| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.9961919268849961, | |
| "eval_steps": 500, | |
| "global_step": 218, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.00913937547600914, | |
| "grad_norm": 3.3369264602661133, | |
| "learning_rate": 9.090909090909091e-07, | |
| "loss": 0.6708, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.01827875095201828, | |
| "grad_norm": 3.4893620014190674, | |
| "learning_rate": 1.8181818181818183e-06, | |
| "loss": 0.6794, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.027418126428027417, | |
| "grad_norm": 3.4242701530456543, | |
| "learning_rate": 2.7272727272727272e-06, | |
| "loss": 0.6599, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.03655750190403656, | |
| "grad_norm": 3.155944347381592, | |
| "learning_rate": 3.6363636363636366e-06, | |
| "loss": 0.6689, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.0456968773800457, | |
| "grad_norm": 2.458157539367676, | |
| "learning_rate": 4.5454545454545455e-06, | |
| "loss": 0.6512, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.05483625285605483, | |
| "grad_norm": 1.8310825824737549, | |
| "learning_rate": 5.4545454545454545e-06, | |
| "loss": 0.6187, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.06397562833206398, | |
| "grad_norm": 1.7105038166046143, | |
| "learning_rate": 6.363636363636364e-06, | |
| "loss": 0.6179, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.07311500380807312, | |
| "grad_norm": 1.7482938766479492, | |
| "learning_rate": 7.272727272727273e-06, | |
| "loss": 0.6075, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.08225437928408226, | |
| "grad_norm": 1.6708910465240479, | |
| "learning_rate": 8.181818181818183e-06, | |
| "loss": 0.5964, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.0913937547600914, | |
| "grad_norm": 1.1437326669692993, | |
| "learning_rate": 9.090909090909091e-06, | |
| "loss": 0.5533, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.10053313023610053, | |
| "grad_norm": 1.5597337484359741, | |
| "learning_rate": 1e-05, | |
| "loss": 0.5382, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.10967250571210967, | |
| "grad_norm": 1.0245758295059204, | |
| "learning_rate": 1.0909090909090909e-05, | |
| "loss": 0.5104, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.1188118811881188, | |
| "grad_norm": 0.8286274671554565, | |
| "learning_rate": 1.181818181818182e-05, | |
| "loss": 0.5173, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.12795125666412796, | |
| "grad_norm": 1.0838040113449097, | |
| "learning_rate": 1.2727272727272728e-05, | |
| "loss": 0.5133, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.1370906321401371, | |
| "grad_norm": 1.0685889720916748, | |
| "learning_rate": 1.3636363636363637e-05, | |
| "loss": 0.5109, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.14623000761614624, | |
| "grad_norm": 0.794349193572998, | |
| "learning_rate": 1.4545454545454546e-05, | |
| "loss": 0.4618, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.15536938309215537, | |
| "grad_norm": 0.8422227501869202, | |
| "learning_rate": 1.5454545454545454e-05, | |
| "loss": 0.4867, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.16450875856816452, | |
| "grad_norm": 0.7492879629135132, | |
| "learning_rate": 1.6363636363636366e-05, | |
| "loss": 0.4731, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.17364813404417365, | |
| "grad_norm": 0.6844476461410522, | |
| "learning_rate": 1.7272727272727274e-05, | |
| "loss": 0.4439, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.1827875095201828, | |
| "grad_norm": 0.7250885367393494, | |
| "learning_rate": 1.8181818181818182e-05, | |
| "loss": 0.4716, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.19192688499619193, | |
| "grad_norm": 0.67299485206604, | |
| "learning_rate": 1.9090909090909094e-05, | |
| "loss": 0.4412, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.20106626047220105, | |
| "grad_norm": 0.6760995388031006, | |
| "learning_rate": 2e-05, | |
| "loss": 0.4459, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.2102056359482102, | |
| "grad_norm": 0.5327174067497253, | |
| "learning_rate": 1.9998715457999313e-05, | |
| "loss": 0.45, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.21934501142421933, | |
| "grad_norm": 0.5728371143341064, | |
| "learning_rate": 1.999486216200688e-05, | |
| "loss": 0.4362, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.2284843869002285, | |
| "grad_norm": 0.4978387951850891, | |
| "learning_rate": 1.9988441101966807e-05, | |
| "loss": 0.4562, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.2376237623762376, | |
| "grad_norm": 0.5106445550918579, | |
| "learning_rate": 1.9979453927503366e-05, | |
| "loss": 0.4684, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.24676313785224677, | |
| "grad_norm": 0.4982711374759674, | |
| "learning_rate": 1.9967902947497158e-05, | |
| "loss": 0.4526, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.2559025133282559, | |
| "grad_norm": 0.5087048411369324, | |
| "learning_rate": 1.9953791129491985e-05, | |
| "loss": 0.4468, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.26504188880426505, | |
| "grad_norm": 0.47533684968948364, | |
| "learning_rate": 1.9937122098932428e-05, | |
| "loss": 0.4272, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.2741812642802742, | |
| "grad_norm": 0.4664396643638611, | |
| "learning_rate": 1.991790013823246e-05, | |
| "loss": 0.4456, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.2833206397562833, | |
| "grad_norm": 0.44873371720314026, | |
| "learning_rate": 1.9896130185675263e-05, | |
| "loss": 0.4436, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.2924600152322925, | |
| "grad_norm": 0.46020084619522095, | |
| "learning_rate": 1.9871817834144506e-05, | |
| "loss": 0.4285, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.3015993907083016, | |
| "grad_norm": 0.6484828591346741, | |
| "learning_rate": 1.9844969329687526e-05, | |
| "loss": 0.4385, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.31073876618431073, | |
| "grad_norm": 0.40165239572525024, | |
| "learning_rate": 1.9815591569910654e-05, | |
| "loss": 0.4159, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.31987814166031986, | |
| "grad_norm": 0.42455559968948364, | |
| "learning_rate": 1.9783692102207156e-05, | |
| "loss": 0.4407, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.32901751713632904, | |
| "grad_norm": 0.4265769124031067, | |
| "learning_rate": 1.9749279121818235e-05, | |
| "loss": 0.3885, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.33815689261233817, | |
| "grad_norm": 0.38099807500839233, | |
| "learning_rate": 1.971236146972764e-05, | |
| "loss": 0.3883, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.3472962680883473, | |
| "grad_norm": 0.43669813871383667, | |
| "learning_rate": 1.9672948630390296e-05, | |
| "loss": 0.3881, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.3564356435643564, | |
| "grad_norm": 0.39924174547195435, | |
| "learning_rate": 1.9631050729295705e-05, | |
| "loss": 0.4378, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.3655750190403656, | |
| "grad_norm": 0.44854018092155457, | |
| "learning_rate": 1.9586678530366607e-05, | |
| "loss": 0.4125, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.3747143945163747, | |
| "grad_norm": 0.3688861131668091, | |
| "learning_rate": 1.953984343319364e-05, | |
| "loss": 0.3944, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.38385376999238385, | |
| "grad_norm": 0.3711301386356354, | |
| "learning_rate": 1.949055747010669e-05, | |
| "loss": 0.4164, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.392993145468393, | |
| "grad_norm": 0.4008907675743103, | |
| "learning_rate": 1.9438833303083677e-05, | |
| "loss": 0.4196, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.4021325209444021, | |
| "grad_norm": 0.42354318499565125, | |
| "learning_rate": 1.9384684220497605e-05, | |
| "loss": 0.3882, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.4112718964204113, | |
| "grad_norm": 0.42434418201446533, | |
| "learning_rate": 1.932812413370265e-05, | |
| "loss": 0.4301, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.4204112718964204, | |
| "grad_norm": 0.4077893793582916, | |
| "learning_rate": 1.926916757346022e-05, | |
| "loss": 0.3943, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.42955064737242954, | |
| "grad_norm": 0.3807850182056427, | |
| "learning_rate": 1.9207829686205882e-05, | |
| "loss": 0.3888, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.43869002284843867, | |
| "grad_norm": 0.3933984637260437, | |
| "learning_rate": 1.9144126230158127e-05, | |
| "loss": 0.4084, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.44782939832444785, | |
| "grad_norm": 0.45636460185050964, | |
| "learning_rate": 1.9078073571269922e-05, | |
| "loss": 0.3892, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.456968773800457, | |
| "grad_norm": 0.3460674583911896, | |
| "learning_rate": 1.900968867902419e-05, | |
| "loss": 0.3911, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.4661081492764661, | |
| "grad_norm": 0.40278884768486023, | |
| "learning_rate": 1.8938989122074195e-05, | |
| "loss": 0.3985, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.4752475247524752, | |
| "grad_norm": 0.4178655743598938, | |
| "learning_rate": 1.8865993063730003e-05, | |
| "loss": 0.4156, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.4843869002284844, | |
| "grad_norm": 0.3947036862373352, | |
| "learning_rate": 1.8790719257292175e-05, | |
| "loss": 0.4189, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.49352627570449353, | |
| "grad_norm": 0.3686043322086334, | |
| "learning_rate": 1.8713187041233896e-05, | |
| "loss": 0.4105, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.5026656511805027, | |
| "grad_norm": 0.4284152090549469, | |
| "learning_rate": 1.8633416334232754e-05, | |
| "loss": 0.389, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.5118050266565118, | |
| "grad_norm": 0.40143266320228577, | |
| "learning_rate": 1.8551427630053464e-05, | |
| "loss": 0.4168, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.5209444021325209, | |
| "grad_norm": 0.37129855155944824, | |
| "learning_rate": 1.8467241992282842e-05, | |
| "loss": 0.3745, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.5300837776085301, | |
| "grad_norm": 0.4992600679397583, | |
| "learning_rate": 1.8380881048918406e-05, | |
| "loss": 0.4024, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.5392231530845393, | |
| "grad_norm": 0.403962641954422, | |
| "learning_rate": 1.8292366986811952e-05, | |
| "loss": 0.379, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.5483625285605483, | |
| "grad_norm": 0.3711932301521301, | |
| "learning_rate": 1.820172254596956e-05, | |
| "loss": 0.3959, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.5575019040365575, | |
| "grad_norm": 0.45666763186454773, | |
| "learning_rate": 1.8108971013709512e-05, | |
| "loss": 0.3833, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.5666412795125666, | |
| "grad_norm": 0.34401726722717285, | |
| "learning_rate": 1.8014136218679566e-05, | |
| "loss": 0.3847, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.5757806549885758, | |
| "grad_norm": 0.36564773321151733, | |
| "learning_rate": 1.79172425247352e-05, | |
| "loss": 0.384, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.584920030464585, | |
| "grad_norm": 0.3763687014579773, | |
| "learning_rate": 1.78183148246803e-05, | |
| "loss": 0.3827, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.594059405940594, | |
| "grad_norm": 0.3825934827327728, | |
| "learning_rate": 1.771737853387202e-05, | |
| "loss": 0.3971, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.6031987814166032, | |
| "grad_norm": 0.3348219394683838, | |
| "learning_rate": 1.7614459583691346e-05, | |
| "loss": 0.3965, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.6123381568926123, | |
| "grad_norm": 0.3712629973888397, | |
| "learning_rate": 1.7509584414881114e-05, | |
| "loss": 0.3933, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.6214775323686215, | |
| "grad_norm": 0.3879682123661041, | |
| "learning_rate": 1.7402779970753156e-05, | |
| "loss": 0.4008, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.6306169078446306, | |
| "grad_norm": 0.3735283315181732, | |
| "learning_rate": 1.7294073690266343e-05, | |
| "loss": 0.4058, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.6397562833206397, | |
| "grad_norm": 0.36928698420524597, | |
| "learning_rate": 1.7183493500977277e-05, | |
| "loss": 0.3911, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.6488956587966489, | |
| "grad_norm": 0.32693177461624146, | |
| "learning_rate": 1.7071067811865477e-05, | |
| "loss": 0.389, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.6580350342726581, | |
| "grad_norm": 0.35049450397491455, | |
| "learning_rate": 1.6956825506034866e-05, | |
| "loss": 0.3824, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.6671744097486672, | |
| "grad_norm": 0.3910599946975708, | |
| "learning_rate": 1.6840795933293464e-05, | |
| "loss": 0.3797, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.6763137852246763, | |
| "grad_norm": 0.34450212121009827, | |
| "learning_rate": 1.672300890261317e-05, | |
| "loss": 0.3769, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.6854531607006854, | |
| "grad_norm": 0.3638371229171753, | |
| "learning_rate": 1.6603494674471595e-05, | |
| "loss": 0.3765, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.6945925361766946, | |
| "grad_norm": 0.3630662262439728, | |
| "learning_rate": 1.6482283953077887e-05, | |
| "loss": 0.402, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.7037319116527038, | |
| "grad_norm": 0.3464076519012451, | |
| "learning_rate": 1.635940787848455e-05, | |
| "loss": 0.3776, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.7128712871287128, | |
| "grad_norm": 0.35827362537384033, | |
| "learning_rate": 1.6234898018587336e-05, | |
| "loss": 0.381, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.722010662604722, | |
| "grad_norm": 0.3732692301273346, | |
| "learning_rate": 1.6108786361015145e-05, | |
| "loss": 0.3723, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.7311500380807312, | |
| "grad_norm": 0.3525165617465973, | |
| "learning_rate": 1.598110530491216e-05, | |
| "loss": 0.4152, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.7402894135567403, | |
| "grad_norm": 0.39010149240493774, | |
| "learning_rate": 1.5851887652614238e-05, | |
| "loss": 0.3937, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.7494287890327495, | |
| "grad_norm": 0.349071741104126, | |
| "learning_rate": 1.5721166601221697e-05, | |
| "loss": 0.4073, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.7585681645087585, | |
| "grad_norm": 0.33521854877471924, | |
| "learning_rate": 1.5588975734070717e-05, | |
| "loss": 0.3756, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.7677075399847677, | |
| "grad_norm": 0.36783722043037415, | |
| "learning_rate": 1.5455349012105488e-05, | |
| "loss": 0.3902, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.7768469154607769, | |
| "grad_norm": 0.3330143690109253, | |
| "learning_rate": 1.5320320765153367e-05, | |
| "loss": 0.3879, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.785986290936786, | |
| "grad_norm": 0.35695841908454895, | |
| "learning_rate": 1.5183925683105254e-05, | |
| "loss": 0.3887, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.7951256664127951, | |
| "grad_norm": 0.35579657554626465, | |
| "learning_rate": 1.504619880700346e-05, | |
| "loss": 0.3859, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.8042650418888042, | |
| "grad_norm": 0.3615005314350128, | |
| "learning_rate": 1.4907175520039381e-05, | |
| "loss": 0.381, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.8134044173648134, | |
| "grad_norm": 0.37881505489349365, | |
| "learning_rate": 1.4766891538463255e-05, | |
| "loss": 0.3798, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.8225437928408226, | |
| "grad_norm": 0.3448372483253479, | |
| "learning_rate": 1.4625382902408356e-05, | |
| "loss": 0.376, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.8316831683168316, | |
| "grad_norm": 0.34421786665916443, | |
| "learning_rate": 1.448268596663197e-05, | |
| "loss": 0.3692, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.8408225437928408, | |
| "grad_norm": 0.32549339532852173, | |
| "learning_rate": 1.4338837391175582e-05, | |
| "loss": 0.385, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.84996191926885, | |
| "grad_norm": 0.34452754259109497, | |
| "learning_rate": 1.419387413194657e-05, | |
| "loss": 0.3711, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.8591012947448591, | |
| "grad_norm": 0.3506733477115631, | |
| "learning_rate": 1.4047833431223938e-05, | |
| "loss": 0.3772, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.8682406702208683, | |
| "grad_norm": 0.3662183880805969, | |
| "learning_rate": 1.390075280809047e-05, | |
| "loss": 0.3791, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.8773800456968773, | |
| "grad_norm": 0.3463532030582428, | |
| "learning_rate": 1.3752670048793744e-05, | |
| "loss": 0.3767, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.8865194211728865, | |
| "grad_norm": 0.3397160768508911, | |
| "learning_rate": 1.3603623197038536e-05, | |
| "loss": 0.3831, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.8956587966488957, | |
| "grad_norm": 0.3651241660118103, | |
| "learning_rate": 1.3453650544213078e-05, | |
| "loss": 0.3933, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.9047981721249048, | |
| "grad_norm": 0.3426673710346222, | |
| "learning_rate": 1.3302790619551673e-05, | |
| "loss": 0.366, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.913937547600914, | |
| "grad_norm": 0.32238805294036865, | |
| "learning_rate": 1.315108218023621e-05, | |
| "loss": 0.3824, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.9230769230769231, | |
| "grad_norm": 0.3311628997325897, | |
| "learning_rate": 1.2998564201439117e-05, | |
| "loss": 0.3845, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.9322162985529322, | |
| "grad_norm": 0.34884315729141235, | |
| "learning_rate": 1.2845275866310325e-05, | |
| "loss": 0.3819, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.9413556740289414, | |
| "grad_norm": 0.32266736030578613, | |
| "learning_rate": 1.2691256555910769e-05, | |
| "loss": 0.3759, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.9504950495049505, | |
| "grad_norm": 0.3028276264667511, | |
| "learning_rate": 1.2536545839095074e-05, | |
| "loss": 0.3963, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.9596344249809596, | |
| "grad_norm": 0.3258204460144043, | |
| "learning_rate": 1.2381183462345983e-05, | |
| "loss": 0.3805, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.9687738004569688, | |
| "grad_norm": 0.30626070499420166, | |
| "learning_rate": 1.2225209339563144e-05, | |
| "loss": 0.3781, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.9779131759329779, | |
| "grad_norm": 0.29698455333709717, | |
| "learning_rate": 1.206866354180891e-05, | |
| "loss": 0.3777, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.9870525514089871, | |
| "grad_norm": 0.30517199635505676, | |
| "learning_rate": 1.1911586287013726e-05, | |
| "loss": 0.3822, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.9961919268849961, | |
| "grad_norm": 0.30879127979278564, | |
| "learning_rate": 1.1754017929643818e-05, | |
| "loss": 0.3761, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 1.0053313023610053, | |
| "grad_norm": 0.547318160533905, | |
| "learning_rate": 1.1595998950333794e-05, | |
| "loss": 0.5449, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 1.0144706778370145, | |
| "grad_norm": 0.37278491258621216, | |
| "learning_rate": 1.143756994548682e-05, | |
| "loss": 0.3479, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 1.0236100533130237, | |
| "grad_norm": 0.35126781463623047, | |
| "learning_rate": 1.1278771616845061e-05, | |
| "loss": 0.338, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 1.0327494287890326, | |
| "grad_norm": 0.37905386090278625, | |
| "learning_rate": 1.1119644761033079e-05, | |
| "loss": 0.3353, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 1.0418888042650418, | |
| "grad_norm": 0.41645359992980957, | |
| "learning_rate": 1.0960230259076819e-05, | |
| "loss": 0.3257, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 1.051028179741051, | |
| "grad_norm": 0.418813019990921, | |
| "learning_rate": 1.0800569065900935e-05, | |
| "loss": 0.3521, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 1.0601675552170602, | |
| "grad_norm": 0.41467365622520447, | |
| "learning_rate": 1.064070219980713e-05, | |
| "loss": 0.3117, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 1.0693069306930694, | |
| "grad_norm": 0.4041651487350464, | |
| "learning_rate": 1.0480670731936209e-05, | |
| "loss": 0.3093, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 1.0784463061690786, | |
| "grad_norm": 0.3533031642436981, | |
| "learning_rate": 1.0320515775716556e-05, | |
| "loss": 0.2998, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 1.0875856816450875, | |
| "grad_norm": 0.41691580414772034, | |
| "learning_rate": 1.0160278476301739e-05, | |
| "loss": 0.3737, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 1.0967250571210967, | |
| "grad_norm": 0.36301714181900024, | |
| "learning_rate": 1e-05, | |
| "loss": 0.3267, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 1.1058644325971059, | |
| "grad_norm": 0.3264053761959076, | |
| "learning_rate": 9.839721523698265e-06, | |
| "loss": 0.2584, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 1.115003808073115, | |
| "grad_norm": 0.4043513238430023, | |
| "learning_rate": 9.67948422428345e-06, | |
| "loss": 0.3526, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 1.1241431835491242, | |
| "grad_norm": 0.3653642237186432, | |
| "learning_rate": 9.519329268063795e-06, | |
| "loss": 0.3425, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 1.1332825590251332, | |
| "grad_norm": 0.33410775661468506, | |
| "learning_rate": 9.359297800192873e-06, | |
| "loss": 0.3097, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 1.1424219345011424, | |
| "grad_norm": 0.340742826461792, | |
| "learning_rate": 9.199430934099068e-06, | |
| "loss": 0.3052, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 1.1515613099771516, | |
| "grad_norm": 0.3352021872997284, | |
| "learning_rate": 9.039769740923183e-06, | |
| "loss": 0.2977, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 1.1607006854531607, | |
| "grad_norm": 0.3389112949371338, | |
| "learning_rate": 8.880355238966923e-06, | |
| "loss": 0.3113, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 1.16984006092917, | |
| "grad_norm": 0.3363248407840729, | |
| "learning_rate": 8.721228383154939e-06, | |
| "loss": 0.3205, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 1.178979436405179, | |
| "grad_norm": 0.34208640456199646, | |
| "learning_rate": 8.562430054513184e-06, | |
| "loss": 0.3176, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 1.188118811881188, | |
| "grad_norm": 0.3295469284057617, | |
| "learning_rate": 8.404001049666211e-06, | |
| "loss": 0.3201, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 1.1972581873571972, | |
| "grad_norm": 0.3238345682621002, | |
| "learning_rate": 8.245982070356186e-06, | |
| "loss": 0.297, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 1.2063975628332064, | |
| "grad_norm": 0.3155887722969055, | |
| "learning_rate": 8.08841371298628e-06, | |
| "loss": 0.3163, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 1.2155369383092156, | |
| "grad_norm": 0.30575722455978394, | |
| "learning_rate": 7.931336458191092e-06, | |
| "loss": 0.3421, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 1.2246763137852246, | |
| "grad_norm": 0.31150344014167786, | |
| "learning_rate": 7.774790660436857e-06, | |
| "loss": 0.3097, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 1.2338156892612338, | |
| "grad_norm": 0.32414835691452026, | |
| "learning_rate": 7.618816537654018e-06, | |
| "loss": 0.3201, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 1.242955064737243, | |
| "grad_norm": 0.2960313558578491, | |
| "learning_rate": 7.463454160904928e-06, | |
| "loss": 0.2807, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 1.2520944402132521, | |
| "grad_norm": 0.308651864528656, | |
| "learning_rate": 7.308743444089232e-06, | |
| "loss": 0.3258, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 1.2612338156892613, | |
| "grad_norm": 0.3236067593097687, | |
| "learning_rate": 7.154724133689677e-06, | |
| "loss": 0.2958, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 1.2703731911652705, | |
| "grad_norm": 0.3053479492664337, | |
| "learning_rate": 7.001435798560884e-06, | |
| "loss": 0.321, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 1.2795125666412794, | |
| "grad_norm": 0.3023087978363037, | |
| "learning_rate": 6.848917819763794e-06, | |
| "loss": 0.3006, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 1.2886519421172886, | |
| "grad_norm": 0.3476320505142212, | |
| "learning_rate": 6.697209380448333e-06, | |
| "loss": 0.3311, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 1.2977913175932978, | |
| "grad_norm": 0.3587212562561035, | |
| "learning_rate": 6.546349455786926e-06, | |
| "loss": 0.3646, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 1.306930693069307, | |
| "grad_norm": 0.28520774841308594, | |
| "learning_rate": 6.396376802961468e-06, | |
| "loss": 0.3165, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 1.316070068545316, | |
| "grad_norm": 0.28903257846832275, | |
| "learning_rate": 6.24732995120626e-06, | |
| "loss": 0.3019, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 1.3252094440213251, | |
| "grad_norm": 0.3147772252559662, | |
| "learning_rate": 6.099247191909532e-06, | |
| "loss": 0.2998, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 1.3343488194973343, | |
| "grad_norm": 0.32797104120254517, | |
| "learning_rate": 5.952166568776062e-06, | |
| "loss": 0.3368, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 1.3434881949733435, | |
| "grad_norm": 0.2969732880592346, | |
| "learning_rate": 5.806125868053433e-06, | |
| "loss": 0.2851, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 1.3526275704493527, | |
| "grad_norm": 0.3113080561161041, | |
| "learning_rate": 5.66116260882442e-06, | |
| "loss": 0.2959, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 1.3617669459253618, | |
| "grad_norm": 0.32516226172447205, | |
| "learning_rate": 5.517314033368031e-06, | |
| "loss": 0.3239, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 1.370906321401371, | |
| "grad_norm": 0.32933881878852844, | |
| "learning_rate": 5.37461709759165e-06, | |
| "loss": 0.3288, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.3838537699923839, | |
| "grad_norm": 0.29292726516723633, | |
| "learning_rate": 5.233108461536749e-06, | |
| "loss": 0.3226, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 1.392993145468393, | |
| "grad_norm": 0.3006382882595062, | |
| "learning_rate": 5.092824479960625e-06, | |
| "loss": 0.3039, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 1.4021325209444022, | |
| "grad_norm": 0.29254230856895447, | |
| "learning_rate": 4.9538011929965436e-06, | |
| "loss": 0.3441, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 1.4112718964204114, | |
| "grad_norm": 0.30929845571517944, | |
| "learning_rate": 4.81607431689475e-06, | |
| "loss": 0.3127, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 1.4204112718964204, | |
| "grad_norm": 0.28565043210983276, | |
| "learning_rate": 4.679679234846636e-06, | |
| "loss": 0.3119, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 1.4295506473724295, | |
| "grad_norm": 0.3046690821647644, | |
| "learning_rate": 4.544650987894514e-06, | |
| "loss": 0.2909, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 1.4386900228484387, | |
| "grad_norm": 0.279197096824646, | |
| "learning_rate": 4.411024265929283e-06, | |
| "loss": 0.3118, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 1.447829398324448, | |
| "grad_norm": 0.29681769013404846, | |
| "learning_rate": 4.278833398778306e-06, | |
| "loss": 0.3122, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 1.4569687738004569, | |
| "grad_norm": 0.28999167680740356, | |
| "learning_rate": 4.148112347385762e-06, | |
| "loss": 0.3073, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 1.466108149276466, | |
| "grad_norm": 0.3117202818393707, | |
| "learning_rate": 4.01889469508784e-06, | |
| "loss": 0.3206, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 1.4752475247524752, | |
| "grad_norm": 0.28692707419395447, | |
| "learning_rate": 3.891213638984858e-06, | |
| "loss": 0.3147, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 1.4843869002284844, | |
| "grad_norm": 0.2778036296367645, | |
| "learning_rate": 3.7651019814126656e-06, | |
| "loss": 0.3249, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 1.4935262757044936, | |
| "grad_norm": 0.30285775661468506, | |
| "learning_rate": 3.6405921215154492e-06, | |
| "loss": 0.3121, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 1.5026656511805028, | |
| "grad_norm": 0.3020969033241272, | |
| "learning_rate": 3.5177160469221184e-06, | |
| "loss": 0.3003, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 1.511805026656512, | |
| "grad_norm": 0.2875075340270996, | |
| "learning_rate": 3.3965053255284085e-06, | |
| "loss": 0.3115, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 1.520944402132521, | |
| "grad_norm": 0.3074340522289276, | |
| "learning_rate": 3.2769910973868314e-06, | |
| "loss": 0.3131, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 1.53008377760853, | |
| "grad_norm": 0.2820333242416382, | |
| "learning_rate": 3.1592040667065393e-06, | |
| "loss": 0.2978, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 1.5392231530845393, | |
| "grad_norm": 0.3219679296016693, | |
| "learning_rate": 3.0431744939651365e-06, | |
| "loss": 0.3075, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 1.5483625285605482, | |
| "grad_norm": 0.2894546389579773, | |
| "learning_rate": 2.9289321881345257e-06, | |
| "loss": 0.3107, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 1.5575019040365574, | |
| "grad_norm": 0.33814314007759094, | |
| "learning_rate": 2.8165064990227255e-06, | |
| "loss": 0.3338, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 1.5666412795125666, | |
| "grad_norm": 0.330655962228775, | |
| "learning_rate": 2.7059263097336595e-06, | |
| "loss": 0.3184, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 1.5757806549885758, | |
| "grad_norm": 0.29267704486846924, | |
| "learning_rate": 2.597220029246846e-06, | |
| "loss": 0.3241, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 1.584920030464585, | |
| "grad_norm": 0.30491986870765686, | |
| "learning_rate": 2.490415585118887e-06, | |
| "loss": 0.3033, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 1.5940594059405941, | |
| "grad_norm": 0.30731192231178284, | |
| "learning_rate": 2.3855404163086558e-06, | |
| "loss": 0.2965, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 1.6031987814166033, | |
| "grad_norm": 0.2904992699623108, | |
| "learning_rate": 2.282621466127982e-06, | |
| "loss": 0.3046, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 1.6123381568926123, | |
| "grad_norm": 0.31819531321525574, | |
| "learning_rate": 2.1816851753197023e-06, | |
| "loss": 0.2998, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 1.6214775323686215, | |
| "grad_norm": 0.30998897552490234, | |
| "learning_rate": 2.082757475264804e-06, | |
| "loss": 0.3047, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 1.6306169078446306, | |
| "grad_norm": 0.2906753718852997, | |
| "learning_rate": 1.9858637813204352e-06, | |
| "loss": 0.2906, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 1.6397562833206396, | |
| "grad_norm": 0.29853326082229614, | |
| "learning_rate": 1.8910289862904917e-06, | |
| "loss": 0.3223, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 1.6488956587966488, | |
| "grad_norm": 0.31702253222465515, | |
| "learning_rate": 1.7982774540304404e-06, | |
| "loss": 0.3286, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 1.658035034272658, | |
| "grad_norm": 0.2962207794189453, | |
| "learning_rate": 1.7076330131880525e-06, | |
| "loss": 0.3162, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 1.6671744097486672, | |
| "grad_norm": 0.2774161100387573, | |
| "learning_rate": 1.6191189510815942e-06, | |
| "loss": 0.3147, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 1.6763137852246763, | |
| "grad_norm": 0.2778419852256775, | |
| "learning_rate": 1.5327580077171589e-06, | |
| "loss": 0.3086, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 1.6854531607006855, | |
| "grad_norm": 0.31520137190818787, | |
| "learning_rate": 1.4485723699465392e-06, | |
| "loss": 0.3259, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 1.6945925361766947, | |
| "grad_norm": 0.29337337613105774, | |
| "learning_rate": 1.3665836657672493e-06, | |
| "loss": 0.3126, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 1.7037319116527039, | |
| "grad_norm": 0.3039644956588745, | |
| "learning_rate": 1.286812958766106e-06, | |
| "loss": 0.3194, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 1.7128712871287128, | |
| "grad_norm": 0.31187689304351807, | |
| "learning_rate": 1.209280742707828e-06, | |
| "loss": 0.3051, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 1.722010662604722, | |
| "grad_norm": 0.30652114748954773, | |
| "learning_rate": 1.134006936269999e-06, | |
| "loss": 0.3267, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 1.7311500380807312, | |
| "grad_norm": 0.33868396282196045, | |
| "learning_rate": 1.0610108779258043e-06, | |
| "loss": 0.3171, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 1.7402894135567402, | |
| "grad_norm": 0.3216879963874817, | |
| "learning_rate": 9.903113209758098e-07, | |
| "loss": 0.2924, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 1.7494287890327493, | |
| "grad_norm": 0.2990855872631073, | |
| "learning_rate": 9.2192642873008e-07, | |
| "loss": 0.3193, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 1.7585681645087585, | |
| "grad_norm": 0.2846771478652954, | |
| "learning_rate": 8.558737698418762e-07, | |
| "loss": 0.2915, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 1.7677075399847677, | |
| "grad_norm": 0.30743762850761414, | |
| "learning_rate": 7.921703137941172e-07, | |
| "loss": 0.3139, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 1.7768469154607769, | |
| "grad_norm": 0.30846983194351196, | |
| "learning_rate": 7.308324265397837e-07, | |
| "loss": 0.31, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 1.785986290936786, | |
| "grad_norm": 0.26738885045051575, | |
| "learning_rate": 6.718758662973524e-07, | |
| "loss": 0.3264, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 1.7951256664127953, | |
| "grad_norm": 0.3010248839855194, | |
| "learning_rate": 6.153157795023956e-07, | |
| "loss": 0.3187, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 1.8042650418888042, | |
| "grad_norm": 0.29657527804374695, | |
| "learning_rate": 5.611666969163243e-07, | |
| "loss": 0.3249, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 1.8134044173648134, | |
| "grad_norm": 0.29564720392227173, | |
| "learning_rate": 5.094425298933136e-07, | |
| "loss": 0.3174, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 1.8225437928408226, | |
| "grad_norm": 0.2934928834438324, | |
| "learning_rate": 4.6015656680636234e-07, | |
| "loss": 0.3278, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 1.8316831683168315, | |
| "grad_norm": 0.2969776690006256, | |
| "learning_rate": 4.133214696333943e-07, | |
| "loss": 0.3167, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.8408225437928407, | |
| "grad_norm": 0.3145069181919098, | |
| "learning_rate": 3.6894927070429744e-07, | |
| "loss": 0.3189, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 1.84996191926885, | |
| "grad_norm": 0.2936561703681946, | |
| "learning_rate": 3.2705136960970554e-07, | |
| "loss": 0.3153, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 1.859101294744859, | |
| "grad_norm": 0.3092540502548218, | |
| "learning_rate": 2.8763853027236277e-07, | |
| "loss": 0.3178, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 1.8682406702208683, | |
| "grad_norm": 0.3078237473964691, | |
| "learning_rate": 2.507208781817638e-07, | |
| "loss": 0.3042, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 1.8773800456968774, | |
| "grad_norm": 0.31245988607406616, | |
| "learning_rate": 2.1630789779284677e-07, | |
| "loss": 0.3121, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 1.8865194211728866, | |
| "grad_norm": 0.29274481534957886, | |
| "learning_rate": 1.844084300893456e-07, | |
| "loss": 0.3083, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 1.8956587966488958, | |
| "grad_norm": 0.27361243963241577, | |
| "learning_rate": 1.55030670312476e-07, | |
| "loss": 0.3198, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 1.9047981721249048, | |
| "grad_norm": 0.28884297609329224, | |
| "learning_rate": 1.2818216585549824e-07, | |
| "loss": 0.3137, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 1.913937547600914, | |
| "grad_norm": 0.3390906751155853, | |
| "learning_rate": 1.0386981432474075e-07, | |
| "loss": 0.3199, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 1.9230769230769231, | |
| "grad_norm": 0.3025045096874237, | |
| "learning_rate": 8.209986176753947e-08, | |
| "loss": 0.3091, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 1.932216298552932, | |
| "grad_norm": 0.30933359265327454, | |
| "learning_rate": 6.287790106757396e-08, | |
| "loss": 0.3267, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 1.9413556740289413, | |
| "grad_norm": 0.3036198914051056, | |
| "learning_rate": 4.6208870508017703e-08, | |
| "loss": 0.3084, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 1.9504950495049505, | |
| "grad_norm": 0.2938263416290283, | |
| "learning_rate": 3.2097052502843005e-08, | |
| "loss": 0.308, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 1.9596344249809596, | |
| "grad_norm": 0.2950015664100647, | |
| "learning_rate": 2.054607249663665e-08, | |
| "loss": 0.2959, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 1.9687738004569688, | |
| "grad_norm": 0.28699854016304016, | |
| "learning_rate": 1.1558898033191545e-08, | |
| "loss": 0.325, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 1.977913175932978, | |
| "grad_norm": 0.3103504478931427, | |
| "learning_rate": 5.137837993121064e-09, | |
| "loss": 0.3208, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 1.9870525514089872, | |
| "grad_norm": 0.30485469102859497, | |
| "learning_rate": 1.2845420006879494e-09, | |
| "loss": 0.3123, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 1.9961919268849961, | |
| "grad_norm": 0.3055669367313385, | |
| "learning_rate": 0.0, | |
| "loss": 0.3137, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 1.9961919268849961, | |
| "step": 218, | |
| "total_flos": 418179526819840.0, | |
| "train_loss": 0.09776416790047916, | |
| "train_runtime": 5460.3908, | |
| "train_samples_per_second": 3.846, | |
| "train_steps_per_second": 0.04 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 218, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 2, | |
| "save_steps": 50, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 418179526819840.0, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |