e-05, | |
| "loss": 0.0003, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 0.34838224228743414, | |
| "grad_norm": 0.001998197054490447, | |
| "learning_rate": 1.4585130112460214e-05, | |
| "loss": 0.0001, | |
| "step": 463 | |
| }, | |
| { | |
| "epoch": 0.3491346877351392, | |
| "grad_norm": 0.4198460578918457, | |
| "learning_rate": 1.4564109836551763e-05, | |
| "loss": 0.0066, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 0.34988713318284426, | |
| "grad_norm": 0.5744123458862305, | |
| "learning_rate": 1.4543064056797826e-05, | |
| "loss": 0.0147, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.3506395786305493, | |
| "grad_norm": 0.0055709234438836575, | |
| "learning_rate": 1.4521992890800379e-05, | |
| "loss": 0.0002, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 0.3513920240782543, | |
| "grad_norm": 0.0019197987858206034, | |
| "learning_rate": 1.4500896456303241e-05, | |
| "loss": 0.0001, | |
| "step": 467 | |
| }, | |
| { | |
| "epoch": 0.35214446952595935, | |
| "grad_norm": 0.8345107436180115, | |
| "learning_rate": 1.4479774871191447e-05, | |
| "loss": 0.1345, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 0.35289691497366443, | |
| "grad_norm": 0.6766433715820312, | |
| "learning_rate": 1.4458628253490555e-05, | |
| "loss": 0.179, | |
| "step": 469 | |
| }, | |
| { | |
| "epoch": 0.35364936042136946, | |
| "grad_norm": 0.014497430063784122, | |
| "learning_rate": 1.4437456721366013e-05, | |
| "loss": 0.0007, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.3544018058690745, | |
| "grad_norm": 0.11778023838996887, | |
| "learning_rate": 1.4416260393122487e-05, | |
| "loss": 0.0046, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 0.3551542513167795, | |
| "grad_norm": 0.47652357816696167, | |
| "learning_rate": 1.4395039387203197e-05, | |
| "loss": 0.0069, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 0.35590669676448455, | |
| "grad_norm": 0.3849281668663025, | |
| "learning_rate": 1.4373793822189266e-05, | |
| "loss": 0.11, | |
| "step": 473 | |
| }, | |
| { | |
| "epoch": 0.35665914221218964, | |
| "grad_norm": 2.7059948444366455, | |
| "learning_rate": 1.4352523816799046e-05, | |
| "loss": 0.1874, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 0.35741158765989467, | |
| "grad_norm": 0.7493427395820618, | |
| "learning_rate": 1.4331229489887463e-05, | |
| "loss": 0.0055, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.3581640331075997, | |
| "grad_norm": 0.002171468921005726, | |
| "learning_rate": 1.430991096044535e-05, | |
| "loss": 0.0001, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 0.35891647855530473, | |
| "grad_norm": 12.364990234375, | |
| "learning_rate": 1.4288568347598777e-05, | |
| "loss": 0.0784, | |
| "step": 477 | |
| }, | |
| { | |
| "epoch": 0.35966892400300976, | |
| "grad_norm": 0.022090526297688484, | |
| "learning_rate": 1.4267201770608392e-05, | |
| "loss": 0.0008, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 0.36042136945071485, | |
| "grad_norm": 1.5892497301101685, | |
| "learning_rate": 1.4245811348868753e-05, | |
| "loss": 0.0918, | |
| "step": 479 | |
| }, | |
| { | |
| "epoch": 0.3611738148984199, | |
| "grad_norm": 0.005540026817470789, | |
| "learning_rate": 1.4224397201907664e-05, | |
| "loss": 0.0003, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.3619262603461249, | |
| "grad_norm": 0.06051875278353691, | |
| "learning_rate": 1.42029594493855e-05, | |
| "loss": 0.0014, | |
| "step": 481 | |
| }, | |
| { | |
| "epoch": 0.36267870579382994, | |
| "grad_norm": 0.011512890458106995, | |
| "learning_rate": 1.418149821109454e-05, | |
| "loss": 0.0005, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 0.36343115124153497, | |
| "grad_norm": 0.002135386224836111, | |
| "learning_rate": 1.4160013606958303e-05, | |
| "loss": 0.0001, | |
| "step": 483 | |
| }, | |
| { | |
| "epoch": 0.36418359668924005, | |
| "grad_norm": 0.018821068108081818, | |
| "learning_rate": 1.4138505757030869e-05, | |
| "loss": 0.0004, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 0.3649360421369451, | |
| "grad_norm": 0.0008346544927917421, | |
| "learning_rate": 1.411697478149622e-05, | |
| "loss": 0.0, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.3656884875846501, | |
| "grad_norm": 0.18350642919540405, | |
| "learning_rate": 1.409542080066756e-05, | |
| "loss": 0.004, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 0.36644093303235514, | |
| "grad_norm": 2.6405723094940186, | |
| "learning_rate": 1.4073843934986644e-05, | |
| "loss": 0.3307, | |
| "step": 487 | |
| }, | |
| { | |
| "epoch": 0.3671933784800602, | |
| "grad_norm": 0.0034803429152816534, | |
| "learning_rate": 1.4052244305023101e-05, | |
| "loss": 0.0001, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 0.36794582392776526, | |
| "grad_norm": 0.006721539422869682, | |
| "learning_rate": 1.403062203147377e-05, | |
| "loss": 0.0003, | |
| "step": 489 | |
| }, | |
| { | |
| "epoch": 0.3686982693754703, | |
| "grad_norm": 0.6599304676055908, | |
| "learning_rate": 1.4008977235162024e-05, | |
| "loss": 0.0107, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.3694507148231753, | |
| "grad_norm": 0.05510720983147621, | |
| "learning_rate": 1.3987310037037081e-05, | |
| "loss": 0.0022, | |
| "step": 491 | |
| }, | |
| { | |
| "epoch": 0.37020316027088035, | |
| "grad_norm": 0.0020602026488631964, | |
| "learning_rate": 1.3965620558173345e-05, | |
| "loss": 0.0001, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 0.3709556057185854, | |
| "grad_norm": 3.3849992752075195, | |
| "learning_rate": 1.3943908919769724e-05, | |
| "loss": 0.0947, | |
| "step": 493 | |
| }, | |
| { | |
| "epoch": 0.37170805116629047, | |
| "grad_norm": 0.0011134468950331211, | |
| "learning_rate": 1.3922175243148948e-05, | |
| "loss": 0.0001, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 0.3724604966139955, | |
| "grad_norm": 0.4826012849807739, | |
| "learning_rate": 1.3900419649756895e-05, | |
| "loss": 0.0075, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.3732129420617005, | |
| "grad_norm": 1.0771576166152954, | |
| "learning_rate": 1.3878642261161916e-05, | |
| "loss": 0.015, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 0.37396538750940556, | |
| "grad_norm": 0.14827294647693634, | |
| "learning_rate": 1.3856843199054144e-05, | |
| "loss": 0.0036, | |
| "step": 497 | |
| }, | |
| { | |
| "epoch": 0.3747178329571106, | |
| "grad_norm": 0.0043605901300907135, | |
| "learning_rate": 1.3835022585244829e-05, | |
| "loss": 0.0002, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 0.37547027840481567, | |
| "grad_norm": 0.05336622893810272, | |
| "learning_rate": 1.3813180541665646e-05, | |
| "loss": 0.0025, | |
| "step": 499 | |
| }, | |
| { | |
| "epoch": 0.3762227238525207, | |
| "grad_norm": 0.00369238737039268, | |
| "learning_rate": 1.3791317190368018e-05, | |
| "loss": 0.0002, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.37697516930022573, | |
| "grad_norm": 1.3321623802185059, | |
| "learning_rate": 1.3769432653522436e-05, | |
| "loss": 0.0112, | |
| "step": 501 | |
| }, | |
| { | |
| "epoch": 0.37772761474793076, | |
| "grad_norm": 0.0012655869359150529, | |
| "learning_rate": 1.3747527053417776e-05, | |
| "loss": 0.0001, | |
| "step": 502 | |
| }, | |
| { | |
| "epoch": 0.3784800601956358, | |
| "grad_norm": 0.0007887445390224457, | |
| "learning_rate": 1.3725600512460606e-05, | |
| "loss": 0.0, | |
| "step": 503 | |
| }, | |
| { | |
| "epoch": 0.3792325056433409, | |
| "grad_norm": 0.1122029721736908, | |
| "learning_rate": 1.3703653153174513e-05, | |
| "loss": 0.0007, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 0.3799849510910459, | |
| "grad_norm": 2.8246965408325195, | |
| "learning_rate": 1.3681685098199418e-05, | |
| "loss": 0.2841, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 0.38073739653875094, | |
| "grad_norm": 0.006240316201001406, | |
| "learning_rate": 1.3659696470290888e-05, | |
| "loss": 0.0002, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 0.38148984198645597, | |
| "grad_norm": 0.15288379788398743, | |
| "learning_rate": 1.3637687392319443e-05, | |
| "loss": 0.0032, | |
| "step": 507 | |
| }, | |
| { | |
| "epoch": 0.382242287434161, | |
| "grad_norm": 0.005582885816693306, | |
| "learning_rate": 1.3615657987269882e-05, | |
| "loss": 0.0002, | |
| "step": 508 | |
| }, | |
| { | |
| "epoch": 0.3829947328818661, | |
| "grad_norm": 0.00010864822252187878, | |
| "learning_rate": 1.3593608378240587e-05, | |
| "loss": 0.0, | |
| "step": 509 | |
| }, | |
| { | |
| "epoch": 0.3837471783295711, | |
| "grad_norm": 5.737289905548096, | |
| "learning_rate": 1.3571538688442843e-05, | |
| "loss": 0.1844, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.38449962377727614, | |
| "grad_norm": 0.002173966495320201, | |
| "learning_rate": 1.3549449041200138e-05, | |
| "loss": 0.0001, | |
| "step": 511 | |
| }, | |
| { | |
| "epoch": 0.3852520692249812, | |
| "grad_norm": 0.8293859958648682, | |
| "learning_rate": 1.3527339559947483e-05, | |
| "loss": 0.1965, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 0.3860045146726862, | |
| "grad_norm": 0.0026998259127140045, | |
| "learning_rate": 1.3505210368230723e-05, | |
| "loss": 0.0001, | |
| "step": 513 | |
| }, | |
| { | |
| "epoch": 0.3867569601203913, | |
| "grad_norm": 0.00044145798892714083, | |
| "learning_rate": 1.3483061589705839e-05, | |
| "loss": 0.0, | |
| "step": 514 | |
| }, | |
| { | |
| "epoch": 0.3875094055680963, | |
| "grad_norm": 0.004763697739690542, | |
| "learning_rate": 1.3460893348138262e-05, | |
| "loss": 0.0001, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 0.38826185101580135, | |
| "grad_norm": 9.023599704960361e-05, | |
| "learning_rate": 1.3438705767402185e-05, | |
| "loss": 0.0, | |
| "step": 516 | |
| }, | |
| { | |
| "epoch": 0.3890142964635064, | |
| "grad_norm": 0.0630866289138794, | |
| "learning_rate": 1.341649897147986e-05, | |
| "loss": 0.0006, | |
| "step": 517 | |
| }, | |
| { | |
| "epoch": 0.3897667419112114, | |
| "grad_norm": 0.0039035833906382322, | |
| "learning_rate": 1.3394273084460916e-05, | |
| "loss": 0.0002, | |
| "step": 518 | |
| }, | |
| { | |
| "epoch": 0.3905191873589165, | |
| "grad_norm": 3.1299986839294434, | |
| "learning_rate": 1.3372028230541658e-05, | |
| "loss": 0.4068, | |
| "step": 519 | |
| }, | |
| { | |
| "epoch": 0.3912716328066215, | |
| "grad_norm": 0.0035276322159916162, | |
| "learning_rate": 1.3349764534024385e-05, | |
| "loss": 0.0002, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.39202407825432656, | |
| "grad_norm": 0.5055978894233704, | |
| "learning_rate": 1.3327482119316674e-05, | |
| "loss": 0.0223, | |
| "step": 521 | |
| }, | |
| { | |
| "epoch": 0.3927765237020316, | |
| "grad_norm": 0.0033338728826493025, | |
| "learning_rate": 1.330518111093071e-05, | |
| "loss": 0.0002, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 0.3935289691497366, | |
| "grad_norm": 0.06779402494430542, | |
| "learning_rate": 1.3282861633482566e-05, | |
| "loss": 0.003, | |
| "step": 523 | |
| }, | |
| { | |
| "epoch": 0.3942814145974417, | |
| "grad_norm": 0.6969091892242432, | |
| "learning_rate": 1.3260523811691527e-05, | |
| "loss": 0.0111, | |
| "step": 524 | |
| }, | |
| { | |
| "epoch": 0.39503386004514673, | |
| "grad_norm": 0.3388632535934448, | |
| "learning_rate": 1.3238167770379384e-05, | |
| "loss": 0.1329, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.39578630549285176, | |
| "grad_norm": 0.15506567060947418, | |
| "learning_rate": 1.3215793634469733e-05, | |
| "loss": 0.0035, | |
| "step": 526 | |
| }, | |
| { | |
| "epoch": 0.3965387509405568, | |
| "grad_norm": 3.37923002243042, | |
| "learning_rate": 1.3193401528987286e-05, | |
| "loss": 0.3774, | |
| "step": 527 | |
| }, | |
| { | |
| "epoch": 0.3972911963882618, | |
| "grad_norm": 0.6373834609985352, | |
| "learning_rate": 1.3170991579057163e-05, | |
| "loss": 0.0922, | |
| "step": 528 | |
| }, | |
| { | |
| "epoch": 0.3980436418359669, | |
| "grad_norm": 0.23215818405151367, | |
| "learning_rate": 1.3148563909904195e-05, | |
| "loss": 0.007, | |
| "step": 529 | |
| }, | |
| { | |
| "epoch": 0.39879608728367194, | |
| "grad_norm": 0.000551075441762805, | |
| "learning_rate": 1.3126118646852235e-05, | |
| "loss": 0.0, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.39954853273137697, | |
| "grad_norm": 2.5601372718811035, | |
| "learning_rate": 1.3103655915323444e-05, | |
| "loss": 0.0924, | |
| "step": 531 | |
| }, | |
| { | |
| "epoch": 0.400300978179082, | |
| "grad_norm": 0.029645578935742378, | |
| "learning_rate": 1.3081175840837595e-05, | |
| "loss": 0.0007, | |
| "step": 532 | |
| }, | |
| { | |
| "epoch": 0.40105342362678703, | |
| "grad_norm": 0.0015663238009437919, | |
| "learning_rate": 1.3058678549011371e-05, | |
| "loss": 0.0001, | |
| "step": 533 | |
| }, | |
| { | |
| "epoch": 0.4018058690744921, | |
| "grad_norm": 7.516592025756836, | |
| "learning_rate": 1.3036164165557667e-05, | |
| "loss": 0.1748, | |
| "step": 534 | |
| }, | |
| { | |
| "epoch": 0.40255831452219715, | |
| "grad_norm": 0.02340114861726761, | |
| "learning_rate": 1.3013632816284885e-05, | |
| "loss": 0.001, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 0.4033107599699022, | |
| "grad_norm": 0.04894443601369858, | |
| "learning_rate": 1.2991084627096226e-05, | |
| "loss": 0.0016, | |
| "step": 536 | |
| }, | |
| { | |
| "epoch": 0.4040632054176072, | |
| "grad_norm": 0.053800590336322784, | |
| "learning_rate": 1.2968519723988994e-05, | |
| "loss": 0.0021, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 0.40481565086531224, | |
| "grad_norm": 0.01409213524311781, | |
| "learning_rate": 1.2945938233053892e-05, | |
| "loss": 0.0005, | |
| "step": 538 | |
| }, | |
| { | |
| "epoch": 0.4055680963130173, | |
| "grad_norm": 0.4299611747264862, | |
| "learning_rate": 1.2923340280474306e-05, | |
| "loss": 0.0091, | |
| "step": 539 | |
| }, | |
| { | |
| "epoch": 0.40632054176072235, | |
| "grad_norm": 3.0297110080718994, | |
| "learning_rate": 1.2900725992525618e-05, | |
| "loss": 0.1769, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.4070729872084274, | |
| "grad_norm": 0.012766638770699501, | |
| "learning_rate": 1.2878095495574484e-05, | |
| "loss": 0.0006, | |
| "step": 541 | |
| }, | |
| { | |
| "epoch": 0.4078254326561324, | |
| "grad_norm": 0.019243692979216576, | |
| "learning_rate": 1.285544891607813e-05, | |
| "loss": 0.0007, | |
| "step": 542 | |
| }, | |
| { | |
| "epoch": 0.40857787810383744, | |
| "grad_norm": 0.4208061099052429, | |
| "learning_rate": 1.2832786380583664e-05, | |
| "loss": 0.0103, | |
| "step": 543 | |
| }, | |
| { | |
| "epoch": 0.40933032355154253, | |
| "grad_norm": 0.07195150852203369, | |
| "learning_rate": 1.2810108015727345e-05, | |
| "loss": 0.0022, | |
| "step": 544 | |
| }, | |
| { | |
| "epoch": 0.41008276899924756, | |
| "grad_norm": 1.117520809173584, | |
| "learning_rate": 1.2787413948233885e-05, | |
| "loss": 0.0159, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 0.4108352144469526, | |
| "grad_norm": 0.15800736844539642, | |
| "learning_rate": 1.2764704304915743e-05, | |
| "loss": 0.0048, | |
| "step": 546 | |
| }, | |
| { | |
| "epoch": 0.4115876598946576, | |
| "grad_norm": 0.026513785123825073, | |
| "learning_rate": 1.2741979212672418e-05, | |
| "loss": 0.0011, | |
| "step": 547 | |
| }, | |
| { | |
| "epoch": 0.4123401053423627, | |
| "grad_norm": 0.27647465467453003, | |
| "learning_rate": 1.2719238798489725e-05, | |
| "loss": 0.0094, | |
| "step": 548 | |
| }, | |
| { | |
| "epoch": 0.41309255079006774, | |
| "grad_norm": 0.003962705843150616, | |
| "learning_rate": 1.2696483189439113e-05, | |
| "loss": 0.0002, | |
| "step": 549 | |
| }, | |
| { | |
| "epoch": 0.41384499623777277, | |
| "grad_norm": 0.006578361615538597, | |
| "learning_rate": 1.2673712512676923e-05, | |
| "loss": 0.0002, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.4145974416854778, | |
| "grad_norm": 0.4227232038974762, | |
| "learning_rate": 1.2650926895443705e-05, | |
| "loss": 0.0132, | |
| "step": 551 | |
| }, | |
| { | |
| "epoch": 0.4153498871331828, | |
| "grad_norm": 0.1907581388950348, | |
| "learning_rate": 1.2628126465063483e-05, | |
| "loss": 0.0062, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 0.4161023325808879, | |
| "grad_norm": 0.02955441363155842, | |
| "learning_rate": 1.2605311348943066e-05, | |
| "loss": 0.0006, | |
| "step": 553 | |
| }, | |
| { | |
| "epoch": 0.41685477802859294, | |
| "grad_norm": 0.582580029964447, | |
| "learning_rate": 1.2582481674571325e-05, | |
| "loss": 0.0115, | |
| "step": 554 | |
| }, | |
| { | |
| "epoch": 0.417607223476298, | |
| "grad_norm": 0.0009048896026797593, | |
| "learning_rate": 1.2559637569518472e-05, | |
| "loss": 0.0, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 0.418359668924003, | |
| "grad_norm": 0.03770684078335762, | |
| "learning_rate": 1.2536779161435368e-05, | |
| "loss": 0.0006, | |
| "step": 556 | |
| }, | |
| { | |
| "epoch": 0.41911211437170803, | |
| "grad_norm": 0.25430363416671753, | |
| "learning_rate": 1.251390657805279e-05, | |
| "loss": 0.1052, | |
| "step": 557 | |
| }, | |
| { | |
| "epoch": 0.4198645598194131, | |
| "grad_norm": 0.8613071441650391, | |
| "learning_rate": 1.2491019947180727e-05, | |
| "loss": 0.0129, | |
| "step": 558 | |
| }, | |
| { | |
| "epoch": 0.42061700526711815, | |
| "grad_norm": 0.00920679047703743, | |
| "learning_rate": 1.2468119396707668e-05, | |
| "loss": 0.0003, | |
| "step": 559 | |
| }, | |
| { | |
| "epoch": 0.4213694507148232, | |
| "grad_norm": 0.49765467643737793, | |
| "learning_rate": 1.2445205054599879e-05, | |
| "loss": 0.0065, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.4221218961625282, | |
| "grad_norm": 0.006361035164445639, | |
| "learning_rate": 1.2422277048900694e-05, | |
| "loss": 0.0003, | |
| "step": 561 | |
| }, | |
| { | |
| "epoch": 0.42287434161023324, | |
| "grad_norm": 0.3172746002674103, | |
| "learning_rate": 1.23993355077298e-05, | |
| "loss": 0.2152, | |
| "step": 562 | |
| }, | |
| { | |
| "epoch": 0.4236267870579383, | |
| "grad_norm": 0.010659880936145782, | |
| "learning_rate": 1.237638055928251e-05, | |
| "loss": 0.0004, | |
| "step": 563 | |
| }, | |
| { | |
| "epoch": 0.42437923250564336, | |
| "grad_norm": 0.18647299706935883, | |
| "learning_rate": 1.2353412331829073e-05, | |
| "loss": 0.0062, | |
| "step": 564 | |
| }, | |
| { | |
| "epoch": 0.4251316779533484, | |
| "grad_norm": 7.033252716064453, | |
| "learning_rate": 1.2330430953713921e-05, | |
| "loss": 0.3394, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 0.4258841234010534, | |
| "grad_norm": 0.035930879414081573, | |
| "learning_rate": 1.2307436553354985e-05, | |
| "loss": 0.0009, | |
| "step": 566 | |
| }, | |
| { | |
| "epoch": 0.42663656884875845, | |
| "grad_norm": 0.0031391908414661884, | |
| "learning_rate": 1.2284429259242958e-05, | |
| "loss": 0.0001, | |
| "step": 567 | |
| }, | |
| { | |
| "epoch": 0.42738901429646353, | |
| "grad_norm": 9.076556205749512, | |
| "learning_rate": 1.2261409199940574e-05, | |
| "loss": 0.3117, | |
| "step": 568 | |
| }, | |
| { | |
| "epoch": 0.42814145974416856, | |
| "grad_norm": 0.3694950044155121, | |
| "learning_rate": 1.2238376504081913e-05, | |
| "loss": 0.0936, | |
| "step": 569 | |
| }, | |
| { | |
| "epoch": 0.4288939051918736, | |
| "grad_norm": 0.0012056067353114486, | |
| "learning_rate": 1.221533130037166e-05, | |
| "loss": 0.0, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.4296463506395786, | |
| "grad_norm": 3.5604007244110107, | |
| "learning_rate": 1.2192273717584386e-05, | |
| "loss": 0.0581, | |
| "step": 571 | |
| }, | |
| { | |
| "epoch": 0.43039879608728365, | |
| "grad_norm": 0.005317636765539646, | |
| "learning_rate": 1.2169203884563846e-05, | |
| "loss": 0.0002, | |
| "step": 572 | |
| }, | |
| { | |
| "epoch": 0.43115124153498874, | |
| "grad_norm": 0.413285493850708, | |
| "learning_rate": 1.2146121930222241e-05, | |
| "loss": 0.0035, | |
| "step": 573 | |
| }, | |
| { | |
| "epoch": 0.43190368698269377, | |
| "grad_norm": 4.266297340393066, | |
| "learning_rate": 1.2123027983539511e-05, | |
| "loss": 0.1756, | |
| "step": 574 | |
| }, | |
| { | |
| "epoch": 0.4326561324303988, | |
| "grad_norm": 10.099092483520508, | |
| "learning_rate": 1.2099922173562602e-05, | |
| "loss": 0.0599, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.43340857787810383, | |
| "grad_norm": 0.0031585677061229944, | |
| "learning_rate": 1.2076804629404752e-05, | |
| "loss": 0.0001, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 0.43416102332580886, | |
| "grad_norm": 0.13065661489963531, | |
| "learning_rate": 1.2053675480244777e-05, | |
| "loss": 0.0047, | |
| "step": 577 | |
| }, | |
| { | |
| "epoch": 0.43491346877351394, | |
| "grad_norm": 8.336974143981934, | |
| "learning_rate": 1.2030534855326326e-05, | |
| "loss": 0.1777, | |
| "step": 578 | |
| }, | |
| { | |
| "epoch": 0.435665914221219, | |
| "grad_norm": 0.007145630661398172, | |
| "learning_rate": 1.2007382883957186e-05, | |
| "loss": 0.0003, | |
| "step": 579 | |
| }, | |
| { | |
| "epoch": 0.436418359668924, | |
| "grad_norm": 1.3916585445404053, | |
| "learning_rate": 1.1984219695508546e-05, | |
| "loss": 0.0053, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.43717080511662904, | |
| "grad_norm": 19.2365665435791, | |
| "learning_rate": 1.1961045419414264e-05, | |
| "loss": 0.0929, | |
| "step": 581 | |
| }, | |
| { | |
| "epoch": 0.43792325056433407, | |
| "grad_norm": 1.4760831618332304e-05, | |
| "learning_rate": 1.1937860185170164e-05, | |
| "loss": 0.0, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 0.43867569601203915, | |
| "grad_norm": 0.003603702411055565, | |
| "learning_rate": 1.1914664122333305e-05, | |
| "loss": 0.0001, | |
| "step": 583 | |
| }, | |
| { | |
| "epoch": 0.4394281414597442, | |
| "grad_norm": 2.7130890885018744e-05, | |
| "learning_rate": 1.1891457360521253e-05, | |
| "loss": 0.0, | |
| "step": 584 | |
| }, | |
| { | |
| "epoch": 0.4401805869074492, | |
| "grad_norm": 0.004726966377347708, | |
| "learning_rate": 1.1868240029411351e-05, | |
| "loss": 0.0002, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 0.44093303235515424, | |
| "grad_norm": 0.007764340378344059, | |
| "learning_rate": 1.1845012258740016e-05, | |
| "loss": 0.0002, | |
| "step": 586 | |
| }, | |
| { | |
| "epoch": 0.44168547780285927, | |
| "grad_norm": 0.003988485783338547, | |
| "learning_rate": 1.182177417830199e-05, | |
| "loss": 0.0001, | |
| "step": 587 | |
| }, | |
| { | |
| "epoch": 0.44243792325056436, | |
| "grad_norm": 0.013868695124983788, | |
| "learning_rate": 1.1798525917949626e-05, | |
| "loss": 0.0004, | |
| "step": 588 | |
| }, | |
| { | |
| "epoch": 0.4431903686982694, | |
| "grad_norm": 0.4925473630428314, | |
| "learning_rate": 1.177526760759217e-05, | |
| "loss": 0.0091, | |
| "step": 589 | |
| }, | |
| { | |
| "epoch": 0.4439428141459744, | |
| "grad_norm": 0.26699769496917725, | |
| "learning_rate": 1.1751999377195014e-05, | |
| "loss": 0.0881, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.44469525959367945, | |
| "grad_norm": 2.3798282146453857, | |
| "learning_rate": 1.1728721356778994e-05, | |
| "loss": 0.0538, | |
| "step": 591 | |
| }, | |
| { | |
| "epoch": 0.4454477050413845, | |
| "grad_norm": 0.001672322629019618, | |
| "learning_rate": 1.1705433676419644e-05, | |
| "loss": 0.0, | |
| "step": 592 | |
| }, | |
| { | |
| "epoch": 0.44620015048908956, | |
| "grad_norm": 0.005258211866021156, | |
| "learning_rate": 1.168213646624648e-05, | |
| "loss": 0.0002, | |
| "step": 593 | |
| }, | |
| { | |
| "epoch": 0.4469525959367946, | |
| "grad_norm": 0.10926241427659988, | |
| "learning_rate": 1.1658829856442269e-05, | |
| "loss": 0.0013, | |
| "step": 594 | |
| }, | |
| { | |
| "epoch": 0.4477050413844996, | |
| "grad_norm": 0.0057695843279361725, | |
| "learning_rate": 1.1635513977242304e-05, | |
| "loss": 0.0002, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 0.44845748683220465, | |
| "grad_norm": 0.059808410704135895, | |
| "learning_rate": 1.1612188958933673e-05, | |
| "loss": 0.0007, | |
| "step": 596 | |
| }, | |
| { | |
| "epoch": 0.4492099322799097, | |
| "grad_norm": 0.0013155878987163305, | |
| "learning_rate": 1.158885493185453e-05, | |
| "loss": 0.0, | |
| "step": 597 | |
| }, | |
| { | |
| "epoch": 0.44996237772761477, | |
| "grad_norm": 10.205440521240234, | |
| "learning_rate": 1.1565512026393371e-05, | |
| "loss": 0.2459, | |
| "step": 598 | |
| }, | |
| { | |
| "epoch": 0.4507148231753198, | |
| "grad_norm": 0.0017919522942975163, | |
| "learning_rate": 1.1542160372988312e-05, | |
| "loss": 0.0001, | |
| "step": 599 | |
| }, | |
| { | |
| "epoch": 0.45146726862302483, | |
| "grad_norm": 0.0012058173306286335, | |
| "learning_rate": 1.1518800102126334e-05, | |
| "loss": 0.0, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.45221971407072986, | |
| "grad_norm": 0.0011316149029880762, | |
| "learning_rate": 1.149543134434259e-05, | |
| "loss": 0.0001, | |
| "step": 601 | |
| }, | |
| { | |
| "epoch": 0.4529721595184349, | |
| "grad_norm": 0.03514290601015091, | |
| "learning_rate": 1.1472054230219644e-05, | |
| "loss": 0.001, | |
| "step": 602 | |
| }, | |
| { | |
| "epoch": 0.45372460496614, | |
| "grad_norm": 0.742960512638092, | |
| "learning_rate": 1.1448668890386765e-05, | |
| "loss": 0.0066, | |
| "step": 603 | |
| }, | |
| { | |
| "epoch": 0.454477050413845, | |
| "grad_norm": 3.21566104888916, | |
| "learning_rate": 1.1425275455519176e-05, | |
| "loss": 0.3373, | |
| "step": 604 | |
| }, | |
| { | |
| "epoch": 0.45522949586155004, | |
| "grad_norm": 0.0010515082394704223, | |
| "learning_rate": 1.140187405633734e-05, | |
| "loss": 0.0, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 0.45598194130925507, | |
| "grad_norm": 0.431357204914093, | |
| "learning_rate": 1.1378464823606228e-05, | |
| "loss": 0.0023, | |
| "step": 606 | |
| }, | |
| { | |
| "epoch": 0.4567343867569601, | |
| "grad_norm": 0.43157681822776794, | |
| "learning_rate": 1.1355047888134571e-05, | |
| "loss": 0.0708, | |
| "step": 607 | |
| }, | |
| { | |
| "epoch": 0.4574868322046652, | |
| "grad_norm": 0.0020544701255857944, | |
| "learning_rate": 1.1331623380774156e-05, | |
| "loss": 0.0001, | |
| "step": 608 | |
| }, | |
| { | |
| "epoch": 0.4582392776523702, | |
| "grad_norm": 2.6385327146272175e-05, | |
| "learning_rate": 1.1308191432419078e-05, | |
| "loss": 0.0, | |
| "step": 609 | |
| }, | |
| { | |
| "epoch": 0.45899172310007524, | |
| "grad_norm": 0.006066860631108284, | |
| "learning_rate": 1.1284752174005005e-05, | |
| "loss": 0.0002, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.4597441685477803, | |
| "grad_norm": 0.9779205322265625, | |
| "learning_rate": 1.1261305736508458e-05, | |
| "loss": 0.1798, | |
| "step": 611 | |
| }, | |
| { | |
| "epoch": 0.4604966139954853, | |
| "grad_norm": 0.0006280777743086219, | |
| "learning_rate": 1.1237852250946077e-05, | |
| "loss": 0.0, | |
| "step": 612 | |
| }, | |
| { | |
| "epoch": 0.4612490594431904, | |
| "grad_norm": 1.0666285753250122, | |
| "learning_rate": 1.1214391848373876e-05, | |
| "loss": 0.0166, | |
| "step": 613 | |
| }, | |
| { | |
| "epoch": 0.4620015048908954, | |
| "grad_norm": 0.018536796793341637, | |
| "learning_rate": 1.1190924659886532e-05, | |
| "loss": 0.0006, | |
| "step": 614 | |
| }, | |
| { | |
| "epoch": 0.46275395033860045, | |
| "grad_norm": 0.17641091346740723, | |
| "learning_rate": 1.1167450816616639e-05, | |
| "loss": 0.0047, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 0.4635063957863055, | |
| "grad_norm": 0.02300580032169819, | |
| "learning_rate": 1.1143970449733968e-05, | |
| "loss": 0.0007, | |
| "step": 616 | |
| }, | |
| { | |
| "epoch": 0.4642588412340105, | |
| "grad_norm": 0.22667698562145233, | |
| "learning_rate": 1.112048369044475e-05, | |
| "loss": 0.0041, | |
| "step": 617 | |
| }, | |
| { | |
| "epoch": 0.4650112866817156, | |
| "grad_norm": 0.0003491557145025581, | |
| "learning_rate": 1.1096990669990942e-05, | |
| "loss": 0.0, | |
| "step": 618 | |
| }, | |
| { | |
| "epoch": 0.4657637321294206, | |
| "grad_norm": 8.98726939340122e-05, | |
| "learning_rate": 1.1073491519649475e-05, | |
| "loss": 0.0, | |
| "step": 619 | |
| }, | |
| { | |
| "epoch": 0.46651617757712566, | |
| "grad_norm": 0.23045086860656738, | |
| "learning_rate": 1.1049986370731545e-05, | |
| "loss": 0.0034, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.4672686230248307, | |
| "grad_norm": 7.242973327636719, | |
| "learning_rate": 1.102647535458186e-05, | |
| "loss": 0.3392, | |
| "step": 621 | |
| }, | |
| { | |
| "epoch": 0.4680210684725357, | |
| "grad_norm": 0.00343369715847075, | |
| "learning_rate": 1.1002958602577922e-05, | |
| "loss": 0.0001, | |
| "step": 622 | |
| }, | |
| { | |
| "epoch": 0.4687735139202408, | |
| "grad_norm": 0.001794908195734024, | |
| "learning_rate": 1.0979436246129267e-05, | |
| "loss": 0.0001, | |
| "step": 623 | |
| }, | |
| { | |
| "epoch": 0.46952595936794583, | |
| "grad_norm": 0.040094684809446335, | |
| "learning_rate": 1.0955908416676772e-05, | |
| "loss": 0.0014, | |
| "step": 624 | |
| }, | |
| { | |
| "epoch": 0.47027840481565086, | |
| "grad_norm": 1.4908933735569008e-05, | |
| "learning_rate": 1.093237524569188e-05, | |
| "loss": 0.0, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.4710308502633559, | |
| "grad_norm": 0.007998337037861347, | |
| "learning_rate": 1.0908836864675884e-05, | |
| "loss": 0.0004, | |
| "step": 626 | |
| }, | |
| { | |
| "epoch": 0.4717832957110609, | |
| "grad_norm": 0.039706043899059296, | |
| "learning_rate": 1.0885293405159196e-05, | |
| "loss": 0.0013, | |
| "step": 627 | |
| }, | |
| { | |
| "epoch": 0.472535741158766, | |
| "grad_norm": 0.002229271689429879, | |
| "learning_rate": 1.0861744998700603e-05, | |
| "loss": 0.0001, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 0.47328818660647104, | |
| "grad_norm": 10.943811416625977, | |
| "learning_rate": 1.0838191776886531e-05, | |
| "loss": 0.0575, | |
| "step": 629 | |
| }, | |
| { | |
| "epoch": 0.47404063205417607, | |
| "grad_norm": 1.3392058610916138, | |
| "learning_rate": 1.0814633871330323e-05, | |
| "loss": 0.1315, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.4747930775018811, | |
| "grad_norm": 0.21532350778579712, | |
| "learning_rate": 1.0791071413671487e-05, | |
| "loss": 0.0045, | |
| "step": 631 | |
| }, | |
| { | |
| "epoch": 0.47554552294958613, | |
| "grad_norm": 0.27968382835388184, | |
| "learning_rate": 1.0767504535574971e-05, | |
| "loss": 0.0037, | |
| "step": 632 | |
| }, | |
| { | |
| "epoch": 0.4762979683972912, | |
| "grad_norm": 0.000970702909398824, | |
| "learning_rate": 1.0743933368730417e-05, | |
| "loss": 0.0, | |
| "step": 633 | |
| }, | |
| { | |
| "epoch": 0.47705041384499625, | |
| "grad_norm": 0.013053186237812042, | |
| "learning_rate": 1.0720358044851448e-05, | |
| "loss": 0.0005, | |
| "step": 634 | |
| }, | |
| { | |
| "epoch": 0.4778028592927013, | |
| "grad_norm": 0.01000749971717596, | |
| "learning_rate": 1.0696778695674899e-05, | |
| "loss": 0.0004, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 0.4785553047404063, | |
| "grad_norm": 0.007056588772684336, | |
| "learning_rate": 1.0673195452960107e-05, | |
| "loss": 0.0002, | |
| "step": 636 | |
| }, | |
| { | |
| "epoch": 0.47930775018811134, | |
| "grad_norm": 0.014011607505381107, | |
| "learning_rate": 1.0649608448488166e-05, | |
| "loss": 0.0004, | |
| "step": 637 | |
| }, | |
| { | |
| "epoch": 0.4800601956358164, | |
| "grad_norm": 0.014452227391302586, | |
| "learning_rate": 1.0626017814061186e-05, | |
| "loss": 0.0004, | |
| "step": 638 | |
| }, | |
| { | |
| "epoch": 0.48081264108352145, | |
| "grad_norm": 0.10185810178518295, | |
| "learning_rate": 1.0602423681501564e-05, | |
| "loss": 0.0018, | |
| "step": 639 | |
| }, | |
| { | |
| "epoch": 0.4815650865312265, | |
| "grad_norm": 0.0015508810756728053, | |
| "learning_rate": 1.0578826182651243e-05, | |
| "loss": 0.0001, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.4823175319789315, | |
| "grad_norm": 4.458479404449463, | |
| "learning_rate": 1.055522544937098e-05, | |
| "loss": 0.2037, | |
| "step": 641 | |
| }, | |
| { | |
| "epoch": 0.48306997742663654, | |
| "grad_norm": 0.028539393097162247, | |
| "learning_rate": 1.0531621613539598e-05, | |
| "loss": 0.0004, | |
| "step": 642 | |
| }, | |
| { | |
| "epoch": 0.48382242287434163, | |
| "grad_norm": 0.02204856462776661, | |
| "learning_rate": 1.0508014807053261e-05, | |
| "loss": 0.0006, | |
| "step": 643 | |
| }, | |
| { | |
| "epoch": 0.48457486832204666, | |
| "grad_norm": 0.0006058391300030053, | |
| "learning_rate": 1.0484405161824743e-05, | |
| "loss": 0.0, | |
| "step": 644 | |
| }, | |
| { | |
| "epoch": 0.4853273137697517, | |
| "grad_norm": 0.00671573868021369, | |
| "learning_rate": 1.0460792809782659e-05, | |
| "loss": 0.0002, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 0.4860797592174567, | |
| "grad_norm": 0.004197990987449884, | |
| "learning_rate": 1.0437177882870768e-05, | |
| "loss": 0.0001, | |
| "step": 646 | |
| }, | |
| { | |
| "epoch": 0.48683220466516175, | |
| "grad_norm": 9.33966064453125, | |
| "learning_rate": 1.0413560513047208e-05, | |
| "loss": 0.0376, | |
| "step": 647 | |
| }, | |
| { | |
| "epoch": 0.48758465011286684, | |
| "grad_norm": 0.0011327359825372696, | |
| "learning_rate": 1.038994083228377e-05, | |
| "loss": 0.0, | |
| "step": 648 | |
| }, | |
| { | |
| "epoch": 0.48833709556057187, | |
| "grad_norm": 3.5501651763916016, | |
| "learning_rate": 1.0366318972565162e-05, | |
| "loss": 0.0192, | |
| "step": 649 | |
| }, | |
| { | |
| "epoch": 0.4890895410082769, | |
| "grad_norm": 8.738161087036133, | |
| "learning_rate": 1.0342695065888262e-05, | |
| "loss": 0.2822, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.4898419864559819, | |
| "grad_norm": 0.0026988936588168144, | |
| "learning_rate": 1.031906924426139e-05, | |
| "loss": 0.0001, | |
| "step": 651 | |
| }, | |
| { | |
| "epoch": 0.49059443190368696, | |
| "grad_norm": 0.011756551451981068, | |
| "learning_rate": 1.0295441639703563e-05, | |
| "loss": 0.0002, | |
| "step": 652 | |
| }, | |
| { | |
| "epoch": 0.49134687735139204, | |
| "grad_norm": 0.04992348700761795, | |
| "learning_rate": 1.027181238424376e-05, | |
| "loss": 0.0009, | |
| "step": 653 | |
| }, | |
| { | |
| "epoch": 0.49209932279909707, | |
| "grad_norm": 0.17335185408592224, | |
| "learning_rate": 1.0248181609920198e-05, | |
| "loss": 0.0035, | |
| "step": 654 | |
| }, | |
| { | |
| "epoch": 0.4928517682468021, | |
| "grad_norm": 0.0004414380819071084, | |
| "learning_rate": 1.0224549448779564e-05, | |
| "loss": 0.0, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 0.49360421369450713, | |
| "grad_norm": 1.6693496704101562, | |
| "learning_rate": 1.0200916032876303e-05, | |
| "loss": 0.0686, | |
| "step": 656 | |
| }, | |
| { | |
| "epoch": 0.49435665914221216, | |
| "grad_norm": 0.0005949286860413849, | |
| "learning_rate": 1.0177281494271873e-05, | |
| "loss": 0.0, | |
| "step": 657 | |
| }, | |
| { | |
| "epoch": 0.49510910458991725, | |
| "grad_norm": 0.09166250377893448, | |
| "learning_rate": 1.0153645965033998e-05, | |
| "loss": 0.0029, | |
| "step": 658 | |
| }, | |
| { | |
| "epoch": 0.4958615500376223, | |
| "grad_norm": 0.0018533958354964852, | |
| "learning_rate": 1.0130009577235946e-05, | |
| "loss": 0.0001, | |
| "step": 659 | |
| }, | |
| { | |
| "epoch": 0.4966139954853273, | |
| "grad_norm": 0.004575713537633419, | |
| "learning_rate": 1.010637246295578e-05, | |
| "loss": 0.0002, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.49736644093303234, | |
| "grad_norm": 0.49716654419898987, | |
| "learning_rate": 1.008273475427562e-05, | |
| "loss": 0.0048, | |
| "step": 661 | |
| }, | |
| { | |
| "epoch": 0.49811888638073737, | |
| "grad_norm": 1.1305954456329346, | |
| "learning_rate": 1.0059096583280907e-05, | |
| "loss": 0.0953, | |
| "step": 662 | |
| }, | |
| { | |
| "epoch": 0.49887133182844245, | |
| "grad_norm": 0.001503349863924086, | |
| "learning_rate": 1.0035458082059672e-05, | |
| "loss": 0.0001, | |
| "step": 663 | |
| }, | |
| { | |
| "epoch": 0.4996237772761475, | |
| "grad_norm": 0.010958625003695488, | |
| "learning_rate": 1.0011819382701784e-05, | |
| "loss": 0.0003, | |
| "step": 664 | |
| }, | |
| { | |
| "epoch": 0.5003762227238525, | |
| "grad_norm": 0.0015685193939134479, | |
| "learning_rate": 9.98818061729822e-06, | |
| "loss": 0.0, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 0.5011286681715575, | |
| "grad_norm": 0.010029254481196404, | |
| "learning_rate": 9.964541917940331e-06, | |
| "loss": 0.0003, | |
| "step": 666 | |
| }, | |
| { | |
| "epoch": 0.5018811136192626, | |
| "grad_norm": 0.004363952670246363, | |
| "learning_rate": 9.940903416719097e-06, | |
| "loss": 0.0001, | |
| "step": 667 | |
| }, | |
| { | |
| "epoch": 0.5026335590669676, | |
| "grad_norm": 0.01587914675474167, | |
| "learning_rate": 9.917265245724385e-06, | |
| "loss": 0.0003, | |
| "step": 668 | |
| }, | |
| { | |
| "epoch": 0.5033860045146726, | |
| "grad_norm": 0.005167737137526274, | |
| "learning_rate": 9.893627537044223e-06, | |
| "loss": 0.0002, | |
| "step": 669 | |
| }, | |
| { | |
| "epoch": 0.5041384499623778, | |
| "grad_norm": 8.244839668273926, | |
| "learning_rate": 9.869990422764056e-06, | |
| "loss": 0.1592, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.5048908954100828, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.846354034966003e-06, | |
| "loss": 0.0, | |
| "step": 671 | |
| }, | |
| { | |
| "epoch": 0.5056433408577878, | |
| "grad_norm": 18.1772518157959, | |
| "learning_rate": 9.822718505728129e-06, | |
| "loss": 0.3513, | |
| "step": 672 | |
| }, | |
| { | |
| "epoch": 0.5063957863054929, | |
| "grad_norm": 0.007274544797837734, | |
| "learning_rate": 9.7990839671237e-06, | |
| "loss": 0.0002, | |
| "step": 673 | |
| }, | |
| { | |
| "epoch": 0.5071482317531979, | |
| "grad_norm": 0.31394076347351074, | |
| "learning_rate": 9.77545055122044e-06, | |
| "loss": 0.1283, | |
| "step": 674 | |
| }, | |
| { | |
| "epoch": 0.5079006772009029, | |
| "grad_norm": 0.03855385258793831, | |
| "learning_rate": 9.751818390079805e-06, | |
| "loss": 0.0003, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.508653122648608, | |
| "grad_norm": 0.000805127143394202, | |
| "learning_rate": 9.728187615756243e-06, | |
| "loss": 0.0, | |
| "step": 676 | |
| }, | |
| { | |
| "epoch": 0.509405568096313, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.704558360296444e-06, | |
| "loss": 0.0, | |
| "step": 677 | |
| }, | |
| { | |
| "epoch": 0.510158013544018, | |
| "grad_norm": 1.416871190071106, | |
| "learning_rate": 9.680930755738616e-06, | |
| "loss": 0.016, | |
| "step": 678 | |
| }, | |
| { | |
| "epoch": 0.510910458991723, | |
| "grad_norm": 0.006285065319389105, | |
| "learning_rate": 9.657304934111742e-06, | |
| "loss": 0.0002, | |
| "step": 679 | |
| }, | |
| { | |
| "epoch": 0.5116629044394282, | |
| "grad_norm": 1.3052698373794556, | |
| "learning_rate": 9.633681027434838e-06, | |
| "loss": 0.1261, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.5124153498871332, | |
| "grad_norm": 0.0009769117459654808, | |
| "learning_rate": 9.61005916771623e-06, | |
| "loss": 0.0, | |
| "step": 681 | |
| }, | |
| { | |
| "epoch": 0.5131677953348383, | |
| "grad_norm": 0.008825716562569141, | |
| "learning_rate": 9.586439486952796e-06, | |
| "loss": 0.0003, | |
| "step": 682 | |
| }, | |
| { | |
| "epoch": 0.5139202407825433, | |
| "grad_norm": 0.004172160290181637, | |
| "learning_rate": 9.562822117129235e-06, | |
| "loss": 0.0001, | |
| "step": 683 | |
| }, | |
| { | |
| "epoch": 0.5146726862302483, | |
| "grad_norm": 0.03515918552875519, | |
| "learning_rate": 9.539207190217343e-06, | |
| "loss": 0.0009, | |
| "step": 684 | |
| }, | |
| { | |
| "epoch": 0.5154251316779533, | |
| "grad_norm": 0.0004445746308192611, | |
| "learning_rate": 9.51559483817526e-06, | |
| "loss": 0.0, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 0.5161775771256584, | |
| "grad_norm": 0.002501540118828416, | |
| "learning_rate": 9.491985192946742e-06, | |
| "loss": 0.0001, | |
| "step": 686 | |
| }, | |
| { | |
| "epoch": 0.5169300225733634, | |
| "grad_norm": 0.21380773186683655, | |
| "learning_rate": 9.468378386460406e-06, | |
| "loss": 0.0934, | |
| "step": 687 | |
| }, | |
| { | |
| "epoch": 0.5176824680210684, | |
| "grad_norm": 0.005090567748993635, | |
| "learning_rate": 9.444774550629024e-06, | |
| "loss": 0.0002, | |
| "step": 688 | |
| }, | |
| { | |
| "epoch": 0.5184349134687735, | |
| "grad_norm": 0.0015086415223777294, | |
| "learning_rate": 9.42117381734876e-06, | |
| "loss": 0.0001, | |
| "step": 689 | |
| }, | |
| { | |
| "epoch": 0.5191873589164786, | |
| "grad_norm": 0.018775468692183495, | |
| "learning_rate": 9.397576318498438e-06, | |
| "loss": 0.0007, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.5199398043641836, | |
| "grad_norm": 0.8499560356140137, | |
| "learning_rate": 9.373982185938815e-06, | |
| "loss": 0.0083, | |
| "step": 691 | |
| }, | |
| { | |
| "epoch": 0.5206922498118887, | |
| "grad_norm": 0.2695012092590332, | |
| "learning_rate": 9.350391551511837e-06, | |
| "loss": 0.0015, | |
| "step": 692 | |
| }, | |
| { | |
| "epoch": 0.5214446952595937, | |
| "grad_norm": 0.06739958375692368, | |
| "learning_rate": 9.326804547039894e-06, | |
| "loss": 0.0009, | |
| "step": 693 | |
| }, | |
| { | |
| "epoch": 0.5221971407072987, | |
| "grad_norm": 1.989675521850586, | |
| "learning_rate": 9.303221304325103e-06, | |
| "loss": 0.2139, | |
| "step": 694 | |
| }, | |
| { | |
| "epoch": 0.5229495861550038, | |
| "grad_norm": 0.04741863161325455, | |
| "learning_rate": 9.279641955148553e-06, | |
| "loss": 0.0013, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 0.5237020316027088, | |
| "grad_norm": 0.060340896248817444, | |
| "learning_rate": 9.256066631269586e-06, | |
| "loss": 0.0008, | |
| "step": 696 | |
| }, | |
| { | |
| "epoch": 0.5244544770504138, | |
| "grad_norm": 2.983630895614624, | |
| "learning_rate": 9.232495464425034e-06, | |
| "loss": 0.4607, | |
| "step": 697 | |
| }, | |
| { | |
| "epoch": 0.5252069224981188, | |
| "grad_norm": 0.0010294626699760556, | |
| "learning_rate": 9.208928586328518e-06, | |
| "loss": 0.0, | |
| "step": 698 | |
| }, | |
| { | |
| "epoch": 0.5259593679458239, | |
| "grad_norm": 1.120234537665965e-05, | |
| "learning_rate": 9.185366128669682e-06, | |
| "loss": 0.0, | |
| "step": 699 | |
| }, | |
| { | |
| "epoch": 0.526711813393529, | |
| "grad_norm": 2.128415107727051, | |
| "learning_rate": 9.161808223113469e-06, | |
| "loss": 0.1173, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.527464258841234, | |
| "grad_norm": 0.0016029436374083161, | |
| "learning_rate": 9.138255001299402e-06, | |
| "loss": 0.0001, | |
| "step": 701 | |
| }, | |
| { | |
| "epoch": 0.5282167042889391, | |
| "grad_norm": 6.54170560836792, | |
| "learning_rate": 9.114706594840806e-06, | |
| "loss": 0.1784, | |
| "step": 702 | |
| }, | |
| { | |
| "epoch": 0.5289691497366441, | |
| "grad_norm": 0.005230502225458622, | |
| "learning_rate": 9.091163135324119e-06, | |
| "loss": 0.0002, | |
| "step": 703 | |
| }, | |
| { | |
| "epoch": 0.5297215951843491, | |
| "grad_norm": 0.028300784528255463, | |
| "learning_rate": 9.067624754308124e-06, | |
| "loss": 0.0004, | |
| "step": 704 | |
| }, | |
| { | |
| "epoch": 0.5304740406320542, | |
| "grad_norm": 1.312003587372601e-05, | |
| "learning_rate": 9.044091583323231e-06, | |
| "loss": 0.0, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 0.5312264860797592, | |
| "grad_norm": 0.03969345614314079, | |
| "learning_rate": 9.020563753870734e-06, | |
| "loss": 0.0009, | |
| "step": 706 | |
| }, | |
| { | |
| "epoch": 0.5319789315274642, | |
| "grad_norm": 29.23634910583496, | |
| "learning_rate": 8.997041397422083e-06, | |
| "loss": 0.2304, | |
| "step": 707 | |
| }, | |
| { | |
| "epoch": 0.5327313769751693, | |
| "grad_norm": 0.08085115998983383, | |
| "learning_rate": 8.973524645418142e-06, | |
| "loss": 0.0021, | |
| "step": 708 | |
| }, | |
| { | |
| "epoch": 0.5334838224228743, | |
| "grad_norm": 0.001823053928092122, | |
| "learning_rate": 8.95001362926846e-06, | |
| "loss": 0.0001, | |
| "step": 709 | |
| }, | |
| { | |
| "epoch": 0.5342362678705794, | |
| "grad_norm": 0.0950389876961708, | |
| "learning_rate": 8.926508480350525e-06, | |
| "loss": 0.0008, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.5349887133182845, | |
| "grad_norm": 0.0005177335697226226, | |
| "learning_rate": 8.903009330009063e-06, | |
| "loss": 0.0, | |
| "step": 711 | |
| }, | |
| { | |
| "epoch": 0.5357411587659895, | |
| "grad_norm": 0.007335309404879808, | |
| "learning_rate": 8.879516309555252e-06, | |
| "loss": 0.0003, | |
| "step": 712 | |
| }, | |
| { | |
| "epoch": 0.5364936042136945, | |
| "grad_norm": 0.0022024691570550203, | |
| "learning_rate": 8.856029550266036e-06, | |
| "loss": 0.0001, | |
| "step": 713 | |
| }, | |
| { | |
| "epoch": 0.5372460496613995, | |
| "grad_norm": 0.0016011092811822891, | |
| "learning_rate": 8.832549183383363e-06, | |
| "loss": 0.0001, | |
| "step": 714 | |
| }, | |
| { | |
| "epoch": 0.5379984951091046, | |
| "grad_norm": 0.009527268819510937, | |
| "learning_rate": 8.80907534011347e-06, | |
| "loss": 0.0003, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 0.5387509405568096, | |
| "grad_norm": 0.04128224030137062, | |
| "learning_rate": 8.785608151626126e-06, | |
| "loss": 0.0012, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 0.5395033860045146, | |
| "grad_norm": 0.013059580698609352, | |
| "learning_rate": 8.762147749053928e-06, | |
| "loss": 0.0004, | |
| "step": 717 | |
| }, | |
| { | |
| "epoch": 0.5402558314522197, | |
| "grad_norm": 0.0006950248498469591, | |
| "learning_rate": 8.738694263491545e-06, | |
| "loss": 0.0, | |
| "step": 718 | |
| }, | |
| { | |
| "epoch": 0.5410082768999247, | |
| "grad_norm": 0.050438202917575836, | |
| "learning_rate": 8.715247825995e-06, | |
| "loss": 0.0014, | |
| "step": 719 | |
| }, | |
| { | |
| "epoch": 0.5417607223476298, | |
| "grad_norm": 0.007887017913162708, | |
| "learning_rate": 8.691808567580922e-06, | |
| "loss": 0.0002, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.5425131677953349, | |
| "grad_norm": 1.9573593139648438, | |
| "learning_rate": 8.668376619225846e-06, | |
| "loss": 0.1971, | |
| "step": 721 | |
| }, | |
| { | |
| "epoch": 0.5432656132430399, | |
| "grad_norm": 0.005204391200095415, | |
| "learning_rate": 8.64495211186543e-06, | |
| "loss": 0.0002, | |
| "step": 722 | |
| }, | |
| { | |
| "epoch": 0.5440180586907449, | |
| "grad_norm": 0.10841142386198044, | |
| "learning_rate": 8.621535176393776e-06, | |
| "loss": 0.0026, | |
| "step": 723 | |
| }, | |
| { | |
| "epoch": 0.54477050413845, | |
| "grad_norm": 0.006886586546897888, | |
| "learning_rate": 8.598125943662662e-06, | |
| "loss": 0.0003, | |
| "step": 724 | |
| }, | |
| { | |
| "epoch": 0.545522949586155, | |
| "grad_norm": 1.904536485671997, | |
| "learning_rate": 8.574724544480829e-06, | |
| "loss": 0.0237, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.54627539503386, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.551331109613238e-06, | |
| "loss": 0.0, | |
| "step": 726 | |
| }, | |
| { | |
| "epoch": 0.547027840481565, | |
| "grad_norm": 0.048206303268671036, | |
| "learning_rate": 8.527945769780358e-06, | |
| "loss": 0.0006, | |
| "step": 727 | |
| }, | |
| { | |
| "epoch": 0.5477802859292701, | |
| "grad_norm": 0.05634288117289543, | |
| "learning_rate": 8.504568655657415e-06, | |
| "loss": 0.001, | |
| "step": 728 | |
| }, | |
| { | |
| "epoch": 0.5485327313769752, | |
| "grad_norm": 0.43885576725006104, | |
| "learning_rate": 8.481199897873667e-06, | |
| "loss": 0.1386, | |
| "step": 729 | |
| }, | |
| { | |
| "epoch": 0.5492851768246803, | |
| "grad_norm": 2.3144471645355225, | |
| "learning_rate": 8.457839627011693e-06, | |
| "loss": 0.0281, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.5500376222723853, | |
| "grad_norm": 0.016497652977705002, | |
| "learning_rate": 8.43448797360663e-06, | |
| "loss": 0.0004, | |
| "step": 731 | |
| }, | |
| { | |
| "epoch": 0.5507900677200903, | |
| "grad_norm": 3.199171543121338, | |
| "learning_rate": 8.411145068145474e-06, | |
| "loss": 0.5294, | |
| "step": 732 | |
| }, | |
| { | |
| "epoch": 0.5515425131677953, | |
| "grad_norm": 0.08361708372831345, | |
| "learning_rate": 8.38781104106633e-06, | |
| "loss": 0.0013, | |
| "step": 733 | |
| }, | |
| { | |
| "epoch": 0.5522949586155004, | |
| "grad_norm": 0.011733477003872395, | |
| "learning_rate": 8.3644860227577e-06, | |
| "loss": 0.0004, | |
| "step": 734 | |
| }, | |
| { | |
| "epoch": 0.5530474040632054, | |
| "grad_norm": 0.005462857894599438, | |
| "learning_rate": 8.341170143557733e-06, | |
| "loss": 0.0002, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.5537998495109104, | |
| "grad_norm": 0.0030956827104091644, | |
| "learning_rate": 8.317863533753523e-06, | |
| "loss": 0.0001, | |
| "step": 736 | |
| }, | |
| { | |
| "epoch": 0.5545522949586155, | |
| "grad_norm": 1.1541496515274048, | |
| "learning_rate": 8.294566323580359e-06, | |
| "loss": 0.1397, | |
| "step": 737 | |
| }, | |
| { | |
| "epoch": 0.5553047404063205, | |
| "grad_norm": 4.27673864364624, | |
| "learning_rate": 8.27127864322101e-06, | |
| "loss": 0.0181, | |
| "step": 738 | |
| }, | |
| { | |
| "epoch": 0.5560571858540256, | |
| "grad_norm": 0.05562518537044525, | |
| "learning_rate": 8.248000622804986e-06, | |
| "loss": 0.0015, | |
| "step": 739 | |
| }, | |
| { | |
| "epoch": 0.5568096313017307, | |
| "grad_norm": 1.1070342063903809, | |
| "learning_rate": 8.224732392407834e-06, | |
| "loss": 0.0464, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.5575620767494357, | |
| "grad_norm": 0.007482460699975491, | |
| "learning_rate": 8.201474082050376e-06, | |
| "loss": 0.0003, | |
| "step": 741 | |
| }, | |
| { | |
| "epoch": 0.5583145221971407, | |
| "grad_norm": 0.007131911348551512, | |
| "learning_rate": 8.178225821698013e-06, | |
| "loss": 0.0003, | |
| "step": 742 | |
| }, | |
| { | |
| "epoch": 0.5590669676448458, | |
| "grad_norm": 0.010571115650236607, | |
| "learning_rate": 8.154987741259986e-06, | |
| "loss": 0.0002, | |
| "step": 743 | |
| }, | |
| { | |
| "epoch": 0.5598194130925508, | |
| "grad_norm": 0.007416147738695145, | |
| "learning_rate": 8.13175997058865e-06, | |
| "loss": 0.0003, | |
| "step": 744 | |
| }, | |
| { | |
| "epoch": 0.5605718585402558, | |
| "grad_norm": 0.07424626499414444, | |
| "learning_rate": 8.10854263947875e-06, | |
| "loss": 0.002, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 0.5613243039879608, | |
| "grad_norm": 0.009504971094429493, | |
| "learning_rate": 8.085335877666696e-06, | |
| "loss": 0.0004, | |
| "step": 746 | |
| }, | |
| { | |
| "epoch": 0.5620767494356659, | |
| "grad_norm": 0.019785739481449127, | |
| "learning_rate": 8.062139814829839e-06, | |
| "loss": 0.0006, | |
| "step": 747 | |
| }, | |
| { | |
| "epoch": 0.5628291948833709, | |
| "grad_norm": 0.002929375506937504, | |
| "learning_rate": 8.038954580585742e-06, | |
| "loss": 0.0001, | |
| "step": 748 | |
| }, | |
| { | |
| "epoch": 0.563581640331076, | |
| "grad_norm": 3.0780787467956543, | |
| "learning_rate": 8.015780304491457e-06, | |
| "loss": 0.069, | |
| "step": 749 | |
| }, | |
| { | |
| "epoch": 0.5643340857787811, | |
| "grad_norm": 10.312975883483887, | |
| "learning_rate": 7.992617116042813e-06, | |
| "loss": 0.0922, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.5650865312264861, | |
| "grad_norm": 0.4201790690422058, | |
| "learning_rate": 7.969465144673674e-06, | |
| "loss": 0.0038, | |
| "step": 751 | |
| }, | |
| { | |
| "epoch": 0.5658389766741911, | |
| "grad_norm": 0.24568204581737518, | |
| "learning_rate": 7.946324519755225e-06, | |
| "loss": 0.0055, | |
| "step": 752 | |
| }, | |
| { | |
| "epoch": 0.5665914221218962, | |
| "grad_norm": 0.007804738823324442, | |
| "learning_rate": 7.92319537059525e-06, | |
| "loss": 0.0004, | |
| "step": 753 | |
| }, | |
| { | |
| "epoch": 0.5673438675696012, | |
| "grad_norm": 0.002531126607209444, | |
| "learning_rate": 7.900077826437402e-06, | |
| "loss": 0.0001, | |
| "step": 754 | |
| }, | |
| { | |
| "epoch": 0.5680963130173062, | |
| "grad_norm": 0.002385746920481324, | |
| "learning_rate": 7.876972016460492e-06, | |
| "loss": 0.0001, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 0.5688487584650113, | |
| "grad_norm": 0.009147096425294876, | |
| "learning_rate": 7.853878069777762e-06, | |
| "loss": 0.0004, | |
| "step": 756 | |
| }, | |
| { | |
| "epoch": 0.5696012039127163, | |
| "grad_norm": 0.4006285071372986, | |
| "learning_rate": 7.83079611543616e-06, | |
| "loss": 0.0046, | |
| "step": 757 | |
| }, | |
| { | |
| "epoch": 0.5703536493604213, | |
| "grad_norm": 3.700183868408203, | |
| "learning_rate": 7.80772628241562e-06, | |
| "loss": 0.1669, | |
| "step": 758 | |
| }, | |
| { | |
| "epoch": 0.5711060948081265, | |
| "grad_norm": 0.007939686998724937, | |
| "learning_rate": 7.784668699628345e-06, | |
| "loss": 0.0003, | |
| "step": 759 | |
| }, | |
| { | |
| "epoch": 0.5718585402558315, | |
| "grad_norm": 0.001986120129004121, | |
| "learning_rate": 7.761623495918089e-06, | |
| "loss": 0.0001, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.5726109857035365, | |
| "grad_norm": 0.0024444598238915205, | |
| "learning_rate": 7.738590800059427e-06, | |
| "loss": 0.0001, | |
| "step": 761 | |
| }, | |
| { | |
| "epoch": 0.5733634311512416, | |
| "grad_norm": 0.0029928828589618206, | |
| "learning_rate": 7.715570740757045e-06, | |
| "loss": 0.0001, | |
| "step": 762 | |
| }, | |
| { | |
| "epoch": 0.5741158765989466, | |
| "grad_norm": 0.02208460308611393, | |
| "learning_rate": 7.692563446645017e-06, | |
| "loss": 0.0009, | |
| "step": 763 | |
| }, | |
| { | |
| "epoch": 0.5748683220466516, | |
| "grad_norm": 0.02059108205139637, | |
| "learning_rate": 7.66956904628608e-06, | |
| "loss": 0.0005, | |
| "step": 764 | |
| }, | |
| { | |
| "epoch": 0.5756207674943566, | |
| "grad_norm": 4.884920120239258, | |
| "learning_rate": 7.64658766817093e-06, | |
| "loss": 0.0456, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 0.5763732129420617, | |
| "grad_norm": 0.04961561784148216, | |
| "learning_rate": 7.623619440717493e-06, | |
| "loss": 0.0016, | |
| "step": 766 | |
| }, | |
| { | |
| "epoch": 0.5771256583897667, | |
| "grad_norm": 0.024083776399493217, | |
| "learning_rate": 7.600664492270206e-06, | |
| "loss": 0.0006, | |
| "step": 767 | |
| }, | |
| { | |
| "epoch": 0.5778781038374717, | |
| "grad_norm": 0.017353009432554245, | |
| "learning_rate": 7.57772295109931e-06, | |
| "loss": 0.0006, | |
| "step": 768 | |
| }, | |
| { | |
| "epoch": 0.5786305492851769, | |
| "grad_norm": 0.00047786516370251775, | |
| "learning_rate": 7.554794945400122e-06, | |
| "loss": 0.0, | |
| "step": 769 | |
| }, | |
| { | |
| "epoch": 0.5793829947328819, | |
| "grad_norm": 0.2889784276485443, | |
| "learning_rate": 7.531880603292333e-06, | |
| "loss": 0.0052, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.5801354401805869, | |
| "grad_norm": 5.133121490478516, | |
| "learning_rate": 7.508980052819274e-06, | |
| "loss": 0.1127, | |
| "step": 771 | |
| }, | |
| { | |
| "epoch": 0.580887885628292, | |
| "grad_norm": 0.06616966426372528, | |
| "learning_rate": 7.486093421947214e-06, | |
| "loss": 0.0023, | |
| "step": 772 | |
| }, | |
| { | |
| "epoch": 0.581640331075997, | |
| "grad_norm": 0.0013333017705008388, | |
| "learning_rate": 7.463220838564635e-06, | |
| "loss": 0.0001, | |
| "step": 773 | |
| }, | |
| { | |
| "epoch": 0.582392776523702, | |
| "grad_norm": 1.0273889303207397, | |
| "learning_rate": 7.440362430481529e-06, | |
| "loss": 0.08, | |
| "step": 774 | |
| }, | |
| { | |
| "epoch": 0.5831452219714071, | |
| "grad_norm": 0.7515692710876465, | |
| "learning_rate": 7.417518325428678e-06, | |
| "loss": 0.11, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.5838976674191121, | |
| "grad_norm": 3.273358743172139e-05, | |
| "learning_rate": 7.3946886510569385e-06, | |
| "loss": 0.0, | |
| "step": 776 | |
| }, | |
| { | |
| "epoch": 0.5846501128668171, | |
| "grad_norm": 3.785158634185791, | |
| "learning_rate": 7.371873534936522e-06, | |
| "loss": 0.448, | |
| "step": 777 | |
| }, | |
| { | |
| "epoch": 0.5854025583145221, | |
| "grad_norm": 0.004100090358406305, | |
| "learning_rate": 7.349073104556301e-06, | |
| "loss": 0.0002, | |
| "step": 778 | |
| }, | |
| { | |
| "epoch": 0.5861550037622273, | |
| "grad_norm": 0.28187495470046997, | |
| "learning_rate": 7.326287487323078e-06, | |
| "loss": 0.0012, | |
| "step": 779 | |
| }, | |
| { | |
| "epoch": 0.5869074492099323, | |
| "grad_norm": 5.85366678237915, | |
| "learning_rate": 7.3035168105608885e-06, | |
| "loss": 0.0818, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.5876598946576373, | |
| "grad_norm": 0.005729475524276495, | |
| "learning_rate": 7.280761201510275e-06, | |
| "loss": 0.0002, | |
| "step": 781 | |
| }, | |
| { | |
| "epoch": 0.5884123401053424, | |
| "grad_norm": 0.1717999428510666, | |
| "learning_rate": 7.2580207873275865e-06, | |
| "loss": 0.0992, | |
| "step": 782 | |
| }, | |
| { | |
| "epoch": 0.5891647855530474, | |
| "grad_norm": 0.0329020619392395, | |
| "learning_rate": 7.235295695084259e-06, | |
| "loss": 0.0006, | |
| "step": 783 | |
| }, | |
| { | |
| "epoch": 0.5899172310007524, | |
| "grad_norm": 0.012131314724683762, | |
| "learning_rate": 7.212586051766118e-06, | |
| "loss": 0.0003, | |
| "step": 784 | |
| }, | |
| { | |
| "epoch": 0.5906696764484575, | |
| "grad_norm": 1.1857929229736328, | |
| "learning_rate": 7.189891984272659e-06, | |
| "loss": 0.0395, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 0.5914221218961625, | |
| "grad_norm": 0.0067533692345023155, | |
| "learning_rate": 7.16721361941634e-06, | |
| "loss": 0.0003, | |
| "step": 786 | |
| }, | |
| { | |
| "epoch": 0.5921745673438675, | |
| "grad_norm": 0.0263225045055151, | |
| "learning_rate": 7.144551083921875e-06, | |
| "loss": 0.001, | |
| "step": 787 | |
| }, | |
| { | |
| "epoch": 0.5929270127915726, | |
| "grad_norm": 2.086378812789917, | |
| "learning_rate": 7.121904504425523e-06, | |
| "loss": 0.0377, | |
| "step": 788 | |
| }, | |
| { | |
| "epoch": 0.5936794582392777, | |
| "grad_norm": 2.842104196548462, | |
| "learning_rate": 7.0992740074743835e-06, | |
| "loss": 0.0224, | |
| "step": 789 | |
| }, | |
| { | |
| "epoch": 0.5944319036869827, | |
| "grad_norm": 0.00352369318716228, | |
| "learning_rate": 7.076659719525694e-06, | |
| "loss": 0.0002, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.5951843491346878, | |
| "grad_norm": 0.02450047992169857, | |
| "learning_rate": 7.05406176694611e-06, | |
| "loss": 0.0006, | |
| "step": 791 | |
| }, | |
| { | |
| "epoch": 0.5959367945823928, | |
| "grad_norm": 0.01613771915435791, | |
| "learning_rate": 7.031480276011007e-06, | |
| "loss": 0.0007, | |
| "step": 792 | |
| }, | |
| { | |
| "epoch": 0.5966892400300978, | |
| "grad_norm": 16.005887985229492, | |
| "learning_rate": 7.008915372903775e-06, | |
| "loss": 0.1847, | |
| "step": 793 | |
| }, | |
| { | |
| "epoch": 0.5974416854778029, | |
| "grad_norm": 0.023932024836540222, | |
| "learning_rate": 6.986367183715117e-06, | |
| "loss": 0.0009, | |
| "step": 794 | |
| }, | |
| { | |
| "epoch": 0.5981941309255079, | |
| "grad_norm": 0.07864760607481003, | |
| "learning_rate": 6.963835834442336e-06, | |
| "loss": 0.0028, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 0.5989465763732129, | |
| "grad_norm": 0.09168053418397903, | |
| "learning_rate": 6.941321450988633e-06, | |
| "loss": 0.0023, | |
| "step": 796 | |
| }, | |
| { | |
| "epoch": 0.5996990218209179, | |
| "grad_norm": 2.519742965698242, | |
| "learning_rate": 6.918824159162409e-06, | |
| "loss": 0.3657, | |
| "step": 797 | |
| }, | |
| { | |
| "epoch": 0.600451467268623, | |
| "grad_norm": 0.7624035477638245, | |
| "learning_rate": 6.89634408467656e-06, | |
| "loss": 0.057, | |
| "step": 798 | |
| }, | |
| { | |
| "epoch": 0.6012039127163281, | |
| "grad_norm": 3.3864188194274902, | |
| "learning_rate": 6.873881353147766e-06, | |
| "loss": 0.3926, | |
| "step": 799 | |
| }, | |
| { | |
| "epoch": 0.6019563581640331, | |
| "grad_norm": 1.0742292404174805, | |
| "learning_rate": 6.851436090095807e-06, | |
| "loss": 0.0186, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.6027088036117382, | |
| "grad_norm": 3.5620031356811523, | |
| "learning_rate": 6.829008420942842e-06, | |
| "loss": 0.0536, | |
| "step": 801 | |
| }, | |
| { | |
| "epoch": 0.6034612490594432, | |
| "grad_norm": 2.453146457672119, | |
| "learning_rate": 6.806598471012717e-06, | |
| "loss": 0.0645, | |
| "step": 802 | |
| }, | |
| { | |
| "epoch": 0.6042136945071482, | |
| "grad_norm": 0.018604706972837448, | |
| "learning_rate": 6.784206365530268e-06, | |
| "loss": 0.0006, | |
| "step": 803 | |
| }, | |
| { | |
| "epoch": 0.6049661399548533, | |
| "grad_norm": 0.6108080744743347, | |
| "learning_rate": 6.761832229620618e-06, | |
| "loss": 0.0065, | |
| "step": 804 | |
| }, | |
| { | |
| "epoch": 0.6057185854025583, | |
| "grad_norm": 0.9389123916625977, | |
| "learning_rate": 6.739476188308476e-06, | |
| "loss": 0.0085, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 0.6064710308502633, | |
| "grad_norm": 0.0035854075103998184, | |
| "learning_rate": 6.717138366517438e-06, | |
| "loss": 0.0002, | |
| "step": 806 | |
| }, | |
| { | |
| "epoch": 0.6072234762979684, | |
| "grad_norm": 0.003520939266309142, | |
| "learning_rate": 6.694818889069294e-06, | |
| "loss": 0.0002, | |
| "step": 807 | |
| }, | |
| { | |
| "epoch": 0.6079759217456734, | |
| "grad_norm": 0.9055103659629822, | |
| "learning_rate": 6.672517880683332e-06, | |
| "loss": 0.0895, | |
| "step": 808 | |
| }, | |
| { | |
| "epoch": 0.6087283671933785, | |
| "grad_norm": 0.1636822521686554, | |
| "learning_rate": 6.6502354659756165e-06, | |
| "loss": 0.0064, | |
| "step": 809 | |
| }, | |
| { | |
| "epoch": 0.6094808126410836, | |
| "grad_norm": 4.655453681945801, | |
| "learning_rate": 6.627971769458341e-06, | |
| "loss": 0.1029, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.6102332580887886, | |
| "grad_norm": 0.3717281222343445, | |
| "learning_rate": 6.605726915539088e-06, | |
| "loss": 0.2568, | |
| "step": 811 | |
| }, | |
| { | |
| "epoch": 0.6109857035364936, | |
| "grad_norm": 0.27854296565055847, | |
| "learning_rate": 6.583501028520143e-06, | |
| "loss": 0.107, | |
| "step": 812 | |
| }, | |
| { | |
| "epoch": 0.6117381489841986, | |
| "grad_norm": 1.342350959777832, | |
| "learning_rate": 6.561294232597817e-06, | |
| "loss": 0.2153, | |
| "step": 813 | |
| }, | |
| { | |
| "epoch": 0.6124905944319037, | |
| "grad_norm": 0.017684536054730415, | |
| "learning_rate": 6.539106651861741e-06, | |
| "loss": 0.0005, | |
| "step": 814 | |
| }, | |
| { | |
| "epoch": 0.6132430398796087, | |
| "grad_norm": 0.005973528604954481, | |
| "learning_rate": 6.516938410294165e-06, | |
| "loss": 0.0002, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 0.6139954853273137, | |
| "grad_norm": 0.12207470089197159, | |
| "learning_rate": 6.494789631769281e-06, | |
| "loss": 0.0047, | |
| "step": 816 | |
| }, | |
| { | |
| "epoch": 0.6147479307750188, | |
| "grad_norm": 0.2868960499763489, | |
| "learning_rate": 6.472660440052521e-06, | |
| "loss": 0.1167, | |
| "step": 817 | |
| }, | |
| { | |
| "epoch": 0.6155003762227238, | |
| "grad_norm": 0.023519231006503105, | |
| "learning_rate": 6.450550958799868e-06, | |
| "loss": 0.0009, | |
| "step": 818 | |
| }, | |
| { | |
| "epoch": 0.6162528216704289, | |
| "grad_norm": 0.001310076448135078, | |
| "learning_rate": 6.428461311557159e-06, | |
| "loss": 0.0001, | |
| "step": 819 | |
| }, | |
| { | |
| "epoch": 0.617005267118134, | |
| "grad_norm": 0.009395822882652283, | |
| "learning_rate": 6.406391621759416e-06, | |
| "loss": 0.0005, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.617757712565839, | |
| "grad_norm": 0.0022374300751835108, | |
| "learning_rate": 6.384342012730122e-06, | |
| "loss": 0.0001, | |
| "step": 821 | |
| }, | |
| { | |
| "epoch": 0.618510158013544, | |
| "grad_norm": 0.02684149146080017, | |
| "learning_rate": 6.362312607680559e-06, | |
| "loss": 0.001, | |
| "step": 822 | |
| }, | |
| { | |
| "epoch": 0.6192626034612491, | |
| "grad_norm": 0.02339085564017296, | |
| "learning_rate": 6.3403035297091145e-06, | |
| "loss": 0.0011, | |
| "step": 823 | |
| }, | |
| { | |
| "epoch": 0.6200150489089541, | |
| "grad_norm": 0.3724048435688019, | |
| "learning_rate": 6.318314901800584e-06, | |
| "loss": 0.1327, | |
| "step": 824 | |
| }, | |
| { | |
| "epoch": 0.6207674943566591, | |
| "grad_norm": 0.10902447998523712, | |
| "learning_rate": 6.29634684682549e-06, | |
| "loss": 0.0025, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.6215199398043642, | |
| "grad_norm": 1.5880934000015259, | |
| "learning_rate": 6.274399487539397e-06, | |
| "loss": 0.09, | |
| "step": 826 | |
| }, | |
| { | |
| "epoch": 0.6222723852520692, | |
| "grad_norm": 0.024421492591500282, | |
| "learning_rate": 6.2524729465822265e-06, | |
| "loss": 0.0008, | |
| "step": 827 | |
| }, | |
| { | |
| "epoch": 0.6230248306997742, | |
| "grad_norm": 0.3446272909641266, | |
| "learning_rate": 6.230567346477567e-06, | |
| "loss": 0.0032, | |
| "step": 828 | |
| }, | |
| { | |
| "epoch": 0.6237772761474794, | |
| "grad_norm": 0.007315884344279766, | |
| "learning_rate": 6.208682809631983e-06, | |
| "loss": 0.0002, | |
| "step": 829 | |
| }, | |
| { | |
| "epoch": 0.6245297215951844, | |
| "grad_norm": 0.3761526346206665, | |
| "learning_rate": 6.1868194583343585e-06, | |
| "loss": 0.1232, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.6252821670428894, | |
| "grad_norm": 0.005384144373238087, | |
| "learning_rate": 6.1649774147551755e-06, | |
| "loss": 0.0002, | |
| "step": 831 | |
| }, | |
| { | |
| "epoch": 0.6260346124905944, | |
| "grad_norm": 2.6692087650299072, | |
| "learning_rate": 6.1431568009458596e-06, | |
| "loss": 0.4896, | |
| "step": 832 | |
| }, | |
| { | |
| "epoch": 0.6267870579382995, | |
| "grad_norm": 0.008824239484965801, | |
| "learning_rate": 6.121357738838088e-06, | |
| "loss": 0.0004, | |
| "step": 833 | |
| }, | |
| { | |
| "epoch": 0.6275395033860045, | |
| "grad_norm": 3.028167963027954, | |
| "learning_rate": 6.099580350243109e-06, | |
| "loss": 0.2251, | |
| "step": 834 | |
| }, | |
| { | |
| "epoch": 0.6282919488337095, | |
| "grad_norm": 0.011604844592511654, | |
| "learning_rate": 6.077824756851055e-06, | |
| "loss": 0.0005, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 0.6290443942814146, | |
| "grad_norm": 3.686901092529297, | |
| "learning_rate": 6.056091080230279e-06, | |
| "loss": 0.1939, | |
| "step": 836 | |
| }, | |
| { | |
| "epoch": 0.6297968397291196, | |
| "grad_norm": 0.012510290369391441, | |
| "learning_rate": 6.034379441826659e-06, | |
| "loss": 0.0006, | |
| "step": 837 | |
| }, | |
| { | |
| "epoch": 0.6305492851768246, | |
| "grad_norm": 0.014967096969485283, | |
| "learning_rate": 6.012689962962923e-06, | |
| "loss": 0.0006, | |
| "step": 838 | |
| }, | |
| { | |
| "epoch": 0.6313017306245298, | |
| "grad_norm": 0.00533378915861249, | |
| "learning_rate": 5.991022764837979e-06, | |
| "loss": 0.0003, | |
| "step": 839 | |
| }, | |
| { | |
| "epoch": 0.6320541760722348, | |
| "grad_norm": 0.021024620160460472, | |
| "learning_rate": 5.969377968526231e-06, | |
| "loss": 0.0009, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.6328066215199398, | |
| "grad_norm": 0.15575391054153442, | |
| "learning_rate": 5.947755694976902e-06, | |
| "loss": 0.0038, | |
| "step": 841 | |
| }, | |
| { | |
| "epoch": 0.6335590669676449, | |
| "grad_norm": 0.5761941075325012, | |
| "learning_rate": 5.926156065013359e-06, | |
| "loss": 0.0294, | |
| "step": 842 | |
| }, | |
| { | |
| "epoch": 0.6343115124153499, | |
| "grad_norm": 0.5115591287612915, | |
| "learning_rate": 5.904579199332443e-06, | |
| "loss": 0.081, | |
| "step": 843 | |
| }, | |
| { | |
| "epoch": 0.6350639578630549, | |
| "grad_norm": 0.017021209001541138, | |
| "learning_rate": 5.883025218503781e-06, | |
| "loss": 0.0008, | |
| "step": 844 | |
| }, | |
| { | |
| "epoch": 0.63581640331076, | |
| "grad_norm": 0.0051472182385623455, | |
| "learning_rate": 5.861494242969134e-06, | |
| "loss": 0.0002, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 0.636568848758465, | |
| "grad_norm": 0.007996303029358387, | |
| "learning_rate": 5.839986393041701e-06, | |
| "loss": 0.0003, | |
| "step": 846 | |
| }, | |
| { | |
| "epoch": 0.63732129420617, | |
| "grad_norm": 0.0032203912269324064, | |
| "learning_rate": 5.818501788905464e-06, | |
| "loss": 0.0001, | |
| "step": 847 | |
| }, | |
| { | |
| "epoch": 0.6380737396538751, | |
| "grad_norm": 0.01805216819047928, | |
| "learning_rate": 5.7970405506145e-06, | |
| "loss": 0.0008, | |
| "step": 848 | |
| }, | |
| { | |
| "epoch": 0.6388261851015802, | |
| "grad_norm": 0.015916842967271805, | |
| "learning_rate": 5.775602798092335e-06, | |
| "loss": 0.0007, | |
| "step": 849 | |
| }, | |
| { | |
| "epoch": 0.6395786305492852, | |
| "grad_norm": 0.18560563027858734, | |
| "learning_rate": 5.754188651131246e-06, | |
| "loss": 0.0048, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.6403310759969902, | |
| "grad_norm": 0.02914322167634964, | |
| "learning_rate": 5.732798229391613e-06, | |
| "loss": 0.0012, | |
| "step": 851 | |
| }, | |
| { | |
| "epoch": 0.6410835214446953, | |
| "grad_norm": 0.0026829675771296024, | |
| "learning_rate": 5.711431652401227e-06, | |
| "loss": 0.0001, | |
| "step": 852 | |
| }, | |
| { | |
| "epoch": 0.6418359668924003, | |
| "grad_norm": 0.1754864752292633, | |
| "learning_rate": 5.690089039554654e-06, | |
| "loss": 0.1013, | |
| "step": 853 | |
| }, | |
| { | |
| "epoch": 0.6425884123401053, | |
| "grad_norm": 0.07116664201021194, | |
| "learning_rate": 5.668770510112538e-06, | |
| "loss": 0.0032, | |
| "step": 854 | |
| }, | |
| { | |
| "epoch": 0.6433408577878104, | |
| "grad_norm": 0.8461427092552185, | |
| "learning_rate": 5.6474761832009554e-06, | |
| "loss": 0.0517, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 0.6440933032355154, | |
| "grad_norm": 0.009436607360839844, | |
| "learning_rate": 5.626206177810735e-06, | |
| "loss": 0.0004, | |
| "step": 856 | |
| }, | |
| { | |
| "epoch": 0.6448457486832204, | |
| "grad_norm": 2.8411285877227783, | |
| "learning_rate": 5.604960612796805e-06, | |
| "loss": 0.3936, | |
| "step": 857 | |
| }, | |
| { | |
| "epoch": 0.6455981941309256, | |
| "grad_norm": 0.05091340094804764, | |
| "learning_rate": 5.583739606877516e-06, | |
| "loss": 0.0014, | |
| "step": 858 | |
| }, | |
| { | |
| "epoch": 0.6463506395786306, | |
| "grad_norm": 0.00600972305983305, | |
| "learning_rate": 5.562543278633988e-06, | |
| "loss": 0.0002, | |
| "step": 859 | |
| }, | |
| { | |
| "epoch": 0.6471030850263356, | |
| "grad_norm": 0.0064705973491072655, | |
| "learning_rate": 5.541371746509448e-06, | |
| "loss": 0.0003, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.6478555304740407, | |
| "grad_norm": 5.701414585113525, | |
| "learning_rate": 5.520225128808555e-06, | |
| "loss": 0.0259, | |
| "step": 861 | |
| }, | |
| { | |
| "epoch": 0.6486079759217457, | |
| "grad_norm": 0.021785501390695572, | |
| "learning_rate": 5.4991035436967585e-06, | |
| "loss": 0.0009, | |
| "step": 862 | |
| }, | |
| { | |
| "epoch": 0.6493604213694507, | |
| "grad_norm": 0.017925115302205086, | |
| "learning_rate": 5.478007109199624e-06, | |
| "loss": 0.0008, | |
| "step": 863 | |
| }, | |
| { | |
| "epoch": 0.6501128668171557, | |
| "grad_norm": 0.012886104173958302, | |
| "learning_rate": 5.456935943202177e-06, | |
| "loss": 0.0005, | |
| "step": 864 | |
| }, | |
| { | |
| "epoch": 0.6508653122648608, | |
| "grad_norm": 0.015779219567775726, | |
| "learning_rate": 5.4358901634482404e-06, | |
| "loss": 0.0006, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 0.6516177577125658, | |
| "grad_norm": 0.05147948116064072, | |
| "learning_rate": 5.4148698875397905e-06, | |
| "loss": 0.0025, | |
| "step": 866 | |
| }, | |
| { | |
| "epoch": 0.6523702031602708, | |
| "grad_norm": 0.019827265292406082, | |
| "learning_rate": 5.393875232936283e-06, | |
| "loss": 0.0007, | |
| "step": 867 | |
| }, | |
| { | |
| "epoch": 0.653122648607976, | |
| "grad_norm": 0.01970357820391655, | |
| "learning_rate": 5.372906316954005e-06, | |
| "loss": 0.0006, | |
| "step": 868 | |
| }, | |
| { | |
| "epoch": 0.653875094055681, | |
| "grad_norm": 8.924382209777832, | |
| "learning_rate": 5.351963256765426e-06, | |
| "loss": 0.1568, | |
| "step": 869 | |
| }, | |
| { | |
| "epoch": 0.654627539503386, | |
| "grad_norm": 1.7220969200134277, | |
| "learning_rate": 5.33104616939853e-06, | |
| "loss": 0.0397, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.6553799849510911, | |
| "grad_norm": 0.3469778895378113, | |
| "learning_rate": 5.3101551717361586e-06, | |
| "loss": 0.0112, | |
| "step": 871 | |
| }, | |
| { | |
| "epoch": 0.6561324303987961, | |
| "grad_norm": 0.3850734233856201, | |
| "learning_rate": 5.2892903805153795e-06, | |
| "loss": 0.1255, | |
| "step": 872 | |
| }, | |
| { | |
| "epoch": 0.6568848758465011, | |
| "grad_norm": 2.2157397270202637, | |
| "learning_rate": 5.2684519123268155e-06, | |
| "loss": 0.3644, | |
| "step": 873 | |
| }, | |
| { | |
| "epoch": 0.6576373212942062, | |
| "grad_norm": 0.011840404942631721, | |
| "learning_rate": 5.247639883613999e-06, | |
| "loss": 0.0005, | |
| "step": 874 | |
| }, | |
| { | |
| "epoch": 0.6583897667419112, | |
| "grad_norm": 0.02047601528465748, | |
| "learning_rate": 5.226854410672724e-06, | |
| "loss": 0.0008, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.6591422121896162, | |
| "grad_norm": 4.430492401123047, | |
| "learning_rate": 5.2060956096503854e-06, | |
| "loss": 0.1308, | |
| "step": 876 | |
| }, | |
| { | |
| "epoch": 0.6598946576373212, | |
| "grad_norm": 0.004179192706942558, | |
| "learning_rate": 5.1853635965453495e-06, | |
| "loss": 0.0002, | |
| "step": 877 | |
| }, | |
| { | |
| "epoch": 0.6606471030850264, | |
| "grad_norm": 0.3641025424003601, | |
| "learning_rate": 5.164658487206275e-06, | |
| "loss": 0.1212, | |
| "step": 878 | |
| }, | |
| { | |
| "epoch": 0.6613995485327314, | |
| "grad_norm": 0.061624638736248016, | |
| "learning_rate": 5.143980397331512e-06, | |
| "loss": 0.0025, | |
| "step": 879 | |
| }, | |
| { | |
| "epoch": 0.6621519939804364, | |
| "grad_norm": 5.032725811004639, | |
| "learning_rate": 5.123329442468403e-06, | |
| "loss": 0.035, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.6629044394281415, | |
| "grad_norm": 1.7214125394821167, | |
| "learning_rate": 5.102705738012676e-06, | |
| "loss": 0.0527, | |
| "step": 881 | |
| }, | |
| { | |
| "epoch": 0.6636568848758465, | |
| "grad_norm": 0.7142964005470276, | |
| "learning_rate": 5.082109399207784e-06, | |
| "loss": 0.1174, | |
| "step": 882 | |
| }, | |
| { | |
| "epoch": 0.6644093303235515, | |
| "grad_norm": 0.062142811715602875, | |
| "learning_rate": 5.061540541144265e-06, | |
| "loss": 0.0029, | |
| "step": 883 | |
| }, | |
| { | |
| "epoch": 0.6651617757712566, | |
| "grad_norm": 0.2744309902191162, | |
| "learning_rate": 5.0409992787590845e-06, | |
| "loss": 0.0093, | |
| "step": 884 | |
| }, | |
| { | |
| "epoch": 0.6659142212189616, | |
| "grad_norm": 0.059226468205451965, | |
| "learning_rate": 5.02048572683502e-06, | |
| "loss": 0.0028, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 0.6666666666666666, | |
| "grad_norm": 0.007528245449066162, | |
| "learning_rate": 5.000000000000003e-06, | |
| "loss": 0.0003, | |
| "step": 886 | |
| }, | |
| { | |
| "epoch": 0.6674191121143717, | |
| "grad_norm": 2.5258829593658447, | |
| "learning_rate": 4.979542212726474e-06, | |
| "loss": 0.2801, | |
| "step": 887 | |
| }, | |
| { | |
| "epoch": 0.6681715575620768, | |
| "grad_norm": 3.617130756378174, | |
| "learning_rate": 4.959112479330753e-06, | |
| "loss": 0.1343, | |
| "step": 888 | |
| }, | |
| { | |
| "epoch": 0.6689240030097818, | |
| "grad_norm": 0.3309485912322998, | |
| "learning_rate": 4.93871091397241e-06, | |
| "loss": 0.0082, | |
| "step": 889 | |
| }, | |
| { | |
| "epoch": 0.6696764484574869, | |
| "grad_norm": 0.031886566430330276, | |
| "learning_rate": 4.9183376306535904e-06, | |
| "loss": 0.0009, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.6704288939051919, | |
| "grad_norm": 0.04454849287867546, | |
| "learning_rate": 4.897992743218419e-06, | |
| "loss": 0.0019, | |
| "step": 891 | |
| }, | |
| { | |
| "epoch": 0.6711813393528969, | |
| "grad_norm": 0.19051343202590942, | |
| "learning_rate": 4.877676365352343e-06, | |
| "loss": 0.1, | |
| "step": 892 | |
| }, | |
| { | |
| "epoch": 0.671933784800602, | |
| "grad_norm": 2.1760129928588867, | |
| "learning_rate": 4.857388610581499e-06, | |
| "loss": 0.1941, | |
| "step": 893 | |
| }, | |
| { | |
| "epoch": 0.672686230248307, | |
| "grad_norm": 0.020949775353074074, | |
| "learning_rate": 4.837129592272083e-06, | |
| "loss": 0.0008, | |
| "step": 894 | |
| }, | |
| { | |
| "epoch": 0.673438675696012, | |
| "grad_norm": 0.08075495064258575, | |
| "learning_rate": 4.81689942362971e-06, | |
| "loss": 0.0042, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 0.674191121143717, | |
| "grad_norm": 0.004490719176828861, | |
| "learning_rate": 4.796698217698791e-06, | |
| "loss": 0.0002, | |
| "step": 896 | |
| }, | |
| { | |
| "epoch": 0.6749435665914221, | |
| "grad_norm": 0.04653012752532959, | |
| "learning_rate": 4.776526087361896e-06, | |
| "loss": 0.0022, | |
| "step": 897 | |
| }, | |
| { | |
| "epoch": 0.6756960120391272, | |
| "grad_norm": 0.009343601763248444, | |
| "learning_rate": 4.756383145339107e-06, | |
| "loss": 0.0003, | |
| "step": 898 | |
| }, | |
| { | |
| "epoch": 0.6764484574868322, | |
| "grad_norm": 0.12858974933624268, | |
| "learning_rate": 4.736269504187431e-06, | |
| "loss": 0.0055, | |
| "step": 899 | |
| }, | |
| { | |
| "epoch": 0.6772009029345373, | |
| "grad_norm": 0.020633621141314507, | |
| "learning_rate": 4.716185276300126e-06, | |
| "loss": 0.0009, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.6779533483822423, | |
| "grad_norm": 0.023468296974897385, | |
| "learning_rate": 4.696130573906096e-06, | |
| "loss": 0.0009, | |
| "step": 901 | |
| }, | |
| { | |
| "epoch": 0.6787057938299473, | |
| "grad_norm": 1.5509250164031982, | |
| "learning_rate": 4.676105509069263e-06, | |
| "loss": 0.0409, | |
| "step": 902 | |
| }, | |
| { | |
| "epoch": 0.6794582392776524, | |
| "grad_norm": 0.607180118560791, | |
| "learning_rate": 4.656110193687925e-06, | |
| "loss": 0.0688, | |
| "step": 903 | |
| }, | |
| { | |
| "epoch": 0.6802106847253574, | |
| "grad_norm": 0.04802005738019943, | |
| "learning_rate": 4.636144739494156e-06, | |
| "loss": 0.0021, | |
| "step": 904 | |
| }, | |
| { | |
| "epoch": 0.6809631301730624, | |
| "grad_norm": 0.009008445776998997, | |
| "learning_rate": 4.616209258053163e-06, | |
| "loss": 0.0003, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 0.6817155756207675, | |
| "grad_norm": 0.02914116531610489, | |
| "learning_rate": 4.5963038607626655e-06, | |
| "loss": 0.0009, | |
| "step": 906 | |
| }, | |
| { | |
| "epoch": 0.6824680210684725, | |
| "grad_norm": 1.7884653806686401, | |
| "learning_rate": 4.57642865885228e-06, | |
| "loss": 0.3394, | |
| "step": 907 | |
| }, | |
| { | |
| "epoch": 0.6832204665161776, | |
| "grad_norm": 0.3303041160106659, | |
| "learning_rate": 4.5565837633828904e-06, | |
| "loss": 0.1329, | |
| "step": 908 | |
| }, | |
| { | |
| "epoch": 0.6839729119638827, | |
| "grad_norm": 0.06692855805158615, | |
| "learning_rate": 4.536769285246033e-06, | |
| "loss": 0.003, | |
| "step": 909 | |
| }, | |
| { | |
| "epoch": 0.6847253574115877, | |
| "grad_norm": 0.04765620082616806, | |
| "learning_rate": 4.516985335163274e-06, | |
| "loss": 0.0022, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.6854778028592927, | |
| "grad_norm": 0.18733732402324677, | |
| "learning_rate": 4.4972320236855916e-06, | |
| "loss": 0.0051, | |
| "step": 911 | |
| }, | |
| { | |
| "epoch": 0.6862302483069977, | |
| "grad_norm": 0.07214639335870743, | |
| "learning_rate": 4.477509461192756e-06, | |
| "loss": 0.0036, | |
| "step": 912 | |
| }, | |
| { | |
| "epoch": 0.6869826937547028, | |
| "grad_norm": 0.23084872961044312, | |
| "learning_rate": 4.457817757892718e-06, | |
| "loss": 0.1103, | |
| "step": 913 | |
| }, | |
| { | |
| "epoch": 0.6877351392024078, | |
| "grad_norm": 0.3813030421733856, | |
| "learning_rate": 4.438157023820991e-06, | |
| "loss": 0.0089, | |
| "step": 914 | |
| }, | |
| { | |
| "epoch": 0.6884875846501128, | |
| "grad_norm": 0.15035146474838257, | |
| "learning_rate": 4.4185273688400274e-06, | |
| "loss": 0.0044, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 0.6892400300978179, | |
| "grad_norm": 0.00839236006140709, | |
| "learning_rate": 4.398928902638626e-06, | |
| "loss": 0.0003, | |
| "step": 916 | |
| }, | |
| { | |
| "epoch": 0.6899924755455229, | |
| "grad_norm": 0.14042149484157562, | |
| "learning_rate": 4.379361734731289e-06, | |
| "loss": 0.0057, | |
| "step": 917 | |
| }, | |
| { | |
| "epoch": 0.690744920993228, | |
| "grad_norm": 0.3208416700363159, | |
| "learning_rate": 4.359825974457632e-06, | |
| "loss": 0.1172, | |
| "step": 918 | |
| }, | |
| { | |
| "epoch": 0.6914973664409331, | |
| "grad_norm": 0.1413826197385788, | |
| "learning_rate": 4.340321730981779e-06, | |
| "loss": 0.0033, | |
| "step": 919 | |
| }, | |
| { | |
| "epoch": 0.6922498118886381, | |
| "grad_norm": 0.09464599937200546, | |
| "learning_rate": 4.32084911329173e-06, | |
| "loss": 0.0045, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.6930022573363431, | |
| "grad_norm": 0.04123972728848457, | |
| "learning_rate": 4.301408230198763e-06, | |
| "loss": 0.0015, | |
| "step": 921 | |
| }, | |
| { | |
| "epoch": 0.6937547027840482, | |
| "grad_norm": 0.5074460506439209, | |
| "learning_rate": 4.28199919033683e-06, | |
| "loss": 0.0119, | |
| "step": 922 | |
| }, | |
| { | |
| "epoch": 0.6945071482317532, | |
| "grad_norm": 0.052684567868709564, | |
| "learning_rate": 4.2626221021619396e-06, | |
| "loss": 0.0027, | |
| "step": 923 | |
| }, | |
| { | |
| "epoch": 0.6952595936794582, | |
| "grad_norm": 0.016798803582787514, | |
| "learning_rate": 4.243277073951562e-06, | |
| "loss": 0.0008, | |
| "step": 924 | |
| }, | |
| { | |
| "epoch": 0.6960120391271633, | |
| "grad_norm": 0.006241548340767622, | |
| "learning_rate": 4.223964213804019e-06, | |
| "loss": 0.0002, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.6967644845748683, | |
| "grad_norm": 3.114851474761963, | |
| "learning_rate": 4.204683629637881e-06, | |
| "loss": 0.32, | |
| "step": 926 | |
| }, | |
| { | |
| "epoch": 0.6975169300225733, | |
| "grad_norm": 0.23592227697372437, | |
| "learning_rate": 4.1854354291913594e-06, | |
| "loss": 0.0053, | |
| "step": 927 | |
| }, | |
| { | |
| "epoch": 0.6982693754702785, | |
| "grad_norm": 0.7205729484558105, | |
| "learning_rate": 4.1662197200217116e-06, | |
| "loss": 0.0056, | |
| "step": 928 | |
| }, | |
| { | |
| "epoch": 0.6990218209179835, | |
| "grad_norm": 0.0018878192640841007, | |
| "learning_rate": 4.147036609504633e-06, | |
| "loss": 0.0001, | |
| "step": 929 | |
| }, | |
| { | |
| "epoch": 0.6997742663656885, | |
| "grad_norm": 0.24593710899353027, | |
| "learning_rate": 4.1278862048336645e-06, | |
| "loss": 0.0058, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.7005267118133935, | |
| "grad_norm": 0.035347383469343185, | |
| "learning_rate": 4.108768613019588e-06, | |
| "loss": 0.0015, | |
| "step": 931 | |
| }, | |
| { | |
| "epoch": 0.7012791572610986, | |
| "grad_norm": 0.15606430172920227, | |
| "learning_rate": 4.089683940889829e-06, | |
| "loss": 0.0069, | |
| "step": 932 | |
| }, | |
| { | |
| "epoch": 0.7020316027088036, | |
| "grad_norm": 1.3364535570144653, | |
| "learning_rate": 4.070632295087863e-06, | |
| "loss": 0.0344, | |
| "step": 933 | |
| }, | |
| { | |
| "epoch": 0.7027840481565086, | |
| "grad_norm": 0.1990508735179901, | |
| "learning_rate": 4.051613782072614e-06, | |
| "loss": 0.0034, | |
| "step": 934 | |
| }, | |
| { | |
| "epoch": 0.7035364936042137, | |
| "grad_norm": 0.004956225864589214, | |
| "learning_rate": 4.0326285081178695e-06, | |
| "loss": 0.0002, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 0.7042889390519187, | |
| "grad_norm": 0.09243706613779068, | |
| "learning_rate": 4.013676579311668e-06, | |
| "loss": 0.0032, | |
| "step": 936 | |
| }, | |
| { | |
| "epoch": 0.7050413844996237, | |
| "grad_norm": 0.2956998944282532, | |
| "learning_rate": 3.994758101555729e-06, | |
| "loss": 0.0037, | |
| "step": 937 | |
| }, | |
| { | |
| "epoch": 0.7057938299473289, | |
| "grad_norm": 6.872308254241943, | |
| "learning_rate": 3.975873180564843e-06, | |
| "loss": 0.185, | |
| "step": 938 | |
| }, | |
| { | |
| "epoch": 0.7065462753950339, | |
| "grad_norm": 5.041158676147461, | |
| "learning_rate": 3.957021921866301e-06, | |
| "loss": 0.1589, | |
| "step": 939 | |
| }, | |
| { | |
| "epoch": 0.7072987208427389, | |
| "grad_norm": 0.004548253491520882, | |
| "learning_rate": 3.938204430799278e-06, | |
| "loss": 0.0002, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.708051166290444, | |
| "grad_norm": 4.161418437957764, | |
| "learning_rate": 3.919420812514267e-06, | |
| "loss": 0.0773, | |
| "step": 941 | |
| }, | |
| { | |
| "epoch": 0.708803611738149, | |
| "grad_norm": 8.64138412475586, | |
| "learning_rate": 3.9006711719724755e-06, | |
| "loss": 0.3975, | |
| "step": 942 | |
| }, | |
| { | |
| "epoch": 0.709556057185854, | |
| "grad_norm": 0.10942798107862473, | |
| "learning_rate": 3.881955613945251e-06, | |
| "loss": 0.0037, | |
| "step": 943 | |
| }, | |
| { | |
| "epoch": 0.710308502633559, | |
| "grad_norm": 0.07507241517305374, | |
| "learning_rate": 3.8632742430134905e-06, | |
| "loss": 0.002, | |
| "step": 944 | |
| }, | |
| { | |
| "epoch": 0.7110609480812641, | |
| "grad_norm": 0.020177626982331276, | |
| "learning_rate": 3.844627163567059e-06, | |
| "loss": 0.0008, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 0.7118133935289691, | |
| "grad_norm": 6.14890718460083, | |
| "learning_rate": 3.826014479804198e-06, | |
| "loss": 0.0767, | |
| "step": 946 | |
| }, | |
| { | |
| "epoch": 0.7125658389766741, | |
| "grad_norm": 0.330617755651474, | |
| "learning_rate": 3.807436295730953e-06, | |
| "loss": 0.0074, | |
| "step": 947 | |
| }, | |
| { | |
| "epoch": 0.7133182844243793, | |
| "grad_norm": 0.07356506586074829, | |
| "learning_rate": 3.788892715160588e-06, | |
| "loss": 0.0027, | |
| "step": 948 | |
| }, | |
| { | |
| "epoch": 0.7140707298720843, | |
| "grad_norm": 5.51183557510376, | |
| "learning_rate": 3.7703838417130045e-06, | |
| "loss": 0.1942, | |
| "step": 949 | |
| }, | |
| { | |
| "epoch": 0.7148231753197893, | |
| "grad_norm": 0.07540974020957947, | |
| "learning_rate": 3.7519097788141635e-06, | |
| "loss": 0.0032, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.7155756207674944, | |
| "grad_norm": 0.1256755292415619, | |
| "learning_rate": 3.7334706296955093e-06, | |
| "loss": 0.0027, | |
| "step": 951 | |
| }, | |
| { | |
| "epoch": 0.7163280662151994, | |
| "grad_norm": 0.0014390931464731693, | |
| "learning_rate": 3.7150664973933893e-06, | |
| "loss": 0.0001, | |
| "step": 952 | |
| }, | |
| { | |
| "epoch": 0.7170805116629044, | |
| "grad_norm": 0.19743457436561584, | |
| "learning_rate": 3.6966974847484805e-06, | |
| "loss": 0.0047, | |
| "step": 953 | |
| }, | |
| { | |
| "epoch": 0.7178329571106095, | |
| "grad_norm": 1.814272165298462, | |
| "learning_rate": 3.6783636944052193e-06, | |
| "loss": 0.2478, | |
| "step": 954 | |
| }, | |
| { | |
| "epoch": 0.7185854025583145, | |
| "grad_norm": 0.35399460792541504, | |
| "learning_rate": 3.66006522881121e-06, | |
| "loss": 0.0039, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 0.7193378480060195, | |
| "grad_norm": 0.6553662419319153, | |
| "learning_rate": 3.641802190216678e-06, | |
| "loss": 0.0906, | |
| "step": 956 | |
| }, | |
| { | |
| "epoch": 0.7200902934537246, | |
| "grad_norm": 0.005217390134930611, | |
| "learning_rate": 3.623574680673879e-06, | |
| "loss": 0.0002, | |
| "step": 957 | |
| }, | |
| { | |
| "epoch": 0.7208427389014297, | |
| "grad_norm": 0.006294188555330038, | |
| "learning_rate": 3.605382802036538e-06, | |
| "loss": 0.0003, | |
| "step": 958 | |
| }, | |
| { | |
| "epoch": 0.7215951843491347, | |
| "grad_norm": 4.843064785003662, | |
| "learning_rate": 3.5872266559592817e-06, | |
| "loss": 0.3738, | |
| "step": 959 | |
| }, | |
| { | |
| "epoch": 0.7223476297968398, | |
| "grad_norm": 0.020986851304769516, | |
| "learning_rate": 3.5691063438970618e-06, | |
| "loss": 0.0008, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.7231000752445448, | |
| "grad_norm": 4.1812334060668945, | |
| "learning_rate": 3.5510219671045875e-06, | |
| "loss": 0.4018, | |
| "step": 961 | |
| }, | |
| { | |
| "epoch": 0.7238525206922498, | |
| "grad_norm": 0.0036870623007416725, | |
| "learning_rate": 3.532973626635773e-06, | |
| "loss": 0.0001, | |
| "step": 962 | |
| }, | |
| { | |
| "epoch": 0.7246049661399548, | |
| "grad_norm": 0.014761660248041153, | |
| "learning_rate": 3.5149614233431616e-06, | |
| "loss": 0.0007, | |
| "step": 963 | |
| }, | |
| { | |
| "epoch": 0.7253574115876599, | |
| "grad_norm": 0.03565015643835068, | |
| "learning_rate": 3.4969854578773667e-06, | |
| "loss": 0.0016, | |
| "step": 964 | |
| }, | |
| { | |
| "epoch": 0.7261098570353649, | |
| "grad_norm": 0.24551992118358612, | |
| "learning_rate": 3.479045830686506e-06, | |
| "loss": 0.006, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 0.7268623024830699, | |
| "grad_norm": 0.03166704997420311, | |
| "learning_rate": 3.4611426420156422e-06, | |
| "loss": 0.0013, | |
| "step": 966 | |
| }, | |
| { | |
| "epoch": 0.7276147479307751, | |
| "grad_norm": 0.4878341257572174, | |
| "learning_rate": 3.4432759919062253e-06, | |
| "loss": 0.1401, | |
| "step": 967 | |
| }, | |
| { | |
| "epoch": 0.7283671933784801, | |
| "grad_norm": 8.347463607788086, | |
| "learning_rate": 3.4254459801955276e-06, | |
| "loss": 0.0622, | |
| "step": 968 | |
| }, | |
| { | |
| "epoch": 0.7291196388261851, | |
| "grad_norm": 0.05901602655649185, | |
| "learning_rate": 3.4076527065160914e-06, | |
| "loss": 0.0025, | |
| "step": 969 | |
| }, | |
| { | |
| "epoch": 0.7298720842738902, | |
| "grad_norm": 0.1788146197795868, | |
| "learning_rate": 3.3898962702951687e-06, | |
| "loss": 0.0094, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.7306245297215952, | |
| "grad_norm": 0.008673246949911118, | |
| "learning_rate": 3.3721767707541696e-06, | |
| "loss": 0.0003, | |
| "step": 971 | |
| }, | |
| { | |
| "epoch": 0.7313769751693002, | |
| "grad_norm": 0.05570625886321068, | |
| "learning_rate": 3.3544943069081025e-06, | |
| "loss": 0.0022, | |
| "step": 972 | |
| }, | |
| { | |
| "epoch": 0.7321294206170053, | |
| "grad_norm": 0.04295314475893974, | |
| "learning_rate": 3.3368489775650282e-06, | |
| "loss": 0.0014, | |
| "step": 973 | |
| }, | |
| { | |
| "epoch": 0.7328818660647103, | |
| "grad_norm": 0.025052128359675407, | |
| "learning_rate": 3.3192408813254918e-06, | |
| "loss": 0.0009, | |
| "step": 974 | |
| }, | |
| { | |
| "epoch": 0.7336343115124153, | |
| "grad_norm": 5.636648178100586, | |
| "learning_rate": 3.3016701165819943e-06, | |
| "loss": 0.1106, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.7343867569601203, | |
| "grad_norm": 0.43065962195396423, | |
| "learning_rate": 3.28413678151843e-06, | |
| "loss": 0.0105, | |
| "step": 976 | |
| }, | |
| { | |
| "epoch": 0.7351392024078255, | |
| "grad_norm": 0.006689665839076042, | |
| "learning_rate": 3.2666409741095328e-06, | |
| "loss": 0.0002, | |
| "step": 977 | |
| }, | |
| { | |
| "epoch": 0.7358916478555305, | |
| "grad_norm": 0.009542127139866352, | |
| "learning_rate": 3.2491827921203456e-06, | |
| "loss": 0.0004, | |
| "step": 978 | |
| }, | |
| { | |
| "epoch": 0.7366440933032355, | |
| "grad_norm": 2.010205030441284, | |
| "learning_rate": 3.231762333105661e-06, | |
| "loss": 0.0458, | |
| "step": 979 | |
| }, | |
| { | |
| "epoch": 0.7373965387509406, | |
| "grad_norm": 0.015615508891642094, | |
| "learning_rate": 3.2143796944094675e-06, | |
| "loss": 0.0006, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.7381489841986456, | |
| "grad_norm": 0.6981082558631897, | |
| "learning_rate": 3.197034973164429e-06, | |
| "loss": 0.0136, | |
| "step": 981 | |
| }, | |
| { | |
| "epoch": 0.7389014296463506, | |
| "grad_norm": 0.6512033939361572, | |
| "learning_rate": 3.1797282662913277e-06, | |
| "loss": 0.0279, | |
| "step": 982 | |
| }, | |
| { | |
| "epoch": 0.7396538750940557, | |
| "grad_norm": 1.36346435546875, | |
| "learning_rate": 3.162459670498523e-06, | |
| "loss": 0.1529, | |
| "step": 983 | |
| }, | |
| { | |
| "epoch": 0.7404063205417607, | |
| "grad_norm": 0.0437924787402153, | |
| "learning_rate": 3.1452292822814145e-06, | |
| "loss": 0.0021, | |
| "step": 984 | |
| }, | |
| { | |
| "epoch": 0.7411587659894657, | |
| "grad_norm": 0.21114251017570496, | |
| "learning_rate": 3.1280371979218993e-06, | |
| "loss": 0.0053, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 0.7419112114371708, | |
| "grad_norm": 0.0944240614771843, | |
| "learning_rate": 3.1108835134878367e-06, | |
| "loss": 0.0029, | |
| "step": 986 | |
| }, | |
| { | |
| "epoch": 0.7426636568848759, | |
| "grad_norm": 0.044163547456264496, | |
| "learning_rate": 3.0937683248325133e-06, | |
| "loss": 0.0013, | |
| "step": 987 | |
| }, | |
| { | |
| "epoch": 0.7434161023325809, | |
| "grad_norm": 0.056986529380083084, | |
| "learning_rate": 3.0766917275941e-06, | |
| "loss": 0.0022, | |
| "step": 988 | |
| }, | |
| { | |
| "epoch": 0.744168547780286, | |
| "grad_norm": 0.039454780519008636, | |
| "learning_rate": 3.0596538171951252e-06, | |
| "loss": 0.0016, | |
| "step": 989 | |
| }, | |
| { | |
| "epoch": 0.744920993227991, | |
| "grad_norm": 0.017716459929943085, | |
| "learning_rate": 3.0426546888419385e-06, | |
| "loss": 0.0006, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.745673438675696, | |
| "grad_norm": 2.3982226848602295, | |
| "learning_rate": 3.025694437524177e-06, | |
| "loss": 0.2701, | |
| "step": 991 | |
| }, | |
| { | |
| "epoch": 0.746425884123401, | |
| "grad_norm": 0.17275047302246094, | |
| "learning_rate": 3.008773158014242e-06, | |
| "loss": 0.0043, | |
| "step": 992 | |
| }, | |
| { | |
| "epoch": 0.7471783295711061, | |
| "grad_norm": 0.028289055451750755, | |
| "learning_rate": 2.991890944866752e-06, | |
| "loss": 0.0013, | |
| "step": 993 | |
| }, | |
| { | |
| "epoch": 0.7479307750188111, | |
| "grad_norm": 2.991833448410034, | |
| "learning_rate": 2.9750478924180383e-06, | |
| "loss": 0.0494, | |
| "step": 994 | |
| }, | |
| { | |
| "epoch": 0.7486832204665161, | |
| "grad_norm": 0.15955565869808197, | |
| "learning_rate": 2.9582440947855993e-06, | |
| "loss": 0.0054, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 0.7494356659142212, | |
| "grad_norm": 0.23433822393417358, | |
| "learning_rate": 2.941479645867583e-06, | |
| "loss": 0.0084, | |
| "step": 996 | |
| }, | |
| { | |
| "epoch": 0.7501881113619263, | |
| "grad_norm": 0.08780021965503693, | |
| "learning_rate": 2.9247546393422566e-06, | |
| "loss": 0.004, | |
| "step": 997 | |
| }, | |
| { | |
| "epoch": 0.7509405568096313, | |
| "grad_norm": 0.13379798829555511, | |
| "learning_rate": 2.9080691686674977e-06, | |
| "loss": 0.0026, | |
| "step": 998 | |
| }, | |
| { | |
| "epoch": 0.7516930022573364, | |
| "grad_norm": 0.01313999854028225, | |
| "learning_rate": 2.891423327080246e-06, | |
| "loss": 0.0006, | |
| "step": 999 | |
| }, | |
| { | |
| "epoch": 0.7524454477050414, | |
| "grad_norm": 0.03100876323878765, | |
| "learning_rate": 2.874817207596007e-06, | |
| "loss": 0.0012, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.7531978931527464, | |
| "grad_norm": 0.018929272890090942, | |
| "learning_rate": 2.8582509030083184e-06, | |
| "loss": 0.0007, | |
| "step": 1001 | |
| }, | |
| { | |
| "epoch": 0.7539503386004515, | |
| "grad_norm": 4.446130752563477, | |
| "learning_rate": 2.841724505888239e-06, | |
| "loss": 0.2995, | |
| "step": 1002 | |
| }, | |
| { | |
| "epoch": 0.7547027840481565, | |
| "grad_norm": 0.27879974246025085, | |
| "learning_rate": 2.8252381085838266e-06, | |
| "loss": 0.0064, | |
| "step": 1003 | |
| }, | |
| { | |
| "epoch": 0.7554552294958615, | |
| "grad_norm": 0.06390407681465149, | |
| "learning_rate": 2.8087918032196214e-06, | |
| "loss": 0.003, | |
| "step": 1004 | |
| }, | |
| { | |
| "epoch": 0.7562076749435666, | |
| "grad_norm": 0.05781351029872894, | |
| "learning_rate": 2.792385681696138e-06, | |
| "loss": 0.0029, | |
| "step": 1005 | |
| }, | |
| { | |
| "epoch": 0.7569601203912716, | |
| "grad_norm": 0.035376448184251785, | |
| "learning_rate": 2.7760198356893466e-06, | |
| "loss": 0.0017, | |
| "step": 1006 | |
| }, | |
| { | |
| "epoch": 0.7577125658389767, | |
| "grad_norm": 0.024740157648921013, | |
| "learning_rate": 2.759694356650149e-06, | |
| "loss": 0.0008, | |
| "step": 1007 | |
| }, | |
| { | |
| "epoch": 0.7584650112866818, | |
| "grad_norm": 0.5133036375045776, | |
| "learning_rate": 2.7434093358039003e-06, | |
| "loss": 0.0121, | |
| "step": 1008 | |
| }, | |
| { | |
| "epoch": 0.7592174567343868, | |
| "grad_norm": 0.017775315791368484, | |
| "learning_rate": 2.727164864149867e-06, | |
| "loss": 0.0007, | |
| "step": 1009 | |
| }, | |
| { | |
| "epoch": 0.7599699021820918, | |
| "grad_norm": 0.004680894315242767, | |
| "learning_rate": 2.7109610324607305e-06, | |
| "loss": 0.0002, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.7607223476297968, | |
| "grad_norm": 0.01725710742175579, | |
| "learning_rate": 2.6947979312820825e-06, | |
| "loss": 0.0008, | |
| "step": 1011 | |
| }, | |
| { | |
| "epoch": 0.7614747930775019, | |
| "grad_norm": 0.279935747385025, | |
| "learning_rate": 2.678675650931917e-06, | |
| "loss": 0.0039, | |
| "step": 1012 | |
| }, | |
| { | |
| "epoch": 0.7622272385252069, | |
| "grad_norm": 0.14282621443271637, | |
| "learning_rate": 2.662594281500115e-06, | |
| "loss": 0.0036, | |
| "step": 1013 | |
| }, | |
| { | |
| "epoch": 0.7629796839729119, | |
| "grad_norm": 0.011395514011383057, | |
| "learning_rate": 2.6465539128479646e-06, | |
| "loss": 0.0005, | |
| "step": 1014 | |
| }, | |
| { | |
| "epoch": 0.763732129420617, | |
| "grad_norm": 0.04269295185804367, | |
| "learning_rate": 2.630554634607637e-06, | |
| "loss": 0.0014, | |
| "step": 1015 | |
| }, | |
| { | |
| "epoch": 0.764484574868322, | |
| "grad_norm": 0.012843809090554714, | |
| "learning_rate": 2.614596536181697e-06, | |
| "loss": 0.0006, | |
| "step": 1016 | |
| }, | |
| { | |
| "epoch": 0.7652370203160271, | |
| "grad_norm": 0.018597368150949478, | |
| "learning_rate": 2.5986797067425972e-06, | |
| "loss": 0.0008, | |
| "step": 1017 | |
| }, | |
| { | |
| "epoch": 0.7659894657637322, | |
| "grad_norm": 0.019066110253334045, | |
| "learning_rate": 2.582804235232187e-06, | |
| "loss": 0.0008, | |
| "step": 1018 | |
| }, | |
| { | |
| "epoch": 0.7667419112114372, | |
| "grad_norm": 0.011421947740018368, | |
| "learning_rate": 2.566970210361208e-06, | |
| "loss": 0.0005, | |
| "step": 1019 | |
| }, | |
| { | |
| "epoch": 0.7674943566591422, | |
| "grad_norm": 0.016336999833583832, | |
| "learning_rate": 2.551177720608802e-06, | |
| "loss": 0.0007, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.7682468021068473, | |
| "grad_norm": 1.4327224493026733, | |
| "learning_rate": 2.5354268542220163e-06, | |
| "loss": 0.1008, | |
| "step": 1021 | |
| }, | |
| { | |
| "epoch": 0.7689992475545523, | |
| "grad_norm": 0.014406280592083931, | |
| "learning_rate": 2.5197176992153125e-06, | |
| "loss": 0.0004, | |
| "step": 1022 | |
| }, | |
| { | |
| "epoch": 0.7697516930022573, | |
| "grad_norm": 2.0984880924224854, | |
| "learning_rate": 2.5040503433700702e-06, | |
| "loss": 0.4668, | |
| "step": 1023 | |
| }, | |
| { | |
| "epoch": 0.7705041384499624, | |
| "grad_norm": 0.002694485941901803, | |
| "learning_rate": 2.4884248742340987e-06, | |
| "loss": 0.0001, | |
| "step": 1024 | |
| }, | |
| { | |
| "epoch": 0.7712565838976674, | |
| "grad_norm": 0.16212724149227142, | |
| "learning_rate": 2.472841379121154e-06, | |
| "loss": 0.0048, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.7720090293453724, | |
| "grad_norm": 2.619164228439331, | |
| "learning_rate": 2.457299945110433e-06, | |
| "loss": 0.2616, | |
| "step": 1026 | |
| }, | |
| { | |
| "epoch": 0.7727614747930776, | |
| "grad_norm": 0.003759812330827117, | |
| "learning_rate": 2.441800659046106e-06, | |
| "loss": 0.0001, | |
| "step": 1027 | |
| }, | |
| { | |
| "epoch": 0.7735139202407826, | |
| "grad_norm": 3.9366838932037354, | |
| "learning_rate": 2.4263436075368307e-06, | |
| "loss": 0.2472, | |
| "step": 1028 | |
| }, | |
| { | |
| "epoch": 0.7742663656884876, | |
| "grad_norm": 2.153656005859375, | |
| "learning_rate": 2.4109288769552518e-06, | |
| "loss": 0.1468, | |
| "step": 1029 | |
| }, | |
| { | |
| "epoch": 0.7750188111361926, | |
| "grad_norm": 0.3993730843067169, | |
| "learning_rate": 2.3955565534375326e-06, | |
| "loss": 0.1389, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.7757712565838977, | |
| "grad_norm": 0.0949854701757431, | |
| "learning_rate": 2.3802267228828703e-06, | |
| "loss": 0.0027, | |
| "step": 1031 | |
| }, | |
| { | |
| "epoch": 0.7765237020316027, | |
| "grad_norm": 1.1868627071380615, | |
| "learning_rate": 2.3649394709530093e-06, | |
| "loss": 0.0392, | |
| "step": 1032 | |
| }, | |
| { | |
| "epoch": 0.7772761474793077, | |
| "grad_norm": 0.0136722382158041, | |
| "learning_rate": 2.349694883071775e-06, | |
| "loss": 0.0006, | |
| "step": 1033 | |
| }, | |
| { | |
| "epoch": 0.7780285929270128, | |
| "grad_norm": 2.6067428588867188, | |
| "learning_rate": 2.3344930444245863e-06, | |
| "loss": 0.4598, | |
| "step": 1034 | |
| }, | |
| { | |
| "epoch": 0.7787810383747178, | |
| "grad_norm": 0.005388299468904734, | |
| "learning_rate": 2.3193340399579865e-06, | |
| "loss": 0.0002, | |
| "step": 1035 | |
| }, | |
| { | |
| "epoch": 0.7795334838224228, | |
| "grad_norm": 0.032653093338012695, | |
| "learning_rate": 2.304217954379162e-06, | |
| "loss": 0.0012, | |
| "step": 1036 | |
| }, | |
| { | |
| "epoch": 0.780285929270128, | |
| "grad_norm": 0.11140411347150803, | |
| "learning_rate": 2.2891448721554733e-06, | |
| "loss": 0.0052, | |
| "step": 1037 | |
| }, | |
| { | |
| "epoch": 0.781038374717833, | |
| "grad_norm": 0.6110525727272034, | |
| "learning_rate": 2.274114877513981e-06, | |
| "loss": 0.0083, | |
| "step": 1038 | |
| }, | |
| { | |
| "epoch": 0.781790820165538, | |
| "grad_norm": 0.3631162643432617, | |
| "learning_rate": 2.259128054440979e-06, | |
| "loss": 0.0101, | |
| "step": 1039 | |
| }, | |
| { | |
| "epoch": 0.782543265613243, | |
| "grad_norm": 0.5492317080497742, | |
| "learning_rate": 2.2441844866815188e-06, | |
| "loss": 0.0043, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.7832957110609481, | |
| "grad_norm": 0.29190927743911743, | |
| "learning_rate": 2.229284257738946e-06, | |
| "loss": 0.0063, | |
| "step": 1041 | |
| }, | |
| { | |
| "epoch": 0.7840481565086531, | |
| "grad_norm": 0.012482292018830776, | |
| "learning_rate": 2.2144274508744355e-06, | |
| "loss": 0.0005, | |
| "step": 1042 | |
| }, | |
| { | |
| "epoch": 0.7848006019563581, | |
| "grad_norm": 0.011028854176402092, | |
| "learning_rate": 2.199614149106519e-06, | |
| "loss": 0.0004, | |
| "step": 1043 | |
| }, | |
| { | |
| "epoch": 0.7855530474040632, | |
| "grad_norm": 0.11335982382297516, | |
| "learning_rate": 2.1848444352106314e-06, | |
| "loss": 0.0025, | |
| "step": 1044 | |
| }, | |
| { | |
| "epoch": 0.7863054928517682, | |
| "grad_norm": 0.17785978317260742, | |
| "learning_rate": 2.1701183917186317e-06, | |
| "loss": 0.0053, | |
| "step": 1045 | |
| }, | |
| { | |
| "epoch": 0.7870579382994732, | |
| "grad_norm": 0.0240810327231884, | |
| "learning_rate": 2.155436100918363e-06, | |
| "loss": 0.001, | |
| "step": 1046 | |
| }, | |
| { | |
| "epoch": 0.7878103837471784, | |
| "grad_norm": 0.048680830746889114, | |
| "learning_rate": 2.1407976448531776e-06, | |
| "loss": 0.0021, | |
| "step": 1047 | |
| }, | |
| { | |
| "epoch": 0.7885628291948834, | |
| "grad_norm": 0.03647977486252785, | |
| "learning_rate": 2.126203105321487e-06, | |
| "loss": 0.0012, | |
| "step": 1048 | |
| }, | |
| { | |
| "epoch": 0.7893152746425884, | |
| "grad_norm": 0.012947115115821362, | |
| "learning_rate": 2.1116525638762963e-06, | |
| "loss": 0.0004, | |
| "step": 1049 | |
| }, | |
| { | |
| "epoch": 0.7900677200902935, | |
| "grad_norm": 0.015302013605833054, | |
| "learning_rate": 2.0971461018247586e-06, | |
| "loss": 0.0007, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.7908201655379985, | |
| "grad_norm": 0.9521199464797974, | |
| "learning_rate": 2.082683800227705e-06, | |
| "loss": 0.2254, | |
| "step": 1051 | |
| }, | |
| { | |
| "epoch": 0.7915726109857035, | |
| "grad_norm": 0.011116956360638142, | |
| "learning_rate": 2.0682657398992124e-06, | |
| "loss": 0.0004, | |
| "step": 1052 | |
| }, | |
| { | |
| "epoch": 0.7923250564334086, | |
| "grad_norm": 0.20475442707538605, | |
| "learning_rate": 2.053892001406136e-06, | |
| "loss": 0.1026, | |
| "step": 1053 | |
| }, | |
| { | |
| "epoch": 0.7930775018811136, | |
| "grad_norm": 0.026699619367718697, | |
| "learning_rate": 2.039562665067667e-06, | |
| "loss": 0.0013, | |
| "step": 1054 | |
| }, | |
| { | |
| "epoch": 0.7938299473288186, | |
| "grad_norm": 0.0033015466760843992, | |
| "learning_rate": 2.0252778109548785e-06, | |
| "loss": 0.0001, | |
| "step": 1055 | |
| }, | |
| { | |
| "epoch": 0.7945823927765236, | |
| "grad_norm": 1.792168140411377, | |
| "learning_rate": 2.0110375188902852e-06, | |
| "loss": 0.1835, | |
| "step": 1056 | |
| }, | |
| { | |
| "epoch": 0.7953348382242288, | |
| "grad_norm": 0.4291799068450928, | |
| "learning_rate": 1.996841868447388e-06, | |
| "loss": 0.1194, | |
| "step": 1057 | |
| }, | |
| { | |
| "epoch": 0.7960872836719338, | |
| "grad_norm": 18.169370651245117, | |
| "learning_rate": 1.9826909389502368e-06, | |
| "loss": 0.0432, | |
| "step": 1058 | |
| }, | |
| { | |
| "epoch": 0.7968397291196389, | |
| "grad_norm": 2.5589189529418945, | |
| "learning_rate": 1.9685848094729853e-06, | |
| "loss": 0.1892, | |
| "step": 1059 | |
| }, | |
| { | |
| "epoch": 0.7975921745673439, | |
| "grad_norm": 2.7977375984191895, | |
| "learning_rate": 1.9545235588394484e-06, | |
| "loss": 0.1353, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.7983446200150489, | |
| "grad_norm": 0.035498447716236115, | |
| "learning_rate": 1.940507265622661e-06, | |
| "loss": 0.0014, | |
| "step": 1061 | |
| }, | |
| { | |
| "epoch": 0.7990970654627539, | |
| "grad_norm": 0.04865005239844322, | |
| "learning_rate": 1.9265360081444385e-06, | |
| "loss": 0.0014, | |
| "step": 1062 | |
| }, | |
| { | |
| "epoch": 0.799849510910459, | |
| "grad_norm": 3.3623549938201904, | |
| "learning_rate": 1.9126098644749482e-06, | |
| "loss": 0.2956, | |
| "step": 1063 | |
| }, | |
| { | |
| "epoch": 0.800601956358164, | |
| "grad_norm": 0.05257292836904526, | |
| "learning_rate": 1.8987289124322517e-06, | |
| "loss": 0.0025, | |
| "step": 1064 | |
| }, | |
| { | |
| "epoch": 0.801354401805869, | |
| "grad_norm": 0.003795795841142535, | |
| "learning_rate": 1.8848932295818945e-06, | |
| "loss": 0.0001, | |
| "step": 1065 | |
| }, | |
| { | |
| "epoch": 0.8021068472535741, | |
| "grad_norm": 0.005308138206601143, | |
| "learning_rate": 1.8711028932364604e-06, | |
| "loss": 0.0002, | |
| "step": 1066 | |
| }, | |
| { | |
| "epoch": 0.8028592927012792, | |
| "grad_norm": 0.7294948697090149, | |
| "learning_rate": 1.8573579804551367e-06, | |
| "loss": 0.0244, | |
| "step": 1067 | |
| }, | |
| { | |
| "epoch": 0.8036117381489842, | |
| "grad_norm": 0.005934171844273806, | |
| "learning_rate": 1.8436585680432951e-06, | |
| "loss": 0.0002, | |
| "step": 1068 | |
| }, | |
| { | |
| "epoch": 0.8043641835966893, | |
| "grad_norm": 5.70277738571167, | |
| "learning_rate": 1.8300047325520508e-06, | |
| "loss": 0.1113, | |
| "step": 1069 | |
| }, | |
| { | |
| "epoch": 0.8051166290443943, | |
| "grad_norm": 0.813459038734436, | |
| "learning_rate": 1.8163965502778337e-06, | |
| "loss": 0.0269, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.8058690744920993, | |
| "grad_norm": 1.580662488937378, | |
| "learning_rate": 1.802834097261975e-06, | |
| "loss": 0.0958, | |
| "step": 1071 | |
| }, | |
| { | |
| "epoch": 0.8066215199398044, | |
| "grad_norm": 0.013537387363612652, | |
| "learning_rate": 1.7893174492902742e-06, | |
| "loss": 0.0005, | |
| "step": 1072 | |
| }, | |
| { | |
| "epoch": 0.8073739653875094, | |
| "grad_norm": 3.1393344402313232, | |
| "learning_rate": 1.7758466818925735e-06, | |
| "loss": 0.2465, | |
| "step": 1073 | |
| }, | |
| { | |
| "epoch": 0.8081264108352144, | |
| "grad_norm": 0.17277193069458008, | |
| "learning_rate": 1.7624218703423402e-06, | |
| "loss": 0.0045, | |
| "step": 1074 | |
| }, | |
| { | |
| "epoch": 0.8088788562829194, | |
| "grad_norm": 1.4239376783370972, | |
| "learning_rate": 1.7490430896562439e-06, | |
| "loss": 0.0138, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.8096313017306245, | |
| "grad_norm": 0.3208409249782562, | |
| "learning_rate": 1.7357104145937365e-06, | |
| "loss": 0.1175, | |
| "step": 1076 | |
| }, | |
| { | |
| "epoch": 0.8103837471783296, | |
| "grad_norm": 0.30378955602645874, | |
| "learning_rate": 1.7224239196566395e-06, | |
| "loss": 0.0059, | |
| "step": 1077 | |
| }, | |
| { | |
| "epoch": 0.8111361926260346, | |
| "grad_norm": 0.004885723814368248, | |
| "learning_rate": 1.7091836790887196e-06, | |
| "loss": 0.0002, | |
| "step": 1078 | |
| }, | |
| { | |
| "epoch": 0.8118886380737397, | |
| "grad_norm": 0.04553482308983803, | |
| "learning_rate": 1.695989766875279e-06, | |
| "loss": 0.0019, | |
| "step": 1079 | |
| }, | |
| { | |
| "epoch": 0.8126410835214447, | |
| "grad_norm": 0.12680576741695404, | |
| "learning_rate": 1.682842256742744e-06, | |
| "loss": 0.0056, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.8133935289691497, | |
| "grad_norm": 0.05921807140111923, | |
| "learning_rate": 1.6697412221582477e-06, | |
| "loss": 0.0026, | |
| "step": 1081 | |
| }, | |
| { | |
| "epoch": 0.8141459744168548, | |
| "grad_norm": 0.03163035586476326, | |
| "learning_rate": 1.6566867363292238e-06, | |
| "loss": 0.0014, | |
| "step": 1082 | |
| }, | |
| { | |
| "epoch": 0.8148984198645598, | |
| "grad_norm": 0.027834394946694374, | |
| "learning_rate": 1.6436788722029906e-06, | |
| "loss": 0.001, | |
| "step": 1083 | |
| }, | |
| { | |
| "epoch": 0.8156508653122648, | |
| "grad_norm": 0.3784581422805786, | |
| "learning_rate": 1.6307177024663534e-06, | |
| "loss": 0.1234, | |
| "step": 1084 | |
| }, | |
| { | |
| "epoch": 0.8164033107599699, | |
| "grad_norm": 0.09414243698120117, | |
| "learning_rate": 1.617803299545192e-06, | |
| "loss": 0.0038, | |
| "step": 1085 | |
| }, | |
| { | |
| "epoch": 0.8171557562076749, | |
| "grad_norm": 1.488474726676941, | |
| "learning_rate": 1.6049357356040584e-06, | |
| "loss": 0.1, | |
| "step": 1086 | |
| }, | |
| { | |
| "epoch": 0.81790820165538, | |
| "grad_norm": 0.0046142516657710075, | |
| "learning_rate": 1.5921150825457677e-06, | |
| "loss": 0.0002, | |
| "step": 1087 | |
| }, | |
| { | |
| "epoch": 0.8186606471030851, | |
| "grad_norm": 0.06056587025523186, | |
| "learning_rate": 1.579341412011014e-06, | |
| "loss": 0.0023, | |
| "step": 1088 | |
| }, | |
| { | |
| "epoch": 0.8194130925507901, | |
| "grad_norm": 4.677060127258301, | |
| "learning_rate": 1.5666147953779376e-06, | |
| "loss": 0.1317, | |
| "step": 1089 | |
| }, | |
| { | |
| "epoch": 0.8201655379984951, | |
| "grad_norm": 2.52014422416687, | |
| "learning_rate": 1.553935303761761e-06, | |
| "loss": 0.503, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.8209179834462002, | |
| "grad_norm": 0.023583421483635902, | |
| "learning_rate": 1.5413030080143708e-06, | |
| "loss": 0.0009, | |
| "step": 1091 | |
| }, | |
| { | |
| "epoch": 0.8216704288939052, | |
| "grad_norm": 0.2893773317337036, | |
| "learning_rate": 1.5287179787239282e-06, | |
| "loss": 0.0056, | |
| "step": 1092 | |
| }, | |
| { | |
| "epoch": 0.8224228743416102, | |
| "grad_norm": 1.8263130187988281, | |
| "learning_rate": 1.5161802862144715e-06, | |
| "loss": 0.0311, | |
| "step": 1093 | |
| }, | |
| { | |
| "epoch": 0.8231753197893152, | |
| "grad_norm": 0.2679802477359772, | |
| "learning_rate": 1.503690000545528e-06, | |
| "loss": 0.0109, | |
| "step": 1094 | |
| }, | |
| { | |
| "epoch": 0.8239277652370203, | |
| "grad_norm": 0.004781723953783512, | |
| "learning_rate": 1.4912471915117189e-06, | |
| "loss": 0.0002, | |
| "step": 1095 | |
| }, | |
| { | |
| "epoch": 0.8246802106847254, | |
| "grad_norm": 1.211321473121643, | |
| "learning_rate": 1.4788519286423687e-06, | |
| "loss": 0.0369, | |
| "step": 1096 | |
| }, | |
| { | |
| "epoch": 0.8254326561324304, | |
| "grad_norm": 0.03698015958070755, | |
| "learning_rate": 1.46650428120112e-06, | |
| "loss": 0.0013, | |
| "step": 1097 | |
| }, | |
| { | |
| "epoch": 0.8261851015801355, | |
| "grad_norm": 0.23283882439136505, | |
| "learning_rate": 1.4542043181855447e-06, | |
| "loss": 0.1022, | |
| "step": 1098 | |
| }, | |
| { | |
| "epoch": 0.8269375470278405, | |
| "grad_norm": 0.045804113149642944, | |
| "learning_rate": 1.441952108326755e-06, | |
| "loss": 0.0007, | |
| "step": 1099 | |
| }, | |
| { | |
| "epoch": 0.8276899924755455, | |
| "grad_norm": 0.08991260826587677, | |
| "learning_rate": 1.4297477200890275e-06, | |
| "loss": 0.0031, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.8284424379232506, | |
| "grad_norm": 0.059289537370204926, | |
| "learning_rate": 1.417591221669412e-06, | |
| "loss": 0.0013, | |
| "step": 1101 | |
| }, | |
| { | |
| "epoch": 0.8291948833709556, | |
| "grad_norm": 0.25663039088249207, | |
| "learning_rate": 1.4054826809973576e-06, | |
| "loss": 0.0031, | |
| "step": 1102 | |
| }, | |
| { | |
| "epoch": 0.8299473288186606, | |
| "grad_norm": 0.047977883368730545, | |
| "learning_rate": 1.393422165734325e-06, | |
| "loss": 0.0019, | |
| "step": 1103 | |
| }, | |
| { | |
| "epoch": 0.8306997742663657, | |
| "grad_norm": 0.17930328845977783, | |
| "learning_rate": 1.3814097432734154e-06, | |
| "loss": 0.0026, | |
| "step": 1104 | |
| }, | |
| { | |
| "epoch": 0.8314522197140707, | |
| "grad_norm": 0.05746970325708389, | |
| "learning_rate": 1.3694454807389935e-06, | |
| "loss": 0.0019, | |
| "step": 1105 | |
| }, | |
| { | |
| "epoch": 0.8322046651617758, | |
| "grad_norm": 0.011797560378909111, | |
| "learning_rate": 1.3575294449863063e-06, | |
| "loss": 0.0005, | |
| "step": 1106 | |
| }, | |
| { | |
| "epoch": 0.8329571106094809, | |
| "grad_norm": 0.00394852552562952, | |
| "learning_rate": 1.3456617026011233e-06, | |
| "loss": 0.0001, | |
| "step": 1107 | |
| }, | |
| { | |
| "epoch": 0.8337095560571859, | |
| "grad_norm": 0.07486367225646973, | |
| "learning_rate": 1.3338423198993422e-06, | |
| "loss": 0.003, | |
| "step": 1108 | |
| }, | |
| { | |
| "epoch": 0.8344620015048909, | |
| "grad_norm": 0.05970097705721855, | |
| "learning_rate": 1.322071362926638e-06, | |
| "loss": 0.0027, | |
| "step": 1109 | |
| }, | |
| { | |
| "epoch": 0.835214446952596, | |
| "grad_norm": 0.002417666371911764, | |
| "learning_rate": 1.3103488974580858e-06, | |
| "loss": 0.0001, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.835966892400301, | |
| "grad_norm": 0.029511481523513794, | |
| "learning_rate": 1.2986749889977968e-06, | |
| "loss": 0.0011, | |
| "step": 1111 | |
| }, | |
| { | |
| "epoch": 0.836719337848006, | |
| "grad_norm": 2.6889474391937256, | |
| "learning_rate": 1.2870497027785444e-06, | |
| "loss": 0.0219, | |
| "step": 1112 | |
| }, | |
| { | |
| "epoch": 0.837471783295711, | |
| "grad_norm": 0.015240820124745369, | |
| "learning_rate": 1.2754731037614122e-06, | |
| "loss": 0.0007, | |
| "step": 1113 | |
| }, | |
| { | |
| "epoch": 0.8382242287434161, | |
| "grad_norm": 0.01888788491487503, | |
| "learning_rate": 1.263945256635416e-06, | |
| "loss": 0.0008, | |
| "step": 1114 | |
| }, | |
| { | |
| "epoch": 0.8389766741911211, | |
| "grad_norm": 0.047674112021923065, | |
| "learning_rate": 1.2524662258171605e-06, | |
| "loss": 0.0022, | |
| "step": 1115 | |
| }, | |
| { | |
| "epoch": 0.8397291196388262, | |
| "grad_norm": 0.07721062004566193, | |
| "learning_rate": 1.2410360754504536e-06, | |
| "loss": 0.0032, | |
| "step": 1116 | |
| }, | |
| { | |
| "epoch": 0.8404815650865313, | |
| "grad_norm": 0.002638805890455842, | |
| "learning_rate": 1.2296548694059818e-06, | |
| "loss": 0.0001, | |
| "step": 1117 | |
| }, | |
| { | |
| "epoch": 0.8412340105342363, | |
| "grad_norm": 2.895000457763672, | |
| "learning_rate": 1.2183226712809238e-06, | |
| "loss": 0.2227, | |
| "step": 1118 | |
| }, | |
| { | |
| "epoch": 0.8419864559819413, | |
| "grad_norm": 4.1699957847595215, | |
| "learning_rate": 1.207039544398607e-06, | |
| "loss": 0.1277, | |
| "step": 1119 | |
| }, | |
| { | |
| "epoch": 0.8427389014296464, | |
| "grad_norm": 0.03461211919784546, | |
| "learning_rate": 1.195805551808158e-06, | |
| "loss": 0.0017, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.8434913468773514, | |
| "grad_norm": 0.011402531526982784, | |
| "learning_rate": 1.1846207562841416e-06, | |
| "loss": 0.0004, | |
| "step": 1121 | |
| }, | |
| { | |
| "epoch": 0.8442437923250564, | |
| "grad_norm": 4.339677810668945, | |
| "learning_rate": 1.1734852203262115e-06, | |
| "loss": 0.1059, | |
| "step": 1122 | |
| }, | |
| { | |
| "epoch": 0.8449962377727614, | |
| "grad_norm": 2.331862688064575, | |
| "learning_rate": 1.1623990061587665e-06, | |
| "loss": 0.3458, | |
| "step": 1123 | |
| }, | |
| { | |
| "epoch": 0.8457486832204665, | |
| "grad_norm": 0.40041258931159973, | |
| "learning_rate": 1.1513621757306015e-06, | |
| "loss": 0.133, | |
| "step": 1124 | |
| }, | |
| { | |
| "epoch": 0.8465011286681715, | |
| "grad_norm": 0.1884562075138092, | |
| "learning_rate": 1.1403747907145546e-06, | |
| "loss": 0.0999, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.8472535741158767, | |
| "grad_norm": 0.0410892590880394, | |
| "learning_rate": 1.1294369125071692e-06, | |
| "loss": 0.0019, | |
| "step": 1126 | |
| }, | |
| { | |
| "epoch": 0.8480060195635817, | |
| "grad_norm": 0.2528564929962158, | |
| "learning_rate": 1.1185486022283553e-06, | |
| "loss": 0.1256, | |
| "step": 1127 | |
| }, | |
| { | |
| "epoch": 0.8487584650112867, | |
| "grad_norm": 0.0033206380903720856, | |
| "learning_rate": 1.1077099207210296e-06, | |
| "loss": 0.0001, | |
| "step": 1128 | |
| }, | |
| { | |
| "epoch": 0.8495109104589917, | |
| "grad_norm": 0.05121954157948494, | |
| "learning_rate": 1.0969209285507954e-06, | |
| "loss": 0.0018, | |
| "step": 1129 | |
| }, | |
| { | |
| "epoch": 0.8502633559066968, | |
| "grad_norm": 0.0063232192769646645, | |
| "learning_rate": 1.0861816860055952e-06, | |
| "loss": 0.0002, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.8510158013544018, | |
| "grad_norm": 0.03940580040216446, | |
| "learning_rate": 1.0754922530953737e-06, | |
| "loss": 0.001, | |
| "step": 1131 | |
| }, | |
| { | |
| "epoch": 0.8517682468021068, | |
| "grad_norm": 0.2049761265516281, | |
| "learning_rate": 1.0648526895517464e-06, | |
| "loss": 0.0064, | |
| "step": 1132 | |
| }, | |
| { | |
| "epoch": 0.8525206922498119, | |
| "grad_norm": 0.2495325803756714, | |
| "learning_rate": 1.0542630548276588e-06, | |
| "loss": 0.1243, | |
| "step": 1133 | |
| }, | |
| { | |
| "epoch": 0.8532731376975169, | |
| "grad_norm": 0.04376570135354996, | |
| "learning_rate": 1.043723408097065e-06, | |
| "loss": 0.0021, | |
| "step": 1134 | |
| }, | |
| { | |
| "epoch": 0.8540255831452219, | |
| "grad_norm": 0.8723930716514587, | |
| "learning_rate": 1.0332338082545812e-06, | |
| "loss": 0.0126, | |
| "step": 1135 | |
| }, | |
| { | |
| "epoch": 0.8547780285929271, | |
| "grad_norm": 0.016171878203749657, | |
| "learning_rate": 1.0227943139151719e-06, | |
| "loss": 0.0006, | |
| "step": 1136 | |
| }, | |
| { | |
| "epoch": 0.8555304740406321, | |
| "grad_norm": 0.004397090524435043, | |
| "learning_rate": 1.0124049834138205e-06, | |
| "loss": 0.0001, | |
| "step": 1137 | |
| }, | |
| { | |
| "epoch": 0.8562829194883371, | |
| "grad_norm": 0.10261158645153046, | |
| "learning_rate": 1.0020658748051925e-06, | |
| "loss": 0.0046, | |
| "step": 1138 | |
| }, | |
| { | |
| "epoch": 0.8570353649360422, | |
| "grad_norm": 0.07801058888435364, | |
| "learning_rate": 9.91777045863319e-07, | |
| "loss": 0.0037, | |
| "step": 1139 | |
| }, | |
| { | |
| "epoch": 0.8577878103837472, | |
| "grad_norm": 0.37156495451927185, | |
| "learning_rate": 9.815385540812761e-07, | |
| "loss": 0.1171, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.8585402558314522, | |
| "grad_norm": 0.0733116865158081, | |
| "learning_rate": 9.713504566708554e-07, | |
| "loss": 0.0033, | |
| "step": 1141 | |
| }, | |
| { | |
| "epoch": 0.8592927012791572, | |
| "grad_norm": 0.0857694074511528, | |
| "learning_rate": 9.61212810562252e-07, | |
| "loss": 0.0036, | |
| "step": 1142 | |
| }, | |
| { | |
| "epoch": 0.8600451467268623, | |
| "grad_norm": 0.41983649134635925, | |
| "learning_rate": 9.511256724037443e-07, | |
| "loss": 0.0053, | |
| "step": 1143 | |
| }, | |
| { | |
| "epoch": 0.8607975921745673, | |
| "grad_norm": 7.980868816375732, | |
| "learning_rate": 9.410890985613741e-07, | |
| "loss": 0.2031, | |
| "step": 1144 | |
| }, | |
| { | |
| "epoch": 0.8615500376222723, | |
| "grad_norm": 3.7673096656799316, | |
| "learning_rate": 9.311031451186381e-07, | |
| "loss": 0.3022, | |
| "step": 1145 | |
| }, | |
| { | |
| "epoch": 0.8623024830699775, | |
| "grad_norm": 0.028988122940063477, | |
| "learning_rate": 9.21167867876167e-07, | |
| "loss": 0.0011, | |
| "step": 1146 | |
| }, | |
| { | |
| "epoch": 0.8630549285176825, | |
| "grad_norm": 2.064547300338745, | |
| "learning_rate": 9.112833223514183e-07, | |
| "loss": 0.0675, | |
| "step": 1147 | |
| }, | |
| { | |
| "epoch": 0.8638073739653875, | |
| "grad_norm": 1.2910840511322021, | |
| "learning_rate": 9.014495637783671e-07, | |
| "loss": 0.093, | |
| "step": 1148 | |
| }, | |
| { | |
| "epoch": 0.8645598194130926, | |
| "grad_norm": 5.712249279022217, | |
| "learning_rate": 8.916666471071922e-07, | |
| "loss": 0.439, | |
| "step": 1149 | |
| }, | |
| { | |
| "epoch": 0.8653122648607976, | |
| "grad_norm": 0.04252437502145767, | |
| "learning_rate": 8.819346270039752e-07, | |
| "loss": 0.0019, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.8660647103085026, | |
| "grad_norm": 0.02926601469516754, | |
| "learning_rate": 8.722535578503899e-07, | |
| "loss": 0.0014, | |
| "step": 1151 | |
| }, | |
| { | |
| "epoch": 0.8668171557562077, | |
| "grad_norm": 0.3518989086151123, | |
| "learning_rate": 8.62623493743402e-07, | |
| "loss": 0.0112, | |
| "step": 1152 | |
| }, | |
| { | |
| "epoch": 0.8675696012039127, | |
| "grad_norm": 3.836555242538452, | |
| "learning_rate": 8.530444884949674e-07, | |
| "loss": 0.0891, | |
| "step": 1153 | |
| }, | |
| { | |
| "epoch": 0.8683220466516177, | |
| "grad_norm": 0.04374834522604942, | |
| "learning_rate": 8.435165956317226e-07, | |
| "loss": 0.0021, | |
| "step": 1154 | |
| }, | |
| { | |
| "epoch": 0.8690744920993227, | |
| "grad_norm": 0.10189584642648697, | |
| "learning_rate": 8.340398683947004e-07, | |
| "loss": 0.0039, | |
| "step": 1155 | |
| }, | |
| { | |
| "epoch": 0.8698269375470279, | |
| "grad_norm": 0.058472152799367905, | |
| "learning_rate": 8.2461435973902e-07, | |
| "loss": 0.0029, | |
| "step": 1156 | |
| }, | |
| { | |
| "epoch": 0.8705793829947329, | |
| "grad_norm": 0.01816386543214321, | |
| "learning_rate": 8.152401223335993e-07, | |
| "loss": 0.0007, | |
| "step": 1157 | |
| }, | |
| { | |
| "epoch": 0.871331828442438, | |
| "grad_norm": 13.75079345703125, | |
| "learning_rate": 8.059172085608535e-07, | |
| "loss": 0.1557, | |
| "step": 1158 | |
| }, | |
| { | |
| "epoch": 0.872084273890143, | |
| "grad_norm": 0.01980912685394287, | |
| "learning_rate": 7.966456705164094e-07, | |
| "loss": 0.0008, | |
| "step": 1159 | |
| }, | |
| { | |
| "epoch": 0.872836719337848, | |
| "grad_norm": 0.016307447105646133, | |
| "learning_rate": 7.874255600088043e-07, | |
| "loss": 0.0006, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.873589164785553, | |
| "grad_norm": 0.11552978307008743, | |
| "learning_rate": 7.78256928559209e-07, | |
| "loss": 0.0054, | |
| "step": 1161 | |
| }, | |
| { | |
| "epoch": 0.8743416102332581, | |
| "grad_norm": 0.3584482967853546, | |
| "learning_rate": 7.69139827401132e-07, | |
| "loss": 0.0154, | |
| "step": 1162 | |
| }, | |
| { | |
| "epoch": 0.8750940556809631, | |
| "grad_norm": 0.07163400202989578, | |
| "learning_rate": 7.600743074801353e-07, | |
| "loss": 0.0036, | |
| "step": 1163 | |
| }, | |
| { | |
| "epoch": 0.8758465011286681, | |
| "grad_norm": 0.01637900248169899, | |
| "learning_rate": 7.510604194535487e-07, | |
| "loss": 0.0007, | |
| "step": 1164 | |
| }, | |
| { | |
| "epoch": 0.8765989465763732, | |
| "grad_norm": 0.017222406342625618, | |
| "learning_rate": 7.420982136901888e-07, | |
| "loss": 0.0007, | |
| "step": 1165 | |
| }, | |
| { | |
| "epoch": 0.8773513920240783, | |
| "grad_norm": 0.032630909234285355, | |
| "learning_rate": 7.331877402700737e-07, | |
| "loss": 0.0008, | |
| "step": 1166 | |
| }, | |
| { | |
| "epoch": 0.8781038374717833, | |
| "grad_norm": 0.1266675889492035, | |
| "learning_rate": 7.243290489841493e-07, | |
| "loss": 0.0061, | |
| "step": 1167 | |
| }, | |
| { | |
| "epoch": 0.8788562829194884, | |
| "grad_norm": 0.0033598667941987514, | |
| "learning_rate": 7.155221893340036e-07, | |
| "loss": 0.0001, | |
| "step": 1168 | |
| }, | |
| { | |
| "epoch": 0.8796087283671934, | |
| "grad_norm": 0.004325724206864834, | |
| "learning_rate": 7.067672105315981e-07, | |
| "loss": 0.0001, | |
| "step": 1169 | |
| }, | |
| { | |
| "epoch": 0.8803611738148984, | |
| "grad_norm": 0.009192912839353085, | |
| "learning_rate": 6.980641614989847e-07, | |
| "loss": 0.0004, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.8811136192626035, | |
| "grad_norm": 0.037033502012491226, | |
| "learning_rate": 6.894130908680396e-07, | |
| "loss": 0.0016, | |
| "step": 1171 | |
| }, | |
| { | |
| "epoch": 0.8818660647103085, | |
| "grad_norm": 9.960691452026367, | |
| "learning_rate": 6.808140469801872e-07, | |
| "loss": 0.137, | |
| "step": 1172 | |
| }, | |
| { | |
| "epoch": 0.8826185101580135, | |
| "grad_norm": 1.7104170322418213, | |
| "learning_rate": 6.722670778861284e-07, | |
| "loss": 0.0401, | |
| "step": 1173 | |
| }, | |
| { | |
| "epoch": 0.8833709556057185, | |
| "grad_norm": 0.0026178702246397734, | |
| "learning_rate": 6.637722313455774e-07, | |
| "loss": 0.0001, | |
| "step": 1174 | |
| }, | |
| { | |
| "epoch": 0.8841234010534236, | |
| "grad_norm": 0.03204850107431412, | |
| "learning_rate": 6.553295548269922e-07, | |
| "loss": 0.0011, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.8848758465011287, | |
| "grad_norm": 0.03557371720671654, | |
| "learning_rate": 6.469390955073073e-07, | |
| "loss": 0.0012, | |
| "step": 1176 | |
| }, | |
| { | |
| "epoch": 0.8856282919488337, | |
| "grad_norm": 0.3740195035934448, | |
| "learning_rate": 6.386009002716776e-07, | |
| "loss": 0.0097, | |
| "step": 1177 | |
| }, | |
| { | |
| "epoch": 0.8863807373965388, | |
| "grad_norm": 0.06629537791013718, | |
| "learning_rate": 6.303150157132044e-07, | |
| "loss": 0.0027, | |
| "step": 1178 | |
| }, | |
| { | |
| "epoch": 0.8871331828442438, | |
| "grad_norm": 0.0544668547809124, | |
| "learning_rate": 6.22081488132682e-07, | |
| "loss": 0.0024, | |
| "step": 1179 | |
| }, | |
| { | |
| "epoch": 0.8878856282919488, | |
| "grad_norm": 3.0258607864379883, | |
| "learning_rate": 6.139003635383433e-07, | |
| "loss": 0.0818, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.8886380737396539, | |
| "grad_norm": 3.636326313018799, | |
| "learning_rate": 6.057716876455932e-07, | |
| "loss": 0.0272, | |
| "step": 1181 | |
| }, | |
| { | |
| "epoch": 0.8893905191873589, | |
| "grad_norm": 2.166203260421753, | |
| "learning_rate": 5.976955058767609e-07, | |
| "loss": 0.2443, | |
| "step": 1182 | |
| }, | |
| { | |
| "epoch": 0.8901429646350639, | |
| "grad_norm": 0.9411613345146179, | |
| "learning_rate": 5.896718633608412e-07, | |
| "loss": 0.0109, | |
| "step": 1183 | |
| }, | |
| { | |
| "epoch": 0.890895410082769, | |
| "grad_norm": 0.05662931129336357, | |
| "learning_rate": 5.81700804933244e-07, | |
| "loss": 0.0028, | |
| "step": 1184 | |
| }, | |
| { | |
| "epoch": 0.891647855530474, | |
| "grad_norm": 0.005325346253812313, | |
| "learning_rate": 5.737823751355465e-07, | |
| "loss": 0.0002, | |
| "step": 1185 | |
| }, | |
| { | |
| "epoch": 0.8924003009781791, | |
| "grad_norm": 0.009086296893656254, | |
| "learning_rate": 5.659166182152387e-07, | |
| "loss": 0.0003, | |
| "step": 1186 | |
| }, | |
| { | |
| "epoch": 0.8931527464258842, | |
| "grad_norm": 0.05443952605128288, | |
| "learning_rate": 5.581035781254807e-07, | |
| "loss": 0.0023, | |
| "step": 1187 | |
| }, | |
| { | |
| "epoch": 0.8939051918735892, | |
| "grad_norm": 0.09654070436954498, | |
| "learning_rate": 5.503432985248558e-07, | |
| "loss": 0.0041, | |
| "step": 1188 | |
| }, | |
| { | |
| "epoch": 0.8946576373212942, | |
| "grad_norm": 0.022555604577064514, | |
| "learning_rate": 5.426358227771245e-07, | |
| "loss": 0.0008, | |
| "step": 1189 | |
| }, | |
| { | |
| "epoch": 0.8954100827689992, | |
| "grad_norm": 0.022699879482388496, | |
| "learning_rate": 5.349811939509874e-07, | |
| "loss": 0.0009, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.8961625282167043, | |
| "grad_norm": 0.013877753168344498, | |
| "learning_rate": 5.273794548198374e-07, | |
| "loss": 0.0005, | |
| "step": 1191 | |
| }, | |
| { | |
| "epoch": 0.8969149736644093, | |
| "grad_norm": 0.23715998232364655, | |
| "learning_rate": 5.198306478615278e-07, | |
| "loss": 0.0989, | |
| "step": 1192 | |
| }, | |
| { | |
| "epoch": 0.8976674191121143, | |
| "grad_norm": 0.010840164497494698, | |
| "learning_rate": 5.123348152581264e-07, | |
| "loss": 0.0005, | |
| "step": 1193 | |
| }, | |
| { | |
| "epoch": 0.8984198645598194, | |
| "grad_norm": 0.026456279680132866, | |
| "learning_rate": 5.048919988956913e-07, | |
| "loss": 0.0008, | |
| "step": 1194 | |
| }, | |
| { | |
| "epoch": 0.8991723100075244, | |
| "grad_norm": 0.0075182537548244, | |
| "learning_rate": 4.975022403640273e-07, | |
| "loss": 0.0002, | |
| "step": 1195 | |
| }, | |
| { | |
| "epoch": 0.8999247554552295, | |
| "grad_norm": 0.028369707986712456, | |
| "learning_rate": 4.901655809564543e-07, | |
| "loss": 0.0011, | |
| "step": 1196 | |
| }, | |
| { | |
| "epoch": 0.9006772009029346, | |
| "grad_norm": 0.029247762635350227, | |
| "learning_rate": 4.828820616695873e-07, | |
| "loss": 0.0012, | |
| "step": 1197 | |
| }, | |
| { | |
| "epoch": 0.9014296463506396, | |
| "grad_norm": 0.011888640001416206, | |
| "learning_rate": 4.7565172320308886e-07, | |
| "loss": 0.0004, | |
| "step": 1198 | |
| }, | |
| { | |
| "epoch": 0.9021820917983446, | |
| "grad_norm": 0.20360657572746277, | |
| "learning_rate": 4.684746059594558e-07, | |
| "loss": 0.0057, | |
| "step": 1199 | |
| }, | |
| { | |
| "epoch": 0.9029345372460497, | |
| "grad_norm": 0.020525943487882614, | |
| "learning_rate": 4.6135075004379193e-07, | |
| "loss": 0.0005, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.9036869826937547, | |
| "grad_norm": 5.512493133544922, | |
| "learning_rate": 4.542801952635789e-07, | |
| "loss": 0.0409, | |
| "step": 1201 | |
| }, | |
| { | |
| "epoch": 0.9044394281414597, | |
| "grad_norm": 0.041713930666446686, | |
| "learning_rate": 4.472629811284568e-07, | |
| "loss": 0.0014, | |
| "step": 1202 | |
| }, | |
| { | |
| "epoch": 0.9051918735891648, | |
| "grad_norm": 0.023644356057047844, | |
| "learning_rate": 4.4029914685000176e-07, | |
| "loss": 0.0011, | |
| "step": 1203 | |
| }, | |
| { | |
| "epoch": 0.9059443190368698, | |
| "grad_norm": 0.009974486194550991, | |
| "learning_rate": 4.333887313415097e-07, | |
| "loss": 0.0004, | |
| "step": 1204 | |
| }, | |
| { | |
| "epoch": 0.9066967644845748, | |
| "grad_norm": 10.814530372619629, | |
| "learning_rate": 4.265317732177787e-07, | |
| "loss": 0.0471, | |
| "step": 1205 | |
| }, | |
| { | |
| "epoch": 0.90744920993228, | |
| "grad_norm": 0.04487287253141403, | |
| "learning_rate": 4.1972831079488354e-07, | |
| "loss": 0.0012, | |
| "step": 1206 | |
| }, | |
| { | |
| "epoch": 0.908201655379985, | |
| "grad_norm": 0.304022878408432, | |
| "learning_rate": 4.129783820899802e-07, | |
| "loss": 0.1235, | |
| "step": 1207 | |
| }, | |
| { | |
| "epoch": 0.90895410082769, | |
| "grad_norm": 0.08795291930437088, | |
| "learning_rate": 4.0628202482107747e-07, | |
| "loss": 0.004, | |
| "step": 1208 | |
| }, | |
| { | |
| "epoch": 0.909706546275395, | |
| "grad_norm": 0.07199563086032867, | |
| "learning_rate": 3.9963927640683243e-07, | |
| "loss": 0.0035, | |
| "step": 1209 | |
| }, | |
| { | |
| "epoch": 0.9104589917231001, | |
| "grad_norm": 0.021918121725320816, | |
| "learning_rate": 3.930501739663406e-07, | |
| "loss": 0.0009, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.9112114371708051, | |
| "grad_norm": 0.10286161303520203, | |
| "learning_rate": 3.865147543189296e-07, | |
| "loss": 0.0036, | |
| "step": 1211 | |
| }, | |
| { | |
| "epoch": 0.9119638826185101, | |
| "grad_norm": 0.25747624039649963, | |
| "learning_rate": 3.8003305398394916e-07, | |
| "loss": 0.1216, | |
| "step": 1212 | |
| }, | |
| { | |
| "epoch": 0.9127163280662152, | |
| "grad_norm": 1.8191041946411133, | |
| "learning_rate": 3.7360510918057256e-07, | |
| "loss": 0.021, | |
| "step": 1213 | |
| }, | |
| { | |
| "epoch": 0.9134687735139202, | |
| "grad_norm": 0.15397605299949646, | |
| "learning_rate": 3.672309558275922e-07, | |
| "loss": 0.0057, | |
| "step": 1214 | |
| }, | |
| { | |
| "epoch": 0.9142212189616253, | |
| "grad_norm": 0.03293481469154358, | |
| "learning_rate": 3.6091062954321634e-07, | |
| "loss": 0.0016, | |
| "step": 1215 | |
| }, | |
| { | |
| "epoch": 0.9149736644093304, | |
| "grad_norm": 0.002885388908907771, | |
| "learning_rate": 3.5464416564487734e-07, | |
| "loss": 0.0001, | |
| "step": 1216 | |
| }, | |
| { | |
| "epoch": 0.9157261098570354, | |
| "grad_norm": 0.02011510170996189, | |
| "learning_rate": 3.484315991490261e-07, | |
| "loss": 0.0006, | |
| "step": 1217 | |
| }, | |
| { | |
| "epoch": 0.9164785553047404, | |
| "grad_norm": 0.0348239503800869, | |
| "learning_rate": 3.422729647709355e-07, | |
| "loss": 0.0015, | |
| "step": 1218 | |
| }, | |
| { | |
| "epoch": 0.9172310007524455, | |
| "grad_norm": 0.001900368370115757, | |
| "learning_rate": 3.361682969245161e-07, | |
| "loss": 0.0001, | |
| "step": 1219 | |
| }, | |
| { | |
| "epoch": 0.9179834462001505, | |
| "grad_norm": 0.013635323382914066, | |
| "learning_rate": 3.3011762972211647e-07, | |
| "loss": 0.0006, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.9187358916478555, | |
| "grad_norm": 0.5103453397750854, | |
| "learning_rate": 3.241209969743353e-07, | |
| "loss": 0.0088, | |
| "step": 1221 | |
| }, | |
| { | |
| "epoch": 0.9194883370955605, | |
| "grad_norm": 0.03513436019420624, | |
| "learning_rate": 3.181784321898285e-07, | |
| "loss": 0.001, | |
| "step": 1222 | |
| }, | |
| { | |
| "epoch": 0.9202407825432656, | |
| "grad_norm": 0.019935129210352898, | |
| "learning_rate": 3.1228996857512795e-07, | |
| "loss": 0.0007, | |
| "step": 1223 | |
| }, | |
| { | |
| "epoch": 0.9209932279909706, | |
| "grad_norm": 0.0077493940480053425, | |
| "learning_rate": 3.064556390344542e-07, | |
| "loss": 0.0003, | |
| "step": 1224 | |
| }, | |
| { | |
| "epoch": 0.9217456734386757, | |
| "grad_norm": 1.4610238075256348, | |
| "learning_rate": 3.0067547616952297e-07, | |
| "loss": 0.0216, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.9224981188863808, | |
| "grad_norm": 0.013395492918789387, | |
| "learning_rate": 2.949495122793833e-07, | |
| "loss": 0.0005, | |
| "step": 1226 | |
| }, | |
| { | |
| "epoch": 0.9232505643340858, | |
| "grad_norm": 0.022489206865429878, | |
| "learning_rate": 2.892777793602175e-07, | |
| "loss": 0.0009, | |
| "step": 1227 | |
| }, | |
| { | |
| "epoch": 0.9240030097817908, | |
| "grad_norm": 0.04702683165669441, | |
| "learning_rate": 2.836603091051704e-07, | |
| "loss": 0.0019, | |
| "step": 1228 | |
| }, | |
| { | |
| "epoch": 0.9247554552294959, | |
| "grad_norm": 0.34132277965545654, | |
| "learning_rate": 2.7809713290417486e-07, | |
| "loss": 0.012, | |
| "step": 1229 | |
| }, | |
| { | |
| "epoch": 0.9255079006772009, | |
| "grad_norm": 1.5338892936706543, | |
| "learning_rate": 2.7258828184377086e-07, | |
| "loss": 0.1041, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.9262603461249059, | |
| "grad_norm": 0.00669435178861022, | |
| "learning_rate": 2.6713378670693455e-07, | |
| "loss": 0.0002, | |
| "step": 1231 | |
| }, | |
| { | |
| "epoch": 0.927012791572611, | |
| "grad_norm": 9.580220222473145, | |
| "learning_rate": 2.617336779729063e-07, | |
| "loss": 0.1428, | |
| "step": 1232 | |
| }, | |
| { | |
| "epoch": 0.927765237020316, | |
| "grad_norm": 3.141953945159912, | |
| "learning_rate": 2.563879858170215e-07, | |
| "loss": 0.0465, | |
| "step": 1233 | |
| }, | |
| { | |
| "epoch": 0.928517682468021, | |
| "grad_norm": 0.04021593928337097, | |
| "learning_rate": 2.5109674011053684e-07, | |
| "loss": 0.0016, | |
| "step": 1234 | |
| }, | |
| { | |
| "epoch": 0.9292701279157262, | |
| "grad_norm": 0.12653590738773346, | |
| "learning_rate": 2.458599704204712e-07, | |
| "loss": 0.0028, | |
| "step": 1235 | |
| }, | |
| { | |
| "epoch": 0.9300225733634312, | |
| "grad_norm": 1.136615514755249, | |
| "learning_rate": 2.406777060094345e-07, | |
| "loss": 0.016, | |
| "step": 1236 | |
| }, | |
| { | |
| "epoch": 0.9307750188111362, | |
| "grad_norm": 0.0015482566086575389, | |
| "learning_rate": 2.3554997583546402e-07, | |
| "loss": 0.0001, | |
| "step": 1237 | |
| }, | |
| { | |
| "epoch": 0.9315274642588413, | |
| "grad_norm": 0.025700274854898453, | |
| "learning_rate": 2.3047680855186716e-07, | |
| "loss": 0.0012, | |
| "step": 1238 | |
| }, | |
| { | |
| "epoch": 0.9322799097065463, | |
| "grad_norm": 0.03950847312808037, | |
| "learning_rate": 2.2545823250705867e-07, | |
| "loss": 0.0008, | |
| "step": 1239 | |
| }, | |
| { | |
| "epoch": 0.9330323551542513, | |
| "grad_norm": 0.01254583802074194, | |
| "learning_rate": 2.2049427574439953e-07, | |
| "loss": 0.0005, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.9337848006019563, | |
| "grad_norm": 0.1133909747004509, | |
| "learning_rate": 2.1558496600204703e-07, | |
| "loss": 0.0028, | |
| "step": 1241 | |
| }, | |
| { | |
| "epoch": 0.9345372460496614, | |
| "grad_norm": 0.04309477657079697, | |
| "learning_rate": 2.1073033071279057e-07, | |
| "loss": 0.0021, | |
| "step": 1242 | |
| }, | |
| { | |
| "epoch": 0.9352896914973664, | |
| "grad_norm": 2.5412564277648926, | |
| "learning_rate": 2.059303970039106e-07, | |
| "loss": 0.3464, | |
| "step": 1243 | |
| }, | |
| { | |
| "epoch": 0.9360421369450714, | |
| "grad_norm": 0.013797776773571968, | |
| "learning_rate": 2.011851916970109e-07, | |
| "loss": 0.0006, | |
| "step": 1244 | |
| }, | |
| { | |
| "epoch": 0.9367945823927766, | |
| "grad_norm": 0.0062391371466219425, | |
| "learning_rate": 1.9649474130788438e-07, | |
| "loss": 0.0002, | |
| "step": 1245 | |
| }, | |
| { | |
| "epoch": 0.9375470278404816, | |
| "grad_norm": 0.4302297830581665, | |
| "learning_rate": 1.9185907204635755e-07, | |
| "loss": 0.1256, | |
| "step": 1246 | |
| }, | |
| { | |
| "epoch": 0.9382994732881866, | |
| "grad_norm": 0.002676435513421893, | |
| "learning_rate": 1.8727820981614407e-07, | |
| "loss": 0.0001, | |
| "step": 1247 | |
| }, | |
| { | |
| "epoch": 0.9390519187358917, | |
| "grad_norm": 5.231035232543945, | |
| "learning_rate": 1.827521802146981e-07, | |
| "loss": 0.4477, | |
| "step": 1248 | |
| }, | |
| { | |
| "epoch": 0.9398043641835967, | |
| "grad_norm": 1.7769052982330322, | |
| "learning_rate": 1.7828100853307884e-07, | |
| "loss": 0.2043, | |
| "step": 1249 | |
| }, | |
| { | |
| "epoch": 0.9405568096313017, | |
| "grad_norm": 0.005753179080784321, | |
| "learning_rate": 1.7386471975579854e-07, | |
| "loss": 0.0002, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.9413092550790068, | |
| "grad_norm": 0.04713589698076248, | |
| "learning_rate": 1.6950333856069369e-07, | |
| "loss": 0.0013, | |
| "step": 1251 | |
| }, | |
| { | |
| "epoch": 0.9420617005267118, | |
| "grad_norm": 0.042459528893232346, | |
| "learning_rate": 1.651968893187783e-07, | |
| "loss": 0.002, | |
| "step": 1252 | |
| }, | |
| { | |
| "epoch": 0.9428141459744168, | |
| "grad_norm": 0.1957351565361023, | |
| "learning_rate": 1.609453960941143e-07, | |
| "loss": 0.0071, | |
| "step": 1253 | |
| }, | |
| { | |
| "epoch": 0.9435665914221218, | |
| "grad_norm": 0.028360065072774887, | |
| "learning_rate": 1.567488826436725e-07, | |
| "loss": 0.0013, | |
| "step": 1254 | |
| }, | |
| { | |
| "epoch": 0.944319036869827, | |
| "grad_norm": 1.9565247297286987, | |
| "learning_rate": 1.526073724172028e-07, | |
| "loss": 0.0938, | |
| "step": 1255 | |
| }, | |
| { | |
| "epoch": 0.945071482317532, | |
| "grad_norm": 0.023253565654158592, | |
| "learning_rate": 1.485208885570999e-07, | |
| "loss": 0.0009, | |
| "step": 1256 | |
| }, | |
| { | |
| "epoch": 0.945823927765237, | |
| "grad_norm": 0.005460272543132305, | |
| "learning_rate": 1.4448945389827772e-07, | |
| "loss": 0.0002, | |
| "step": 1257 | |
| }, | |
| { | |
| "epoch": 0.9465763732129421, | |
| "grad_norm": 0.07867200672626495, | |
| "learning_rate": 1.4051309096803967e-07, | |
| "loss": 0.0031, | |
| "step": 1258 | |
| }, | |
| { | |
| "epoch": 0.9473288186606471, | |
| "grad_norm": 0.011308335699141026, | |
| "learning_rate": 1.36591821985953e-07, | |
| "loss": 0.0004, | |
| "step": 1259 | |
| }, | |
| { | |
| "epoch": 0.9480812641083521, | |
| "grad_norm": 0.7972699999809265, | |
| "learning_rate": 1.3272566886372572e-07, | |
| "loss": 0.0923, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.9488337095560572, | |
| "grad_norm": 0.01437709853053093, | |
| "learning_rate": 1.2891465320508113e-07, | |
| "loss": 0.0006, | |
| "step": 1261 | |
| }, | |
| { | |
| "epoch": 0.9495861550037622, | |
| "grad_norm": 0.021911505609750748, | |
| "learning_rate": 1.2515879630564108e-07, | |
| "loss": 0.0009, | |
| "step": 1262 | |
| }, | |
| { | |
| "epoch": 0.9503386004514672, | |
| "grad_norm": 0.018048502504825592, | |
| "learning_rate": 1.2145811915280414e-07, | |
| "loss": 0.0009, | |
| "step": 1263 | |
| }, | |
| { | |
| "epoch": 0.9510910458991723, | |
| "grad_norm": 0.01497666072100401, | |
| "learning_rate": 1.1781264242562984e-07, | |
| "loss": 0.0006, | |
| "step": 1264 | |
| }, | |
| { | |
| "epoch": 0.9518434913468774, | |
| "grad_norm": 0.01695135422050953, | |
| "learning_rate": 1.1422238649472228e-07, | |
| "loss": 0.0005, | |
| "step": 1265 | |
| }, | |
| { | |
| "epoch": 0.9525959367945824, | |
| "grad_norm": 0.05780097469687462, | |
| "learning_rate": 1.1068737142211683e-07, | |
| "loss": 0.0023, | |
| "step": 1266 | |
| }, | |
| { | |
| "epoch": 0.9533483822422875, | |
| "grad_norm": 0.4244280457496643, | |
| "learning_rate": 1.072076169611691e-07, | |
| "loss": 0.0127, | |
| "step": 1267 | |
| }, | |
| { | |
| "epoch": 0.9541008276899925, | |
| "grad_norm": 0.28837308287620544, | |
| "learning_rate": 1.0378314255643951e-07, | |
| "loss": 0.0074, | |
| "step": 1268 | |
| }, | |
| { | |
| "epoch": 0.9548532731376975, | |
| "grad_norm": 0.05638807266950607, | |
| "learning_rate": 1.004139673435922e-07, | |
| "loss": 0.0019, | |
| "step": 1269 | |
| }, | |
| { | |
| "epoch": 0.9556057185854026, | |
| "grad_norm": 0.025965796783566475, | |
| "learning_rate": 9.7100110149283e-08, | |
| "loss": 0.0011, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 0.9563581640331076, | |
| "grad_norm": 0.00950747448951006, | |
| "learning_rate": 9.384158949105382e-08, | |
| "loss": 0.0004, | |
| "step": 1271 | |
| }, | |
| { | |
| "epoch": 0.9571106094808126, | |
| "grad_norm": 0.4918772280216217, | |
| "learning_rate": 9.063842357723284e-08, | |
| "loss": 0.0149, | |
| "step": 1272 | |
| }, | |
| { | |
| "epoch": 0.9578630549285176, | |
| "grad_norm": 4.298508167266846, | |
| "learning_rate": 8.749063030683125e-08, | |
| "loss": 0.0998, | |
| "step": 1273 | |
| }, | |
| { | |
| "epoch": 0.9586155003762227, | |
| "grad_norm": 4.235317230224609, | |
| "learning_rate": 8.439822726943991e-08, | |
| "loss": 0.2552, | |
| "step": 1274 | |
| }, | |
| { | |
| "epoch": 0.9593679458239278, | |
| "grad_norm": 0.04486255720257759, | |
| "learning_rate": 8.136123174513843e-08, | |
| "loss": 0.0019, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.9601203912716328, | |
| "grad_norm": 0.004140100441873074, | |
| "learning_rate": 7.837966070438851e-08, | |
| "loss": 0.0001, | |
| "step": 1276 | |
| }, | |
| { | |
| "epoch": 0.9608728367193379, | |
| "grad_norm": 0.01324907224625349, | |
| "learning_rate": 7.54535308079507e-08, | |
| "loss": 0.0006, | |
| "step": 1277 | |
| }, | |
| { | |
| "epoch": 0.9616252821670429, | |
| "grad_norm": 0.04146108031272888, | |
| "learning_rate": 7.258285840677893e-08, | |
| "loss": 0.0018, | |
| "step": 1278 | |
| }, | |
| { | |
| "epoch": 0.9623777276147479, | |
| "grad_norm": 0.008022352121770382, | |
| "learning_rate": 6.976765954194165e-08, | |
| "loss": 0.0003, | |
| "step": 1279 | |
| }, | |
| { | |
| "epoch": 0.963130173062453, | |
| "grad_norm": 0.03538546711206436, | |
| "learning_rate": 6.700794994452198e-08, | |
| "loss": 0.0018, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.963882618510158, | |
| "grad_norm": 2.600939989089966, | |
| "learning_rate": 6.430374503553439e-08, | |
| "loss": 0.5258, | |
| "step": 1281 | |
| }, | |
| { | |
| "epoch": 0.964635063957863, | |
| "grad_norm": 0.13569378852844238, | |
| "learning_rate": 6.165505992584142e-08, | |
| "loss": 0.0053, | |
| "step": 1282 | |
| }, | |
| { | |
| "epoch": 0.9653875094055681, | |
| "grad_norm": 0.004670869559049606, | |
| "learning_rate": 5.9061909416059385e-08, | |
| "loss": 0.0002, | |
| "step": 1283 | |
| }, | |
| { | |
| "epoch": 0.9661399548532731, | |
| "grad_norm": 0.014692548662424088, | |
| "learning_rate": 5.652430799648945e-08, | |
| "loss": 0.0005, | |
| "step": 1284 | |
| }, | |
| { | |
| "epoch": 0.9668924003009782, | |
| "grad_norm": 0.8171061873435974, | |
| "learning_rate": 5.404226984702221e-08, | |
| "loss": 0.1252, | |
| "step": 1285 | |
| }, | |
| { | |
| "epoch": 0.9676448457486833, | |
| "grad_norm": 0.018849292770028114, | |
| "learning_rate": 5.161580883707218e-08, | |
| "loss": 0.0007, | |
| "step": 1286 | |
| }, | |
| { | |
| "epoch": 0.9683972911963883, | |
| "grad_norm": 3.7725510597229004, | |
| "learning_rate": 4.924493852549006e-08, | |
| "loss": 0.2135, | |
| "step": 1287 | |
| }, | |
| { | |
| "epoch": 0.9691497366440933, | |
| "grad_norm": 0.005706661846488714, | |
| "learning_rate": 4.69296721604906e-08, | |
| "loss": 0.0002, | |
| "step": 1288 | |
| }, | |
| { | |
| "epoch": 0.9699021820917983, | |
| "grad_norm": 1.751198649406433, | |
| "learning_rate": 4.4670022679579314e-08, | |
| "loss": 0.0254, | |
| "step": 1289 | |
| }, | |
| { | |
| "epoch": 0.9706546275395034, | |
| "grad_norm": 0.015075215138494968, | |
| "learning_rate": 4.24660027094792e-08, | |
| "loss": 0.0006, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 0.9714070729872084, | |
| "grad_norm": 0.12630032002925873, | |
| "learning_rate": 4.0317624566060806e-08, | |
| "loss": 0.0049, | |
| "step": 1291 | |
| }, | |
| { | |
| "epoch": 0.9721595184349134, | |
| "grad_norm": 0.01940903812646866, | |
| "learning_rate": 3.822490025427339e-08, | |
| "loss": 0.0005, | |
| "step": 1292 | |
| }, | |
| { | |
| "epoch": 0.9729119638826185, | |
| "grad_norm": 0.025583157315850258, | |
| "learning_rate": 3.618784146807497e-08, | |
| "loss": 0.0011, | |
| "step": 1293 | |
| }, | |
| { | |
| "epoch": 0.9736644093303235, | |
| "grad_norm": 3.0885636806488037, | |
| "learning_rate": 3.42064595903735e-08, | |
| "loss": 0.3399, | |
| "step": 1294 | |
| }, | |
| { | |
| "epoch": 0.9744168547780286, | |
| "grad_norm": 0.04877516254782677, | |
| "learning_rate": 3.2280765692956904e-08, | |
| "loss": 0.0022, | |
| "step": 1295 | |
| }, | |
| { | |
| "epoch": 0.9751693002257337, | |
| "grad_norm": 2.90549635887146, | |
| "learning_rate": 3.0410770536432047e-08, | |
| "loss": 0.2458, | |
| "step": 1296 | |
| }, | |
| { | |
| "epoch": 0.9759217456734387, | |
| "grad_norm": 0.039560992270708084, | |
| "learning_rate": 2.859648457016917e-08, | |
| "loss": 0.0017, | |
| "step": 1297 | |
| }, | |
| { | |
| "epoch": 0.9766741911211437, | |
| "grad_norm": 0.10429071635007858, | |
| "learning_rate": 2.6837917932238667e-08, | |
| "loss": 0.0048, | |
| "step": 1298 | |
| }, | |
| { | |
| "epoch": 0.9774266365688488, | |
| "grad_norm": 0.24316257238388062, | |
| "learning_rate": 2.513508044935775e-08, | |
| "loss": 0.0068, | |
| "step": 1299 | |
| }, | |
| { | |
| "epoch": 0.9781790820165538, | |
| "grad_norm": 0.020967544987797737, | |
| "learning_rate": 2.3487981636831635e-08, | |
| "loss": 0.0006, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.9789315274642588, | |
| "grad_norm": 0.05963461101055145, | |
| "learning_rate": 2.189663069850578e-08, | |
| "loss": 0.0026, | |
| "step": 1301 | |
| }, | |
| { | |
| "epoch": 0.9796839729119639, | |
| "grad_norm": 0.022699451074004173, | |
| "learning_rate": 2.0361036526707067e-08, | |
| "loss": 0.001, | |
| "step": 1302 | |
| }, | |
| { | |
| "epoch": 0.9804364183596689, | |
| "grad_norm": 2.274538278579712, | |
| "learning_rate": 1.8881207702202696e-08, | |
| "loss": 0.1143, | |
| "step": 1303 | |
| }, | |
| { | |
| "epoch": 0.9811888638073739, | |
| "grad_norm": 0.12566229701042175, | |
| "learning_rate": 1.7457152494145814e-08, | |
| "loss": 0.0061, | |
| "step": 1304 | |
| }, | |
| { | |
| "epoch": 0.981941309255079, | |
| "grad_norm": 0.03288928419351578, | |
| "learning_rate": 1.6088878860032187e-08, | |
| "loss": 0.0016, | |
| "step": 1305 | |
| }, | |
| { | |
| "epoch": 0.9826937547027841, | |
| "grad_norm": 0.0259998869150877, | |
| "learning_rate": 1.4776394445655818e-08, | |
| "loss": 0.0009, | |
| "step": 1306 | |
| }, | |
| { | |
| "epoch": 0.9834462001504891, | |
| "grad_norm": 1.0154011249542236, | |
| "learning_rate": 1.3519706585063408e-08, | |
| "loss": 0.0419, | |
| "step": 1307 | |
| }, | |
| { | |
| "epoch": 0.9841986455981941, | |
| "grad_norm": 4.762252330780029, | |
| "learning_rate": 1.231882230051662e-08, | |
| "loss": 0.0761, | |
| "step": 1308 | |
| }, | |
| { | |
| "epoch": 0.9849510910458992, | |
| "grad_norm": 5.905474662780762, | |
| "learning_rate": 1.1173748302450993e-08, | |
| "loss": 0.2441, | |
| "step": 1309 | |
| }, | |
| { | |
| "epoch": 0.9857035364936042, | |
| "grad_norm": 4.897678852081299, | |
| "learning_rate": 1.0084490989441531e-08, | |
| "loss": 0.228, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 0.9864559819413092, | |
| "grad_norm": 0.00369740417227149, | |
| "learning_rate": 9.051056448160511e-09, | |
| "loss": 0.0001, | |
| "step": 1311 | |
| }, | |
| { | |
| "epoch": 0.9872084273890143, | |
| "grad_norm": 0.0034076508600264788, | |
| "learning_rate": 8.07345045334973e-09, | |
| "loss": 0.0001, | |
| "step": 1312 | |
| }, | |
| { | |
| "epoch": 0.9879608728367193, | |
| "grad_norm": 0.007578455843031406, | |
| "learning_rate": 7.151678467787193e-09, | |
| "loss": 0.0003, | |
| "step": 1313 | |
| }, | |
| { | |
| "epoch": 0.9887133182844243, | |
| "grad_norm": 0.1452741026878357, | |
| "learning_rate": 6.285745642253816e-09, | |
| "loss": 0.0057, | |
| "step": 1314 | |
| }, | |
| { | |
| "epoch": 0.9894657637321295, | |
| "grad_norm": 0.007722517475485802, | |
| "learning_rate": 5.475656815504549e-09, | |
| "loss": 0.0003, | |
| "step": 1315 | |
| }, | |
| { | |
| "epoch": 0.9902182091798345, | |
| "grad_norm": 0.04480676352977753, | |
| "learning_rate": 4.721416514245069e-09, | |
| "loss": 0.002, | |
| "step": 1316 | |
| }, | |
| { | |
| "epoch": 0.9909706546275395, | |
| "grad_norm": 0.6687166094779968, | |
| "learning_rate": 4.023028953106245e-09, | |
| "loss": 0.1434, | |
| "step": 1317 | |
| }, | |
| { | |
| "epoch": 0.9917231000752446, | |
| "grad_norm": 0.014944463968276978, | |
| "learning_rate": 3.3804980346141547e-09, | |
| "loss": 0.0006, | |
| "step": 1318 | |
| }, | |
| { | |
| "epoch": 0.9924755455229496, | |
| "grad_norm": 0.01458480954170227, | |
| "learning_rate": 2.7938273491756596e-09, | |
| "loss": 0.0006, | |
| "step": 1319 | |
| }, | |
| { | |
| "epoch": 0.9932279909706546, | |
| "grad_norm": 0.01680012419819832, | |
| "learning_rate": 2.2630201750561965e-09, | |
| "loss": 0.0008, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.9939804364183596, | |
| "grad_norm": 0.054570991545915604, | |
| "learning_rate": 1.7880794783575738e-09, | |
| "loss": 0.0019, | |
| "step": 1321 | |
| }, | |
| { | |
| "epoch": 0.9947328818660647, | |
| "grad_norm": 0.07745349407196045, | |
| "learning_rate": 1.3690079130090905e-09, | |
| "loss": 0.0017, | |
| "step": 1322 | |
| }, | |
| { | |
| "epoch": 0.9954853273137697, | |
| "grad_norm": 9.53114128112793, | |
| "learning_rate": 1.0058078207453303e-09, | |
| "loss": 0.0771, | |
| "step": 1323 | |
| }, | |
| { | |
| "epoch": 0.9962377727614747, | |
| "grad_norm": 0.049212612211704254, | |
| "learning_rate": 6.984812310950606e-10, | |
| "loss": 0.0018, | |
| "step": 1324 | |
| }, | |
| { | |
| "epoch": 0.9969902182091799, | |
| "grad_norm": 0.004117061849683523, | |
| "learning_rate": 4.470298613745705e-10, | |
| "loss": 0.0002, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.9977426636568849, | |
| "grad_norm": 0.06400679796934128, | |
| "learning_rate": 2.514551166699075e-10, | |
| "loss": 0.0028, | |
| "step": 1326 | |
| }, | |
| { | |
| "epoch": 0.9984951091045899, | |
| "grad_norm": 1.3703467845916748, | |
| "learning_rate": 1.1175808983687752e-10, | |
| "loss": 0.0185, | |
| "step": 1327 | |
| }, | |
| { | |
| "epoch": 0.999247554552295, | |
| "grad_norm": 0.01683868281543255, | |
| "learning_rate": 2.793956148994248e-11, | |
| "loss": 0.0007, | |
| "step": 1328 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 0.0036646851804107428, | |
| "learning_rate": 0.0, | |
| "loss": 0.0001, | |
| "step": 1329 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 1329, | |
| "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": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 4.439066769492296e+17, | |
| "train_batch_size": 2, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |