| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.0, | |
| "eval_steps": 500, | |
| "global_step": 1052, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0019011406844106464, | |
| "grad_norm": 3.85232468702254, | |
| "learning_rate": 1.886792452830189e-07, | |
| "loss": 0.6986, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0038022813688212928, | |
| "grad_norm": 3.406695706371158, | |
| "learning_rate": 3.773584905660378e-07, | |
| "loss": 0.7053, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.005703422053231939, | |
| "grad_norm": 3.6315672097786442, | |
| "learning_rate": 5.660377358490567e-07, | |
| "loss": 0.6952, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.0076045627376425855, | |
| "grad_norm": 3.174199847140795, | |
| "learning_rate": 7.547169811320755e-07, | |
| "loss": 0.6555, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.009505703422053232, | |
| "grad_norm": 3.47033914985571, | |
| "learning_rate": 9.433962264150944e-07, | |
| "loss": 0.7149, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.011406844106463879, | |
| "grad_norm": 3.193940149556615, | |
| "learning_rate": 1.1320754716981133e-06, | |
| "loss": 0.6664, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.013307984790874524, | |
| "grad_norm": 3.3122323512930194, | |
| "learning_rate": 1.3207547169811322e-06, | |
| "loss": 0.6893, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.015209125475285171, | |
| "grad_norm": 3.2255575536622523, | |
| "learning_rate": 1.509433962264151e-06, | |
| "loss": 0.716, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.017110266159695818, | |
| "grad_norm": 2.751792870659948, | |
| "learning_rate": 1.6981132075471698e-06, | |
| "loss": 0.6724, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.019011406844106463, | |
| "grad_norm": 2.6660624489041607, | |
| "learning_rate": 1.8867924528301889e-06, | |
| "loss": 0.651, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.02091254752851711, | |
| "grad_norm": 1.9450973263803515, | |
| "learning_rate": 2.075471698113208e-06, | |
| "loss": 0.6265, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.022813688212927757, | |
| "grad_norm": 1.7088614215806133, | |
| "learning_rate": 2.2641509433962266e-06, | |
| "loss": 0.5762, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.024714828897338403, | |
| "grad_norm": 1.644616443985957, | |
| "learning_rate": 2.4528301886792453e-06, | |
| "loss": 0.5882, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.026615969581749048, | |
| "grad_norm": 1.4986850530948101, | |
| "learning_rate": 2.6415094339622644e-06, | |
| "loss": 0.5518, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.028517110266159697, | |
| "grad_norm": 1.9080520756163941, | |
| "learning_rate": 2.830188679245283e-06, | |
| "loss": 0.5697, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.030418250950570342, | |
| "grad_norm": 2.2728626355621264, | |
| "learning_rate": 3.018867924528302e-06, | |
| "loss": 0.557, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.03231939163498099, | |
| "grad_norm": 2.271962656759812, | |
| "learning_rate": 3.207547169811321e-06, | |
| "loss": 0.5686, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.034220532319391636, | |
| "grad_norm": 1.8255382659254265, | |
| "learning_rate": 3.3962264150943395e-06, | |
| "loss": 0.5537, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.03612167300380228, | |
| "grad_norm": 1.5879200808542495, | |
| "learning_rate": 3.5849056603773586e-06, | |
| "loss": 0.5331, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.03802281368821293, | |
| "grad_norm": 1.4416070513323247, | |
| "learning_rate": 3.7735849056603777e-06, | |
| "loss": 0.5279, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.039923954372623575, | |
| "grad_norm": 1.565456787885237, | |
| "learning_rate": 3.962264150943396e-06, | |
| "loss": 0.5288, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.04182509505703422, | |
| "grad_norm": 1.4925085647369567, | |
| "learning_rate": 4.150943396226416e-06, | |
| "loss": 0.5314, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.043726235741444866, | |
| "grad_norm": 1.4120408881327184, | |
| "learning_rate": 4.339622641509435e-06, | |
| "loss": 0.5105, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.045627376425855515, | |
| "grad_norm": 1.1574565248209154, | |
| "learning_rate": 4.528301886792453e-06, | |
| "loss": 0.5191, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.04752851711026616, | |
| "grad_norm": 1.121651772257867, | |
| "learning_rate": 4.716981132075472e-06, | |
| "loss": 0.5257, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.049429657794676805, | |
| "grad_norm": 1.1109759098135985, | |
| "learning_rate": 4.905660377358491e-06, | |
| "loss": 0.5302, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.051330798479087454, | |
| "grad_norm": 1.0923441357297827, | |
| "learning_rate": 5.09433962264151e-06, | |
| "loss": 0.5224, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.053231939163498096, | |
| "grad_norm": 1.0702270093566981, | |
| "learning_rate": 5.283018867924529e-06, | |
| "loss": 0.5191, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.055133079847908745, | |
| "grad_norm": 1.09857830748445, | |
| "learning_rate": 5.4716981132075475e-06, | |
| "loss": 0.4919, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.057034220532319393, | |
| "grad_norm": 1.004091118236633, | |
| "learning_rate": 5.660377358490566e-06, | |
| "loss": 0.5, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.058935361216730035, | |
| "grad_norm": 1.0388699354391888, | |
| "learning_rate": 5.849056603773585e-06, | |
| "loss": 0.4918, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.060836501901140684, | |
| "grad_norm": 0.9197436489231963, | |
| "learning_rate": 6.037735849056604e-06, | |
| "loss": 0.4556, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.06273764258555133, | |
| "grad_norm": 0.9658443965119134, | |
| "learning_rate": 6.226415094339623e-06, | |
| "loss": 0.4786, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.06463878326996197, | |
| "grad_norm": 0.9767807722352592, | |
| "learning_rate": 6.415094339622642e-06, | |
| "loss": 0.4881, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.06653992395437262, | |
| "grad_norm": 0.9678771570174117, | |
| "learning_rate": 6.60377358490566e-06, | |
| "loss": 0.4872, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.06844106463878327, | |
| "grad_norm": 0.9045356908532408, | |
| "learning_rate": 6.792452830188679e-06, | |
| "loss": 0.4674, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.07034220532319392, | |
| "grad_norm": 0.904826088102294, | |
| "learning_rate": 6.981132075471699e-06, | |
| "loss": 0.4546, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.07224334600760456, | |
| "grad_norm": 1.0470805644016272, | |
| "learning_rate": 7.169811320754717e-06, | |
| "loss": 0.4775, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.0741444866920152, | |
| "grad_norm": 0.9331663309918206, | |
| "learning_rate": 7.358490566037736e-06, | |
| "loss": 0.4867, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.07604562737642585, | |
| "grad_norm": 0.963879348960802, | |
| "learning_rate": 7.5471698113207555e-06, | |
| "loss": 0.4565, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.0779467680608365, | |
| "grad_norm": 1.0361589547202308, | |
| "learning_rate": 7.735849056603775e-06, | |
| "loss": 0.4757, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.07984790874524715, | |
| "grad_norm": 0.8787102373463636, | |
| "learning_rate": 7.924528301886793e-06, | |
| "loss": 0.4529, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.0817490494296578, | |
| "grad_norm": 0.9282766286904502, | |
| "learning_rate": 8.113207547169812e-06, | |
| "loss": 0.4594, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.08365019011406843, | |
| "grad_norm": 0.9710964965222922, | |
| "learning_rate": 8.301886792452832e-06, | |
| "loss": 0.4496, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.08555133079847908, | |
| "grad_norm": 0.878993248631418, | |
| "learning_rate": 8.49056603773585e-06, | |
| "loss": 0.4514, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.08745247148288973, | |
| "grad_norm": 0.9380217837364928, | |
| "learning_rate": 8.67924528301887e-06, | |
| "loss": 0.4609, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.08935361216730038, | |
| "grad_norm": 0.898895301278193, | |
| "learning_rate": 8.867924528301887e-06, | |
| "loss": 0.4672, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.09125475285171103, | |
| "grad_norm": 0.9007389663701948, | |
| "learning_rate": 9.056603773584907e-06, | |
| "loss": 0.4641, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.09315589353612168, | |
| "grad_norm": 1.0013000447642713, | |
| "learning_rate": 9.245283018867926e-06, | |
| "loss": 0.456, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.09505703422053231, | |
| "grad_norm": 0.9002493544919404, | |
| "learning_rate": 9.433962264150944e-06, | |
| "loss": 0.4649, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.09695817490494296, | |
| "grad_norm": 0.8834448824894418, | |
| "learning_rate": 9.622641509433963e-06, | |
| "loss": 0.4549, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.09885931558935361, | |
| "grad_norm": 0.8883252786516431, | |
| "learning_rate": 9.811320754716981e-06, | |
| "loss": 0.4597, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.10076045627376426, | |
| "grad_norm": 0.9317907119949653, | |
| "learning_rate": 1e-05, | |
| "loss": 0.4539, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.10266159695817491, | |
| "grad_norm": 0.9069976707570846, | |
| "learning_rate": 1.018867924528302e-05, | |
| "loss": 0.4529, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.10456273764258556, | |
| "grad_norm": 1.0138048694575552, | |
| "learning_rate": 1.0377358490566038e-05, | |
| "loss": 0.4729, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.10646387832699619, | |
| "grad_norm": 1.0672886764115217, | |
| "learning_rate": 1.0566037735849058e-05, | |
| "loss": 0.4562, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.10836501901140684, | |
| "grad_norm": 0.8495301874150771, | |
| "learning_rate": 1.0754716981132076e-05, | |
| "loss": 0.4317, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.11026615969581749, | |
| "grad_norm": 1.1197277415559004, | |
| "learning_rate": 1.0943396226415095e-05, | |
| "loss": 0.4622, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.11216730038022814, | |
| "grad_norm": 0.9242003056122978, | |
| "learning_rate": 1.1132075471698115e-05, | |
| "loss": 0.4498, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.11406844106463879, | |
| "grad_norm": 0.9861245491198054, | |
| "learning_rate": 1.1320754716981132e-05, | |
| "loss": 0.4697, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.11596958174904944, | |
| "grad_norm": 0.9364531126386214, | |
| "learning_rate": 1.1509433962264152e-05, | |
| "loss": 0.4299, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.11787072243346007, | |
| "grad_norm": 0.9269665071240559, | |
| "learning_rate": 1.169811320754717e-05, | |
| "loss": 0.4481, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.11977186311787072, | |
| "grad_norm": 0.9602180260966944, | |
| "learning_rate": 1.188679245283019e-05, | |
| "loss": 0.4424, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.12167300380228137, | |
| "grad_norm": 1.0558108767090975, | |
| "learning_rate": 1.2075471698113209e-05, | |
| "loss": 0.4869, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.12357414448669202, | |
| "grad_norm": 0.917333154745973, | |
| "learning_rate": 1.2264150943396227e-05, | |
| "loss": 0.4591, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.12547528517110265, | |
| "grad_norm": 1.1115485765548476, | |
| "learning_rate": 1.2452830188679246e-05, | |
| "loss": 0.451, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.12737642585551331, | |
| "grad_norm": 0.8535988189387356, | |
| "learning_rate": 1.2641509433962264e-05, | |
| "loss": 0.4506, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.12927756653992395, | |
| "grad_norm": 0.9333091398426233, | |
| "learning_rate": 1.2830188679245283e-05, | |
| "loss": 0.4569, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.1311787072243346, | |
| "grad_norm": 1.121079222420052, | |
| "learning_rate": 1.3018867924528303e-05, | |
| "loss": 0.4736, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.13307984790874525, | |
| "grad_norm": 0.88369673751432, | |
| "learning_rate": 1.320754716981132e-05, | |
| "loss": 0.4373, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.13498098859315588, | |
| "grad_norm": 1.1182305784539315, | |
| "learning_rate": 1.339622641509434e-05, | |
| "loss": 0.4451, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.13688212927756654, | |
| "grad_norm": 1.011565506044473, | |
| "learning_rate": 1.3584905660377358e-05, | |
| "loss": 0.4591, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.13878326996197718, | |
| "grad_norm": 1.1151745156183657, | |
| "learning_rate": 1.3773584905660378e-05, | |
| "loss": 0.4496, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.14068441064638784, | |
| "grad_norm": 0.8577625638905495, | |
| "learning_rate": 1.3962264150943397e-05, | |
| "loss": 0.4224, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.14258555133079848, | |
| "grad_norm": 0.9562724157685919, | |
| "learning_rate": 1.4150943396226415e-05, | |
| "loss": 0.4272, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.1444866920152091, | |
| "grad_norm": 0.8719843282722944, | |
| "learning_rate": 1.4339622641509435e-05, | |
| "loss": 0.454, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.14638783269961977, | |
| "grad_norm": 0.9734976318137991, | |
| "learning_rate": 1.4528301886792452e-05, | |
| "loss": 0.4703, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.1482889733840304, | |
| "grad_norm": 0.8809117960623533, | |
| "learning_rate": 1.4716981132075472e-05, | |
| "loss": 0.4394, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.15019011406844107, | |
| "grad_norm": 1.038804232427567, | |
| "learning_rate": 1.4905660377358491e-05, | |
| "loss": 0.4451, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.1520912547528517, | |
| "grad_norm": 0.9305040800945201, | |
| "learning_rate": 1.5094339622641511e-05, | |
| "loss": 0.4608, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.15399239543726237, | |
| "grad_norm": 0.9203318824616789, | |
| "learning_rate": 1.5283018867924532e-05, | |
| "loss": 0.4573, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.155893536121673, | |
| "grad_norm": 0.8562521890057109, | |
| "learning_rate": 1.547169811320755e-05, | |
| "loss": 0.4304, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.15779467680608364, | |
| "grad_norm": 1.108428974677605, | |
| "learning_rate": 1.5660377358490568e-05, | |
| "loss": 0.4554, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.1596958174904943, | |
| "grad_norm": 1.794791773247, | |
| "learning_rate": 1.5849056603773586e-05, | |
| "loss": 0.4488, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.16159695817490494, | |
| "grad_norm": 0.9743055030407272, | |
| "learning_rate": 1.6037735849056607e-05, | |
| "loss": 0.4349, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.1634980988593156, | |
| "grad_norm": 1.0447461649688048, | |
| "learning_rate": 1.6226415094339625e-05, | |
| "loss": 0.4264, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.16539923954372623, | |
| "grad_norm": 0.9737136332580937, | |
| "learning_rate": 1.6415094339622643e-05, | |
| "loss": 0.4522, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.16730038022813687, | |
| "grad_norm": 1.0794722274137125, | |
| "learning_rate": 1.6603773584905664e-05, | |
| "loss": 0.4486, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.16920152091254753, | |
| "grad_norm": 0.8840290914285559, | |
| "learning_rate": 1.679245283018868e-05, | |
| "loss": 0.4669, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.17110266159695817, | |
| "grad_norm": 1.1269554036060225, | |
| "learning_rate": 1.69811320754717e-05, | |
| "loss": 0.4624, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.17300380228136883, | |
| "grad_norm": 1.006613659299684, | |
| "learning_rate": 1.716981132075472e-05, | |
| "loss": 0.458, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.17490494296577946, | |
| "grad_norm": 1.0643356534252053, | |
| "learning_rate": 1.735849056603774e-05, | |
| "loss": 0.4719, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.17680608365019013, | |
| "grad_norm": 1.0256234697395523, | |
| "learning_rate": 1.7547169811320756e-05, | |
| "loss": 0.4291, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.17870722433460076, | |
| "grad_norm": 0.9876023215844656, | |
| "learning_rate": 1.7735849056603774e-05, | |
| "loss": 0.4542, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.1806083650190114, | |
| "grad_norm": 1.2323269526729317, | |
| "learning_rate": 1.7924528301886795e-05, | |
| "loss": 0.4643, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.18250950570342206, | |
| "grad_norm": 0.9479928339311949, | |
| "learning_rate": 1.8113207547169813e-05, | |
| "loss": 0.4417, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.1844106463878327, | |
| "grad_norm": 1.1816980786569204, | |
| "learning_rate": 1.830188679245283e-05, | |
| "loss": 0.4584, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.18631178707224336, | |
| "grad_norm": 0.8948311016439893, | |
| "learning_rate": 1.8490566037735852e-05, | |
| "loss": 0.4198, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.188212927756654, | |
| "grad_norm": 1.0006387723189951, | |
| "learning_rate": 1.867924528301887e-05, | |
| "loss": 0.4537, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.19011406844106463, | |
| "grad_norm": 1.0644157220591428, | |
| "learning_rate": 1.8867924528301888e-05, | |
| "loss": 0.4495, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.1920152091254753, | |
| "grad_norm": 0.8862896783873369, | |
| "learning_rate": 1.905660377358491e-05, | |
| "loss": 0.455, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.19391634980988592, | |
| "grad_norm": 1.1355058664798832, | |
| "learning_rate": 1.9245283018867927e-05, | |
| "loss": 0.4402, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.1958174904942966, | |
| "grad_norm": 0.8687345304186009, | |
| "learning_rate": 1.9433962264150945e-05, | |
| "loss": 0.4649, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.19771863117870722, | |
| "grad_norm": 1.0233554005425392, | |
| "learning_rate": 1.9622641509433963e-05, | |
| "loss": 0.4404, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.19961977186311788, | |
| "grad_norm": 1.0081198307556802, | |
| "learning_rate": 1.9811320754716984e-05, | |
| "loss": 0.4548, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.20152091254752852, | |
| "grad_norm": 2.5185528300064677, | |
| "learning_rate": 2e-05, | |
| "loss": 0.4474, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.20342205323193915, | |
| "grad_norm": 1.2518103496271424, | |
| "learning_rate": 1.9999944857420527e-05, | |
| "loss": 0.4305, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.20532319391634982, | |
| "grad_norm": 5.833595316332377, | |
| "learning_rate": 1.9999779430290247e-05, | |
| "loss": 0.4642, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.20722433460076045, | |
| "grad_norm": 4.994297472787273, | |
| "learning_rate": 1.9999503720433575e-05, | |
| "loss": 0.503, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.20912547528517111, | |
| "grad_norm": 2.3951510419223476, | |
| "learning_rate": 1.999911773089118e-05, | |
| "loss": 0.4566, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.21102661596958175, | |
| "grad_norm": 1.3724017134479995, | |
| "learning_rate": 1.999862146591996e-05, | |
| "loss": 0.4442, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.21292775665399238, | |
| "grad_norm": 0.9293779651737607, | |
| "learning_rate": 1.9998014930992976e-05, | |
| "loss": 0.4508, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.21482889733840305, | |
| "grad_norm": 1.1116690187368958, | |
| "learning_rate": 1.9997298132799408e-05, | |
| "loss": 0.452, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.21673003802281368, | |
| "grad_norm": 131.41952059960505, | |
| "learning_rate": 1.9996471079244477e-05, | |
| "loss": 0.4875, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.21863117870722434, | |
| "grad_norm": 1.0259474707125407, | |
| "learning_rate": 1.999553377944936e-05, | |
| "loss": 0.4311, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.22053231939163498, | |
| "grad_norm": 0.9163317003904157, | |
| "learning_rate": 1.9994486243751076e-05, | |
| "loss": 0.4643, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.2224334600760456, | |
| "grad_norm": 0.890643493646681, | |
| "learning_rate": 1.9993328483702393e-05, | |
| "loss": 0.4481, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.22433460076045628, | |
| "grad_norm": 0.8363012677932206, | |
| "learning_rate": 1.999206051207169e-05, | |
| "loss": 0.4464, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.2262357414448669, | |
| "grad_norm": 0.9228078870644588, | |
| "learning_rate": 1.9990682342842805e-05, | |
| "loss": 0.4505, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.22813688212927757, | |
| "grad_norm": 0.9111147900984897, | |
| "learning_rate": 1.99891939912149e-05, | |
| "loss": 0.4291, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.2300380228136882, | |
| "grad_norm": 0.882486376849749, | |
| "learning_rate": 1.9987595473602292e-05, | |
| "loss": 0.46, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.23193916349809887, | |
| "grad_norm": 1.0406449481615851, | |
| "learning_rate": 1.9985886807634246e-05, | |
| "loss": 0.4752, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.2338403041825095, | |
| "grad_norm": 0.7553447832039742, | |
| "learning_rate": 1.9984068012154824e-05, | |
| "loss": 0.4235, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.23574144486692014, | |
| "grad_norm": 0.9251205029491636, | |
| "learning_rate": 1.9982139107222634e-05, | |
| "loss": 0.4483, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.2376425855513308, | |
| "grad_norm": 0.9120656132979653, | |
| "learning_rate": 1.9980100114110637e-05, | |
| "loss": 0.4394, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.23954372623574144, | |
| "grad_norm": 0.891000963107638, | |
| "learning_rate": 1.99779510553059e-05, | |
| "loss": 0.4473, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.2414448669201521, | |
| "grad_norm": 1.0108336972790397, | |
| "learning_rate": 1.9975691954509347e-05, | |
| "loss": 0.4597, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.24334600760456274, | |
| "grad_norm": 0.8688698546672481, | |
| "learning_rate": 1.9973322836635517e-05, | |
| "loss": 0.4324, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.24524714828897337, | |
| "grad_norm": 1.1251082491744038, | |
| "learning_rate": 1.997084372781226e-05, | |
| "loss": 0.4694, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.24714828897338403, | |
| "grad_norm": 1.008160592615569, | |
| "learning_rate": 1.9968254655380465e-05, | |
| "loss": 0.4356, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.24904942965779467, | |
| "grad_norm": 1.0086460469716076, | |
| "learning_rate": 1.996555564789376e-05, | |
| "loss": 0.4481, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.2509505703422053, | |
| "grad_norm": 1.100676154598148, | |
| "learning_rate": 1.996274673511819e-05, | |
| "loss": 0.425, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.25285171102661597, | |
| "grad_norm": 0.9237138198861893, | |
| "learning_rate": 1.99598279480319e-05, | |
| "loss": 0.4456, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.25475285171102663, | |
| "grad_norm": 0.9986356041163961, | |
| "learning_rate": 1.9956799318824776e-05, | |
| "loss": 0.4307, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.25665399239543724, | |
| "grad_norm": 0.8970448000295634, | |
| "learning_rate": 1.99536608808981e-05, | |
| "loss": 0.4591, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.2585551330798479, | |
| "grad_norm": 1.1322198663109666, | |
| "learning_rate": 1.995041266886419e-05, | |
| "loss": 0.4398, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.26045627376425856, | |
| "grad_norm": 0.9445960060848516, | |
| "learning_rate": 1.9947054718545996e-05, | |
| "loss": 0.4496, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.2623574144486692, | |
| "grad_norm": 0.8499391062808022, | |
| "learning_rate": 1.994358706697674e-05, | |
| "loss": 0.4514, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.26425855513307983, | |
| "grad_norm": 1.1634659526177034, | |
| "learning_rate": 1.9940009752399462e-05, | |
| "loss": 0.4575, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.2661596958174905, | |
| "grad_norm": 0.9029922581510964, | |
| "learning_rate": 1.9936322814266634e-05, | |
| "loss": 0.4624, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.26806083650190116, | |
| "grad_norm": 1.1061056693858171, | |
| "learning_rate": 1.9932526293239713e-05, | |
| "loss": 0.4482, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.26996197718631176, | |
| "grad_norm": 1.3325525574873587, | |
| "learning_rate": 1.9928620231188694e-05, | |
| "loss": 0.4437, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.2718631178707224, | |
| "grad_norm": 0.8733018579505439, | |
| "learning_rate": 1.992460467119164e-05, | |
| "loss": 0.4506, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.2737642585551331, | |
| "grad_norm": 1.2606397109469378, | |
| "learning_rate": 1.992047965753422e-05, | |
| "loss": 0.4625, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.27566539923954375, | |
| "grad_norm": 0.9916868161280936, | |
| "learning_rate": 1.991624523570922e-05, | |
| "loss": 0.4541, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.27756653992395436, | |
| "grad_norm": 0.98155135523797, | |
| "learning_rate": 1.9911901452416012e-05, | |
| "loss": 0.4386, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.279467680608365, | |
| "grad_norm": 1.0391744667894165, | |
| "learning_rate": 1.9907448355560094e-05, | |
| "loss": 0.4397, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.2813688212927757, | |
| "grad_norm": 0.9766895766674882, | |
| "learning_rate": 1.9902885994252506e-05, | |
| "loss": 0.4534, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.2832699619771863, | |
| "grad_norm": 1.0031287640616235, | |
| "learning_rate": 1.989821441880933e-05, | |
| "loss": 0.4564, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.28517110266159695, | |
| "grad_norm": 1.029930760089183, | |
| "learning_rate": 1.9893433680751105e-05, | |
| "loss": 0.4301, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.2870722433460076, | |
| "grad_norm": 0.8093323822809237, | |
| "learning_rate": 1.9888543832802277e-05, | |
| "loss": 0.4231, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.2889733840304182, | |
| "grad_norm": 0.8822022185067913, | |
| "learning_rate": 1.9883544928890612e-05, | |
| "loss": 0.4494, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.2908745247148289, | |
| "grad_norm": 0.9916974639740077, | |
| "learning_rate": 1.9878437024146603e-05, | |
| "loss": 0.4753, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.29277566539923955, | |
| "grad_norm": 0.9240367951928649, | |
| "learning_rate": 1.9873220174902857e-05, | |
| "loss": 0.445, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.2946768060836502, | |
| "grad_norm": 0.999043605251974, | |
| "learning_rate": 1.986789443869348e-05, | |
| "loss": 0.4553, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.2965779467680608, | |
| "grad_norm": 0.9909532964389867, | |
| "learning_rate": 1.9862459874253438e-05, | |
| "loss": 0.425, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.2984790874524715, | |
| "grad_norm": 0.8096957082828341, | |
| "learning_rate": 1.985691654151791e-05, | |
| "loss": 0.4297, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.30038022813688214, | |
| "grad_norm": 1.046788222724364, | |
| "learning_rate": 1.9851264501621635e-05, | |
| "loss": 0.4518, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.30228136882129275, | |
| "grad_norm": 1.101543376648695, | |
| "learning_rate": 1.984550381689822e-05, | |
| "loss": 0.455, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.3041825095057034, | |
| "grad_norm": 0.9233982122578147, | |
| "learning_rate": 1.983963455087946e-05, | |
| "loss": 0.4555, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.3060836501901141, | |
| "grad_norm": 1.0965011625123815, | |
| "learning_rate": 1.983365676829466e-05, | |
| "loss": 0.4526, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.30798479087452474, | |
| "grad_norm": 0.8496810748805607, | |
| "learning_rate": 1.982757053506989e-05, | |
| "loss": 0.4263, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.30988593155893535, | |
| "grad_norm": 0.9463087299965671, | |
| "learning_rate": 1.9821375918327268e-05, | |
| "loss": 0.436, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.311787072243346, | |
| "grad_norm": 0.9099139061345621, | |
| "learning_rate": 1.981507298638422e-05, | |
| "loss": 0.4247, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.31368821292775667, | |
| "grad_norm": 0.8062359216251765, | |
| "learning_rate": 1.9808661808752735e-05, | |
| "loss": 0.4397, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.3155893536121673, | |
| "grad_norm": 0.9559431598551538, | |
| "learning_rate": 1.980214245613858e-05, | |
| "loss": 0.4453, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.31749049429657794, | |
| "grad_norm": 0.8792602821761563, | |
| "learning_rate": 1.979551500044055e-05, | |
| "loss": 0.4425, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.3193916349809886, | |
| "grad_norm": 0.8295910226677302, | |
| "learning_rate": 1.9788779514749635e-05, | |
| "loss": 0.4615, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.32129277566539927, | |
| "grad_norm": 1.0360052465649607, | |
| "learning_rate": 1.978193607334826e-05, | |
| "loss": 0.4323, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.3231939163498099, | |
| "grad_norm": 0.8262900971663927, | |
| "learning_rate": 1.977498475170941e-05, | |
| "loss": 0.4496, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.32509505703422054, | |
| "grad_norm": 0.8755224178635944, | |
| "learning_rate": 1.9767925626495857e-05, | |
| "loss": 0.4479, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.3269961977186312, | |
| "grad_norm": 0.8713119794820382, | |
| "learning_rate": 1.9760758775559275e-05, | |
| "loss": 0.4495, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.3288973384030418, | |
| "grad_norm": 0.8188246401516779, | |
| "learning_rate": 1.975348427793939e-05, | |
| "loss": 0.4487, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.33079847908745247, | |
| "grad_norm": 0.9161927449797218, | |
| "learning_rate": 1.9746102213863113e-05, | |
| "loss": 0.4454, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.33269961977186313, | |
| "grad_norm": 0.7845024530606038, | |
| "learning_rate": 1.973861266474366e-05, | |
| "loss": 0.4398, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.33460076045627374, | |
| "grad_norm": 0.8720421222011884, | |
| "learning_rate": 1.9731015713179643e-05, | |
| "loss": 0.4444, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.3365019011406844, | |
| "grad_norm": 0.8951116716352723, | |
| "learning_rate": 1.9723311442954163e-05, | |
| "loss": 0.4559, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.33840304182509506, | |
| "grad_norm": 0.8117756458013391, | |
| "learning_rate": 1.9715499939033883e-05, | |
| "loss": 0.4201, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.3403041825095057, | |
| "grad_norm": 0.8306814053317432, | |
| "learning_rate": 1.9707581287568094e-05, | |
| "loss": 0.4222, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.34220532319391633, | |
| "grad_norm": 0.7903513048375812, | |
| "learning_rate": 1.969955557588778e-05, | |
| "loss": 0.4275, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.344106463878327, | |
| "grad_norm": 0.811242191921134, | |
| "learning_rate": 1.9691422892504626e-05, | |
| "loss": 0.4265, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.34600760456273766, | |
| "grad_norm": 0.8185181647002605, | |
| "learning_rate": 1.968318332711006e-05, | |
| "loss": 0.4526, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.34790874524714827, | |
| "grad_norm": 0.8532286623160587, | |
| "learning_rate": 1.9674836970574253e-05, | |
| "loss": 0.4779, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.34980988593155893, | |
| "grad_norm": 0.8889699563409826, | |
| "learning_rate": 1.966638391494514e-05, | |
| "loss": 0.4382, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.3517110266159696, | |
| "grad_norm": 0.8794410877379305, | |
| "learning_rate": 1.9657824253447378e-05, | |
| "loss": 0.4237, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.35361216730038025, | |
| "grad_norm": 0.9542789266542244, | |
| "learning_rate": 1.9649158080481327e-05, | |
| "loss": 0.4532, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.35551330798479086, | |
| "grad_norm": 0.813112710571308, | |
| "learning_rate": 1.964038549162201e-05, | |
| "loss": 0.4193, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.3574144486692015, | |
| "grad_norm": 0.8561337453355, | |
| "learning_rate": 1.963150658361807e-05, | |
| "loss": 0.4379, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.3593155893536122, | |
| "grad_norm": 0.7953064825748455, | |
| "learning_rate": 1.962252145439068e-05, | |
| "loss": 0.4247, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.3612167300380228, | |
| "grad_norm": 0.8207229317764647, | |
| "learning_rate": 1.9613430203032486e-05, | |
| "loss": 0.4213, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.36311787072243346, | |
| "grad_norm": 1.0001161826404577, | |
| "learning_rate": 1.9604232929806493e-05, | |
| "loss": 0.4433, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.3650190114068441, | |
| "grad_norm": 0.8862440782801183, | |
| "learning_rate": 1.9594929736144978e-05, | |
| "loss": 0.4279, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.3669201520912547, | |
| "grad_norm": 0.8290509092227688, | |
| "learning_rate": 1.9585520724648354e-05, | |
| "loss": 0.4369, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.3688212927756654, | |
| "grad_norm": 1.1637531751203498, | |
| "learning_rate": 1.957600599908406e-05, | |
| "loss": 0.4266, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.37072243346007605, | |
| "grad_norm": 0.891249791531336, | |
| "learning_rate": 1.95663856643854e-05, | |
| "loss": 0.4447, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.3726235741444867, | |
| "grad_norm": 0.9941154684468924, | |
| "learning_rate": 1.955665982665038e-05, | |
| "loss": 0.4206, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.3745247148288973, | |
| "grad_norm": 0.8993031553077088, | |
| "learning_rate": 1.9546828593140565e-05, | |
| "loss": 0.4246, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.376425855513308, | |
| "grad_norm": 0.83249046996326, | |
| "learning_rate": 1.9536892072279863e-05, | |
| "loss": 0.4494, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.37832699619771865, | |
| "grad_norm": 1.000558377221497, | |
| "learning_rate": 1.9526850373653356e-05, | |
| "loss": 0.4675, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.38022813688212925, | |
| "grad_norm": 0.7278198919318765, | |
| "learning_rate": 1.9516703608006074e-05, | |
| "loss": 0.3944, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.3821292775665399, | |
| "grad_norm": 0.8639074128919991, | |
| "learning_rate": 1.9506451887241787e-05, | |
| "loss": 0.4258, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.3840304182509506, | |
| "grad_norm": 0.7763042827871615, | |
| "learning_rate": 1.949609532442176e-05, | |
| "loss": 0.4379, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.38593155893536124, | |
| "grad_norm": 0.8157811646368555, | |
| "learning_rate": 1.9485634033763507e-05, | |
| "loss": 0.4221, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.38783269961977185, | |
| "grad_norm": 0.7982160869239662, | |
| "learning_rate": 1.9475068130639543e-05, | |
| "loss": 0.4156, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.3897338403041825, | |
| "grad_norm": 0.8688608532034439, | |
| "learning_rate": 1.9464397731576093e-05, | |
| "loss": 0.4401, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.3916349809885932, | |
| "grad_norm": 0.8280781032688402, | |
| "learning_rate": 1.945362295425183e-05, | |
| "loss": 0.4333, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.3935361216730038, | |
| "grad_norm": 0.7266630066317653, | |
| "learning_rate": 1.944274391749655e-05, | |
| "loss": 0.4225, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.39543726235741444, | |
| "grad_norm": 0.7772785763924482, | |
| "learning_rate": 1.9431760741289886e-05, | |
| "loss": 0.4183, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.3973384030418251, | |
| "grad_norm": 0.8034556650941019, | |
| "learning_rate": 1.942067354675997e-05, | |
| "loss": 0.4405, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.39923954372623577, | |
| "grad_norm": 0.7718834751426096, | |
| "learning_rate": 1.9409482456182105e-05, | |
| "loss": 0.4229, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.4011406844106464, | |
| "grad_norm": 0.836438140607089, | |
| "learning_rate": 1.939818759297741e-05, | |
| "loss": 0.4222, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.40304182509505704, | |
| "grad_norm": 0.7802454057141343, | |
| "learning_rate": 1.9386789081711465e-05, | |
| "loss": 0.434, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.4049429657794677, | |
| "grad_norm": 0.909740774504771, | |
| "learning_rate": 1.9375287048092927e-05, | |
| "loss": 0.4375, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.4068441064638783, | |
| "grad_norm": 0.735083603259127, | |
| "learning_rate": 1.9363681618972166e-05, | |
| "loss": 0.4312, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.40874524714828897, | |
| "grad_norm": 0.762230135943909, | |
| "learning_rate": 1.9351972922339835e-05, | |
| "loss": 0.4381, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.41064638783269963, | |
| "grad_norm": 0.8549835828802903, | |
| "learning_rate": 1.9340161087325483e-05, | |
| "loss": 0.4348, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.41254752851711024, | |
| "grad_norm": 0.7626125950901438, | |
| "learning_rate": 1.9328246244196117e-05, | |
| "loss": 0.4253, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.4144486692015209, | |
| "grad_norm": 0.7425221430424228, | |
| "learning_rate": 1.931622852435478e-05, | |
| "loss": 0.4374, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.41634980988593157, | |
| "grad_norm": 0.7803512780617675, | |
| "learning_rate": 1.930410806033908e-05, | |
| "loss": 0.4258, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.41825095057034223, | |
| "grad_norm": 0.7649241966473099, | |
| "learning_rate": 1.929188498581975e-05, | |
| "loss": 0.4334, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.42015209125475284, | |
| "grad_norm": 0.7957279957101813, | |
| "learning_rate": 1.9279559435599164e-05, | |
| "loss": 0.4471, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.4220532319391635, | |
| "grad_norm": 0.7691072911843975, | |
| "learning_rate": 1.926713154560984e-05, | |
| "loss": 0.4392, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.42395437262357416, | |
| "grad_norm": 0.8377932731205584, | |
| "learning_rate": 1.9254601452912972e-05, | |
| "loss": 0.4508, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.42585551330798477, | |
| "grad_norm": 0.7475366410022324, | |
| "learning_rate": 1.924196929569688e-05, | |
| "loss": 0.4291, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.42775665399239543, | |
| "grad_norm": 0.7530490081190968, | |
| "learning_rate": 1.922923521327551e-05, | |
| "loss": 0.4408, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.4296577946768061, | |
| "grad_norm": 0.7697733237731316, | |
| "learning_rate": 1.9216399346086893e-05, | |
| "loss": 0.4299, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.43155893536121676, | |
| "grad_norm": 0.7828955553546068, | |
| "learning_rate": 1.9203461835691596e-05, | |
| "loss": 0.4376, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.43346007604562736, | |
| "grad_norm": 0.7901782180060598, | |
| "learning_rate": 1.9190422824771158e-05, | |
| "loss": 0.4377, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.435361216730038, | |
| "grad_norm": 0.6944946203439041, | |
| "learning_rate": 1.9177282457126515e-05, | |
| "loss": 0.4151, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.4372623574144487, | |
| "grad_norm": 0.8068792939243996, | |
| "learning_rate": 1.9164040877676425e-05, | |
| "loss": 0.4303, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.4391634980988593, | |
| "grad_norm": 0.7280964026420176, | |
| "learning_rate": 1.9150698232455853e-05, | |
| "loss": 0.3926, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.44106463878326996, | |
| "grad_norm": 0.7437513417574995, | |
| "learning_rate": 1.913725466861438e-05, | |
| "loss": 0.4191, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.4429657794676806, | |
| "grad_norm": 0.7782292420809169, | |
| "learning_rate": 1.9123710334414552e-05, | |
| "loss": 0.4485, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.4448669201520912, | |
| "grad_norm": 0.7401246772782446, | |
| "learning_rate": 1.911006537923029e-05, | |
| "loss": 0.4274, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.4467680608365019, | |
| "grad_norm": 0.7258045723427025, | |
| "learning_rate": 1.9096319953545186e-05, | |
| "loss": 0.4262, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.44866920152091255, | |
| "grad_norm": 0.747883242082474, | |
| "learning_rate": 1.908247420895089e-05, | |
| "loss": 0.4263, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.4505703422053232, | |
| "grad_norm": 0.7508442830311588, | |
| "learning_rate": 1.9068528298145418e-05, | |
| "loss": 0.4134, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.4524714828897338, | |
| "grad_norm": 0.8364901136893746, | |
| "learning_rate": 1.905448237493147e-05, | |
| "loss": 0.437, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.4543726235741445, | |
| "grad_norm": 0.7200277714024923, | |
| "learning_rate": 1.9040336594214727e-05, | |
| "loss": 0.3993, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.45627376425855515, | |
| "grad_norm": 0.7786312935205012, | |
| "learning_rate": 1.9026091112002163e-05, | |
| "loss": 0.4081, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.45817490494296575, | |
| "grad_norm": 0.7978351537887257, | |
| "learning_rate": 1.90117460854003e-05, | |
| "loss": 0.423, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.4600760456273764, | |
| "grad_norm": 0.7878400715410642, | |
| "learning_rate": 1.8997301672613496e-05, | |
| "loss": 0.4468, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.4619771863117871, | |
| "grad_norm": 0.7590104038455995, | |
| "learning_rate": 1.8982758032942184e-05, | |
| "loss": 0.4392, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.46387832699619774, | |
| "grad_norm": 1.0238333348300779, | |
| "learning_rate": 1.896811532678113e-05, | |
| "loss": 0.4398, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.46577946768060835, | |
| "grad_norm": 0.7580263553856883, | |
| "learning_rate": 1.8953373715617646e-05, | |
| "loss": 0.4328, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.467680608365019, | |
| "grad_norm": 0.8028159741521785, | |
| "learning_rate": 1.893853336202983e-05, | |
| "loss": 0.4134, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.4695817490494297, | |
| "grad_norm": 0.8863810718591283, | |
| "learning_rate": 1.892359442968475e-05, | |
| "loss": 0.411, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 0.4714828897338403, | |
| "grad_norm": 0.7165829287536344, | |
| "learning_rate": 1.8908557083336668e-05, | |
| "loss": 0.4222, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.47338403041825095, | |
| "grad_norm": 1.0042802161911777, | |
| "learning_rate": 1.889342148882519e-05, | |
| "loss": 0.4159, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 0.4752851711026616, | |
| "grad_norm": 0.7435479633726337, | |
| "learning_rate": 1.8878187813073465e-05, | |
| "loss": 0.4232, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.47718631178707227, | |
| "grad_norm": 0.8279455678156201, | |
| "learning_rate": 1.886285622408633e-05, | |
| "loss": 0.4172, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 0.4790874524714829, | |
| "grad_norm": 0.7632566963853045, | |
| "learning_rate": 1.8847426890948447e-05, | |
| "loss": 0.4245, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.48098859315589354, | |
| "grad_norm": 0.7847357463254118, | |
| "learning_rate": 1.8831899983822475e-05, | |
| "loss": 0.4283, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 0.4828897338403042, | |
| "grad_norm": 0.7464364006971929, | |
| "learning_rate": 1.8816275673947148e-05, | |
| "loss": 0.4076, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.4847908745247148, | |
| "grad_norm": 0.7463051440159735, | |
| "learning_rate": 1.8800554133635417e-05, | |
| "loss": 0.4036, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.4866920152091255, | |
| "grad_norm": 0.7634108082733331, | |
| "learning_rate": 1.8784735536272543e-05, | |
| "loss": 0.4254, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.48859315589353614, | |
| "grad_norm": 0.7010742759449635, | |
| "learning_rate": 1.8768820056314173e-05, | |
| "loss": 0.411, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 0.49049429657794674, | |
| "grad_norm": 0.754450327875082, | |
| "learning_rate": 1.875280786928444e-05, | |
| "loss": 0.4236, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.4923954372623574, | |
| "grad_norm": 0.872711682630628, | |
| "learning_rate": 1.873669915177399e-05, | |
| "loss": 0.4255, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 0.49429657794676807, | |
| "grad_norm": 0.7298578128493682, | |
| "learning_rate": 1.872049408143808e-05, | |
| "loss": 0.4111, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.49619771863117873, | |
| "grad_norm": 0.9401377415906165, | |
| "learning_rate": 1.8704192836994578e-05, | |
| "loss": 0.4128, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 0.49809885931558934, | |
| "grad_norm": 0.7671547241311577, | |
| "learning_rate": 1.8687795598222024e-05, | |
| "loss": 0.4386, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "grad_norm": 0.8112042700957924, | |
| "learning_rate": 1.8671302545957628e-05, | |
| "loss": 0.4343, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 0.5019011406844106, | |
| "grad_norm": 0.7139927131242524, | |
| "learning_rate": 1.8654713862095272e-05, | |
| "loss": 0.4224, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.5038022813688213, | |
| "grad_norm": 0.7230370775329742, | |
| "learning_rate": 1.8638029729583524e-05, | |
| "loss": 0.4235, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.5057034220532319, | |
| "grad_norm": 0.7574771657169349, | |
| "learning_rate": 1.8621250332423603e-05, | |
| "loss": 0.4174, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.5076045627376425, | |
| "grad_norm": 0.752623774971912, | |
| "learning_rate": 1.860437585566736e-05, | |
| "loss": 0.4168, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 0.5095057034220533, | |
| "grad_norm": 0.7297710766866342, | |
| "learning_rate": 1.8587406485415226e-05, | |
| "loss": 0.4136, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.5114068441064639, | |
| "grad_norm": 0.8022213538204764, | |
| "learning_rate": 1.8570342408814173e-05, | |
| "loss": 0.4291, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 0.5133079847908745, | |
| "grad_norm": 0.7574910965054957, | |
| "learning_rate": 1.855318381405564e-05, | |
| "loss": 0.4381, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.5152091254752852, | |
| "grad_norm": 0.7681272936320716, | |
| "learning_rate": 1.8535930890373467e-05, | |
| "loss": 0.4304, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 0.5171102661596958, | |
| "grad_norm": 0.7808163872744192, | |
| "learning_rate": 1.8518583828041787e-05, | |
| "loss": 0.4194, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.5190114068441065, | |
| "grad_norm": 0.7479753075038985, | |
| "learning_rate": 1.8501142818372964e-05, | |
| "loss": 0.4136, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 0.5209125475285171, | |
| "grad_norm": 0.939272523451366, | |
| "learning_rate": 1.848360805371544e-05, | |
| "loss": 0.4334, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.5228136882129277, | |
| "grad_norm": 0.7409056940678842, | |
| "learning_rate": 1.8465979727451653e-05, | |
| "loss": 0.4208, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.5247148288973384, | |
| "grad_norm": 0.7542467808477448, | |
| "learning_rate": 1.8448258033995877e-05, | |
| "loss": 0.4198, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.526615969581749, | |
| "grad_norm": 0.8121028432796388, | |
| "learning_rate": 1.8430443168792087e-05, | |
| "loss": 0.4029, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 0.5285171102661597, | |
| "grad_norm": 0.7515292651802513, | |
| "learning_rate": 1.8412535328311813e-05, | |
| "loss": 0.4259, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 0.5304182509505704, | |
| "grad_norm": 0.8766278278695824, | |
| "learning_rate": 1.8394534710051956e-05, | |
| "loss": 0.4306, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 0.532319391634981, | |
| "grad_norm": 0.8534670881932418, | |
| "learning_rate": 1.8376441512532617e-05, | |
| "loss": 0.4215, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.5342205323193916, | |
| "grad_norm": 0.7560988615911489, | |
| "learning_rate": 1.835825593529492e-05, | |
| "loss": 0.4008, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 0.5361216730038023, | |
| "grad_norm": 0.9081003545746638, | |
| "learning_rate": 1.833997817889878e-05, | |
| "loss": 0.4205, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 0.5380228136882129, | |
| "grad_norm": 0.8715595059137391, | |
| "learning_rate": 1.8321608444920738e-05, | |
| "loss": 0.403, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 0.5399239543726235, | |
| "grad_norm": 0.8127215138595487, | |
| "learning_rate": 1.830314693595169e-05, | |
| "loss": 0.4082, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.5418250950570342, | |
| "grad_norm": 0.8310304791953129, | |
| "learning_rate": 1.828459385559468e-05, | |
| "loss": 0.4113, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.5437262357414449, | |
| "grad_norm": 0.7234759727808568, | |
| "learning_rate": 1.8265949408462657e-05, | |
| "loss": 0.4123, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 0.5456273764258555, | |
| "grad_norm": 0.7862452804413832, | |
| "learning_rate": 1.8247213800176192e-05, | |
| "loss": 0.4195, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 0.5475285171102662, | |
| "grad_norm": 0.7793367491171196, | |
| "learning_rate": 1.8228387237361245e-05, | |
| "loss": 0.4153, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.5494296577946768, | |
| "grad_norm": 0.7200026961081423, | |
| "learning_rate": 1.8209469927646863e-05, | |
| "loss": 0.4112, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 0.5513307984790875, | |
| "grad_norm": 0.8307757036015943, | |
| "learning_rate": 1.8190462079662897e-05, | |
| "loss": 0.4302, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.5532319391634981, | |
| "grad_norm": 0.7038864985014598, | |
| "learning_rate": 1.81713639030377e-05, | |
| "loss": 0.4155, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.5551330798479087, | |
| "grad_norm": 0.7437893838045984, | |
| "learning_rate": 1.8152175608395814e-05, | |
| "loss": 0.3968, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 0.5570342205323194, | |
| "grad_norm": 0.8670279536714978, | |
| "learning_rate": 1.8132897407355657e-05, | |
| "loss": 0.4236, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 0.55893536121673, | |
| "grad_norm": 0.8055188861797729, | |
| "learning_rate": 1.811352951252717e-05, | |
| "loss": 0.4003, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 0.5608365019011406, | |
| "grad_norm": 0.7564735293025392, | |
| "learning_rate": 1.809407213750949e-05, | |
| "loss": 0.415, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.5627376425855514, | |
| "grad_norm": 0.8519507433653463, | |
| "learning_rate": 1.807452549688859e-05, | |
| "loss": 0.4167, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 0.564638783269962, | |
| "grad_norm": 0.8738355419808232, | |
| "learning_rate": 1.8054889806234906e-05, | |
| "loss": 0.4454, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 0.5665399239543726, | |
| "grad_norm": 0.796305715231978, | |
| "learning_rate": 1.8035165282100963e-05, | |
| "loss": 0.4172, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 0.5684410646387833, | |
| "grad_norm": 0.8179163375507781, | |
| "learning_rate": 1.8015352142018984e-05, | |
| "loss": 0.4235, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 0.5703422053231939, | |
| "grad_norm": 0.7733016650765832, | |
| "learning_rate": 1.799545060449851e-05, | |
| "loss": 0.4099, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.5722433460076045, | |
| "grad_norm": 0.7918473550430336, | |
| "learning_rate": 1.797546088902396e-05, | |
| "loss": 0.4062, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 0.5741444866920152, | |
| "grad_norm": 0.7877029141652803, | |
| "learning_rate": 1.7955383216052224e-05, | |
| "loss": 0.4214, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 0.5760456273764258, | |
| "grad_norm": 0.7607784646298151, | |
| "learning_rate": 1.7935217807010238e-05, | |
| "loss": 0.4251, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 0.5779467680608364, | |
| "grad_norm": 0.8900682264655115, | |
| "learning_rate": 1.7914964884292543e-05, | |
| "loss": 0.4345, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 0.5798479087452472, | |
| "grad_norm": 0.7458783551974968, | |
| "learning_rate": 1.7894624671258813e-05, | |
| "loss": 0.4282, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.5817490494296578, | |
| "grad_norm": 0.8507127005672392, | |
| "learning_rate": 1.7874197392231414e-05, | |
| "loss": 0.4138, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 0.5836501901140685, | |
| "grad_norm": 0.7757679226354866, | |
| "learning_rate": 1.7853683272492913e-05, | |
| "loss": 0.4185, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 0.5855513307984791, | |
| "grad_norm": 0.6985735937620627, | |
| "learning_rate": 1.7833082538283615e-05, | |
| "loss": 0.4108, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 0.5874524714828897, | |
| "grad_norm": 0.7027160594245886, | |
| "learning_rate": 1.7812395416799034e-05, | |
| "loss": 0.3999, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 0.5893536121673004, | |
| "grad_norm": 0.895488177928045, | |
| "learning_rate": 1.7791622136187422e-05, | |
| "loss": 0.4385, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.591254752851711, | |
| "grad_norm": 0.7218198104546929, | |
| "learning_rate": 1.7770762925547235e-05, | |
| "loss": 0.4177, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 0.5931558935361216, | |
| "grad_norm": 0.98705635242737, | |
| "learning_rate": 1.7749818014924612e-05, | |
| "loss": 0.4156, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 0.5950570342205324, | |
| "grad_norm": 0.7049046478709482, | |
| "learning_rate": 1.7728787635310828e-05, | |
| "loss": 0.4008, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 0.596958174904943, | |
| "grad_norm": 0.8782913053327349, | |
| "learning_rate": 1.770767201863976e-05, | |
| "loss": 0.417, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.5988593155893536, | |
| "grad_norm": 0.8412945152771016, | |
| "learning_rate": 1.7686471397785322e-05, | |
| "loss": 0.4018, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.6007604562737643, | |
| "grad_norm": 0.7838505252477733, | |
| "learning_rate": 1.76651860065589e-05, | |
| "loss": 0.4091, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 0.6026615969581749, | |
| "grad_norm": 1.0434152104186716, | |
| "learning_rate": 1.764381607970677e-05, | |
| "loss": 0.42, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 0.6045627376425855, | |
| "grad_norm": 0.729980112821725, | |
| "learning_rate": 1.7622361852907506e-05, | |
| "loss": 0.4144, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 0.6064638783269962, | |
| "grad_norm": 0.979875549732311, | |
| "learning_rate": 1.760082356276939e-05, | |
| "loss": 0.4349, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 0.6083650190114068, | |
| "grad_norm": 0.779496296671776, | |
| "learning_rate": 1.75792014468278e-05, | |
| "loss": 0.4146, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.6102661596958175, | |
| "grad_norm": 0.7364283113479096, | |
| "learning_rate": 1.7557495743542586e-05, | |
| "loss": 0.4235, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 0.6121673003802282, | |
| "grad_norm": 0.8676045275894251, | |
| "learning_rate": 1.7535706692295436e-05, | |
| "loss": 0.4259, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 0.6140684410646388, | |
| "grad_norm": 0.7965226358797538, | |
| "learning_rate": 1.7513834533387256e-05, | |
| "loss": 0.4119, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 0.6159695817490495, | |
| "grad_norm": 0.7850902675847963, | |
| "learning_rate": 1.7491879508035488e-05, | |
| "loss": 0.4125, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 0.6178707224334601, | |
| "grad_norm": 0.8337711132775816, | |
| "learning_rate": 1.746984185837149e-05, | |
| "loss": 0.4012, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.6197718631178707, | |
| "grad_norm": 0.7262087791286067, | |
| "learning_rate": 1.744772182743782e-05, | |
| "loss": 0.4098, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 0.6216730038022814, | |
| "grad_norm": 0.7062522214493603, | |
| "learning_rate": 1.7425519659185596e-05, | |
| "loss": 0.4038, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 0.623574144486692, | |
| "grad_norm": 0.8823688575645542, | |
| "learning_rate": 1.740323559847179e-05, | |
| "loss": 0.4069, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 0.6254752851711026, | |
| "grad_norm": 0.7050399466138574, | |
| "learning_rate": 1.738086989105651e-05, | |
| "loss": 0.4023, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 0.6273764258555133, | |
| "grad_norm": 0.7988069925524706, | |
| "learning_rate": 1.735842278360032e-05, | |
| "loss": 0.4311, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.629277566539924, | |
| "grad_norm": 0.7761644948449151, | |
| "learning_rate": 1.73358945236615e-05, | |
| "loss": 0.4153, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 0.6311787072243346, | |
| "grad_norm": 0.6805322496469458, | |
| "learning_rate": 1.7313285359693322e-05, | |
| "loss": 0.4256, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 0.6330798479087453, | |
| "grad_norm": 0.7835770873118278, | |
| "learning_rate": 1.7290595541041312e-05, | |
| "loss": 0.3867, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 0.6349809885931559, | |
| "grad_norm": 0.7170788522182358, | |
| "learning_rate": 1.7267825317940494e-05, | |
| "loss": 0.4313, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 0.6368821292775665, | |
| "grad_norm": 0.7729388053926101, | |
| "learning_rate": 1.724497494151264e-05, | |
| "loss": 0.4106, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.6387832699619772, | |
| "grad_norm": 0.7351467180563832, | |
| "learning_rate": 1.7222044663763484e-05, | |
| "loss": 0.4084, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 0.6406844106463878, | |
| "grad_norm": 0.7378840439209664, | |
| "learning_rate": 1.7199034737579962e-05, | |
| "loss": 0.4049, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 0.6425855513307985, | |
| "grad_norm": 0.7099720614706242, | |
| "learning_rate": 1.7175945416727405e-05, | |
| "loss": 0.4098, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 0.6444866920152091, | |
| "grad_norm": 0.7382021523891706, | |
| "learning_rate": 1.7152776955846768e-05, | |
| "loss": 0.4065, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 0.6463878326996197, | |
| "grad_norm": 0.7228844460949072, | |
| "learning_rate": 1.7129529610451775e-05, | |
| "loss": 0.3982, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.6482889733840305, | |
| "grad_norm": 0.7282870522921897, | |
| "learning_rate": 1.7106203636926154e-05, | |
| "loss": 0.4186, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 0.6501901140684411, | |
| "grad_norm": 0.6964949991080847, | |
| "learning_rate": 1.7082799292520767e-05, | |
| "loss": 0.4066, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 0.6520912547528517, | |
| "grad_norm": 0.8185154910901153, | |
| "learning_rate": 1.7059316835350806e-05, | |
| "loss": 0.4182, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 0.6539923954372624, | |
| "grad_norm": 0.6677504707396503, | |
| "learning_rate": 1.7035756524392924e-05, | |
| "loss": 0.4154, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 0.655893536121673, | |
| "grad_norm": 0.7063517030224836, | |
| "learning_rate": 1.7012118619482376e-05, | |
| "loss": 0.4071, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.6577946768060836, | |
| "grad_norm": 0.7319455070050461, | |
| "learning_rate": 1.6988403381310177e-05, | |
| "loss": 0.428, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 0.6596958174904943, | |
| "grad_norm": 0.722338581101519, | |
| "learning_rate": 1.696461107142021e-05, | |
| "loss": 0.4158, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 0.6615969581749049, | |
| "grad_norm": 0.7474661510535513, | |
| "learning_rate": 1.6940741952206342e-05, | |
| "loss": 0.3905, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 0.6634980988593155, | |
| "grad_norm": 0.7145738707702087, | |
| "learning_rate": 1.691679628690953e-05, | |
| "loss": 0.4231, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 0.6653992395437263, | |
| "grad_norm": 0.6904476996845786, | |
| "learning_rate": 1.6892774339614927e-05, | |
| "loss": 0.4001, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.6673003802281369, | |
| "grad_norm": 0.7552196032601604, | |
| "learning_rate": 1.686867637524896e-05, | |
| "loss": 0.43, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 0.6692015209125475, | |
| "grad_norm": 0.7111756763700475, | |
| "learning_rate": 1.6844502659576414e-05, | |
| "loss": 0.4169, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 0.6711026615969582, | |
| "grad_norm": 0.7571346775795089, | |
| "learning_rate": 1.6820253459197493e-05, | |
| "loss": 0.4223, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 0.6730038022813688, | |
| "grad_norm": 0.7744954077691227, | |
| "learning_rate": 1.679592904154489e-05, | |
| "loss": 0.4317, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 0.6749049429657795, | |
| "grad_norm": 0.6358230093187863, | |
| "learning_rate": 1.677152967488084e-05, | |
| "loss": 0.4119, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.6768060836501901, | |
| "grad_norm": 0.8270393346418727, | |
| "learning_rate": 1.6747055628294134e-05, | |
| "loss": 0.4182, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 0.6787072243346007, | |
| "grad_norm": 0.7087991924446325, | |
| "learning_rate": 1.6722507171697184e-05, | |
| "loss": 0.419, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 0.6806083650190115, | |
| "grad_norm": 0.7567644550900913, | |
| "learning_rate": 1.669788457582304e-05, | |
| "loss": 0.4146, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.6825095057034221, | |
| "grad_norm": 0.787249130560551, | |
| "learning_rate": 1.6673188112222394e-05, | |
| "loss": 0.4061, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 0.6844106463878327, | |
| "grad_norm": 0.8048502887589264, | |
| "learning_rate": 1.6648418053260585e-05, | |
| "loss": 0.4094, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.6863117870722434, | |
| "grad_norm": 0.8512563959751673, | |
| "learning_rate": 1.6623574672114596e-05, | |
| "loss": 0.4095, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 0.688212927756654, | |
| "grad_norm": 0.6930948411384824, | |
| "learning_rate": 1.6598658242770054e-05, | |
| "loss": 0.4156, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 0.6901140684410646, | |
| "grad_norm": 0.8999684760249871, | |
| "learning_rate": 1.6573669040018202e-05, | |
| "loss": 0.3961, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 0.6920152091254753, | |
| "grad_norm": 0.7727612999636725, | |
| "learning_rate": 1.6548607339452853e-05, | |
| "loss": 0.3942, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 0.6939163498098859, | |
| "grad_norm": 0.6900415781775979, | |
| "learning_rate": 1.652347341746737e-05, | |
| "loss": 0.4125, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.6958174904942965, | |
| "grad_norm": 0.8462656663622575, | |
| "learning_rate": 1.6498267551251618e-05, | |
| "loss": 0.4061, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 0.6977186311787072, | |
| "grad_norm": 0.6487088081153829, | |
| "learning_rate": 1.6472990018788884e-05, | |
| "loss": 0.386, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 0.6996197718631179, | |
| "grad_norm": 0.7244559831015, | |
| "learning_rate": 1.644764109885284e-05, | |
| "loss": 0.4196, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 0.7015209125475285, | |
| "grad_norm": 0.8232611823772114, | |
| "learning_rate": 1.642222107100446e-05, | |
| "loss": 0.4178, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 0.7034220532319392, | |
| "grad_norm": 0.7336858618815859, | |
| "learning_rate": 1.6396730215588913e-05, | |
| "loss": 0.406, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.7053231939163498, | |
| "grad_norm": 0.735460644566425, | |
| "learning_rate": 1.6371168813732514e-05, | |
| "loss": 0.4085, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 0.7072243346007605, | |
| "grad_norm": 0.8461775114052308, | |
| "learning_rate": 1.6345537147339578e-05, | |
| "loss": 0.4074, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 0.7091254752851711, | |
| "grad_norm": 0.6982265813743247, | |
| "learning_rate": 1.6319835499089358e-05, | |
| "loss": 0.3952, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 0.7110266159695817, | |
| "grad_norm": 0.7823104808162714, | |
| "learning_rate": 1.6294064152432878e-05, | |
| "loss": 0.4073, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 0.7129277566539924, | |
| "grad_norm": 0.7412753900313068, | |
| "learning_rate": 1.626822339158985e-05, | |
| "loss": 0.3946, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.714828897338403, | |
| "grad_norm": 0.7029518292817529, | |
| "learning_rate": 1.6242313501545522e-05, | |
| "loss": 0.4017, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 0.7167300380228137, | |
| "grad_norm": 0.7547571332542912, | |
| "learning_rate": 1.621633476804752e-05, | |
| "loss": 0.4058, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 0.7186311787072244, | |
| "grad_norm": 0.85265016552861, | |
| "learning_rate": 1.6190287477602716e-05, | |
| "loss": 0.4161, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 0.720532319391635, | |
| "grad_norm": 0.7504785035494755, | |
| "learning_rate": 1.6164171917474078e-05, | |
| "loss": 0.3985, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 0.7224334600760456, | |
| "grad_norm": 0.7289487156072273, | |
| "learning_rate": 1.6137988375677466e-05, | |
| "loss": 0.4264, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.7243346007604563, | |
| "grad_norm": 0.7073771117569176, | |
| "learning_rate": 1.6111737140978495e-05, | |
| "loss": 0.4033, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 0.7262357414448669, | |
| "grad_norm": 0.7539660374802845, | |
| "learning_rate": 1.6085418502889315e-05, | |
| "loss": 0.4107, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 0.7281368821292775, | |
| "grad_norm": 0.7578534341409431, | |
| "learning_rate": 1.6059032751665454e-05, | |
| "loss": 0.4282, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 0.7300380228136882, | |
| "grad_norm": 0.7409738342990565, | |
| "learning_rate": 1.6032580178302585e-05, | |
| "loss": 0.4235, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 0.7319391634980988, | |
| "grad_norm": 0.709848332589726, | |
| "learning_rate": 1.600606107453333e-05, | |
| "loss": 0.4054, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.7338403041825095, | |
| "grad_norm": 0.7306708012000779, | |
| "learning_rate": 1.597947573282405e-05, | |
| "loss": 0.4033, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 0.7357414448669202, | |
| "grad_norm": 0.6755683118547308, | |
| "learning_rate": 1.5952824446371608e-05, | |
| "loss": 0.4127, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 0.7376425855513308, | |
| "grad_norm": 0.7639408081032738, | |
| "learning_rate": 1.592610750910014e-05, | |
| "loss": 0.4082, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 0.7395437262357415, | |
| "grad_norm": 0.6807687739345376, | |
| "learning_rate": 1.589932521565781e-05, | |
| "loss": 0.3956, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 0.7414448669201521, | |
| "grad_norm": 0.7719886410900012, | |
| "learning_rate": 1.587247786141358e-05, | |
| "loss": 0.4113, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.7433460076045627, | |
| "grad_norm": 0.7375110239072027, | |
| "learning_rate": 1.5845565742453906e-05, | |
| "loss": 0.431, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 0.7452471482889734, | |
| "grad_norm": 0.7287830807140766, | |
| "learning_rate": 1.581858915557953e-05, | |
| "loss": 0.4025, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 0.747148288973384, | |
| "grad_norm": 0.726812873095342, | |
| "learning_rate": 1.5791548398302167e-05, | |
| "loss": 0.4185, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 0.7490494296577946, | |
| "grad_norm": 0.6748570741849, | |
| "learning_rate": 1.5764443768841234e-05, | |
| "loss": 0.3922, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 0.7509505703422054, | |
| "grad_norm": 0.7118608253031248, | |
| "learning_rate": 1.5737275566120577e-05, | |
| "loss": 0.4181, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.752851711026616, | |
| "grad_norm": 0.670282276514105, | |
| "learning_rate": 1.5710044089765144e-05, | |
| "loss": 0.4162, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 0.7547528517110266, | |
| "grad_norm": 0.7440610483438017, | |
| "learning_rate": 1.5682749640097708e-05, | |
| "loss": 0.4038, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 0.7566539923954373, | |
| "grad_norm": 0.6748042604583505, | |
| "learning_rate": 1.565539251813554e-05, | |
| "loss": 0.385, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 0.7585551330798479, | |
| "grad_norm": 0.6598889673850782, | |
| "learning_rate": 1.5627973025587093e-05, | |
| "loss": 0.4137, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 0.7604562737642585, | |
| "grad_norm": 0.7175541625661477, | |
| "learning_rate": 1.560049146484868e-05, | |
| "loss": 0.4097, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.7623574144486692, | |
| "grad_norm": 0.7162069852321954, | |
| "learning_rate": 1.5572948139001128e-05, | |
| "loss": 0.4128, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 0.7642585551330798, | |
| "grad_norm": 0.6393324198770911, | |
| "learning_rate": 1.5545343351806443e-05, | |
| "loss": 0.3991, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 0.7661596958174905, | |
| "grad_norm": 0.8374538646059091, | |
| "learning_rate": 1.551767740770446e-05, | |
| "loss": 0.3966, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 0.7680608365019012, | |
| "grad_norm": 0.6916479778826591, | |
| "learning_rate": 1.5489950611809484e-05, | |
| "loss": 0.3962, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 0.7699619771863118, | |
| "grad_norm": 0.6944095412016408, | |
| "learning_rate": 1.5462163269906928e-05, | |
| "loss": 0.4114, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.7718631178707225, | |
| "grad_norm": 0.8108854241694067, | |
| "learning_rate": 1.5434315688449924e-05, | |
| "loss": 0.4133, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 0.7737642585551331, | |
| "grad_norm": 0.6617789200023629, | |
| "learning_rate": 1.5406408174555978e-05, | |
| "loss": 0.3883, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 0.7756653992395437, | |
| "grad_norm": 0.7536545765286243, | |
| "learning_rate": 1.5378441036003543e-05, | |
| "loss": 0.4499, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 0.7775665399239544, | |
| "grad_norm": 0.7781429005601748, | |
| "learning_rate": 1.535041458122865e-05, | |
| "loss": 0.3983, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 0.779467680608365, | |
| "grad_norm": 0.6885168368258409, | |
| "learning_rate": 1.5322329119321508e-05, | |
| "loss": 0.3972, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.7813688212927756, | |
| "grad_norm": 0.6847222592088383, | |
| "learning_rate": 1.529418496002308e-05, | |
| "loss": 0.4091, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 0.7832699619771863, | |
| "grad_norm": 0.7272406413887, | |
| "learning_rate": 1.5265982413721662e-05, | |
| "loss": 0.3824, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 0.785171102661597, | |
| "grad_norm": 0.7197032035973552, | |
| "learning_rate": 1.5237721791449497e-05, | |
| "loss": 0.4144, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 0.7870722433460076, | |
| "grad_norm": 0.7419100918228936, | |
| "learning_rate": 1.5209403404879305e-05, | |
| "loss": 0.401, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 0.7889733840304183, | |
| "grad_norm": 0.7308464817309022, | |
| "learning_rate": 1.5181027566320858e-05, | |
| "loss": 0.3982, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.7908745247148289, | |
| "grad_norm": 0.7338235939187204, | |
| "learning_rate": 1.5152594588717544e-05, | |
| "loss": 0.4052, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 0.7927756653992395, | |
| "grad_norm": 0.7800234778242071, | |
| "learning_rate": 1.5124104785642909e-05, | |
| "loss": 0.4129, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 0.7946768060836502, | |
| "grad_norm": 0.8050956171248939, | |
| "learning_rate": 1.5095558471297196e-05, | |
| "loss": 0.4025, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 0.7965779467680608, | |
| "grad_norm": 0.6855083217672223, | |
| "learning_rate": 1.5066955960503893e-05, | |
| "loss": 0.3913, | |
| "step": 419 | |
| }, | |
| { | |
| "epoch": 0.7984790874524715, | |
| "grad_norm": 0.8479122296697837, | |
| "learning_rate": 1.5038297568706244e-05, | |
| "loss": 0.3978, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.8003802281368821, | |
| "grad_norm": 0.848678170533286, | |
| "learning_rate": 1.5009583611963772e-05, | |
| "loss": 0.3984, | |
| "step": 421 | |
| }, | |
| { | |
| "epoch": 0.8022813688212928, | |
| "grad_norm": 0.7316858291023512, | |
| "learning_rate": 1.4980814406948806e-05, | |
| "loss": 0.4, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 0.8041825095057035, | |
| "grad_norm": 0.7030752551809762, | |
| "learning_rate": 1.4951990270942991e-05, | |
| "loss": 0.3936, | |
| "step": 423 | |
| }, | |
| { | |
| "epoch": 0.8060836501901141, | |
| "grad_norm": 0.7288865749792287, | |
| "learning_rate": 1.492311152183376e-05, | |
| "loss": 0.3801, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 0.8079847908745247, | |
| "grad_norm": 0.6851009218346807, | |
| "learning_rate": 1.4894178478110856e-05, | |
| "loss": 0.4106, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.8098859315589354, | |
| "grad_norm": 0.7437318632236917, | |
| "learning_rate": 1.4865191458862816e-05, | |
| "loss": 0.3847, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 0.811787072243346, | |
| "grad_norm": 0.6998949415558212, | |
| "learning_rate": 1.4836150783773442e-05, | |
| "loss": 0.3973, | |
| "step": 427 | |
| }, | |
| { | |
| "epoch": 0.8136882129277566, | |
| "grad_norm": 0.7074183705074062, | |
| "learning_rate": 1.4807056773118276e-05, | |
| "loss": 0.4259, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 0.8155893536121673, | |
| "grad_norm": 0.6566427231037307, | |
| "learning_rate": 1.4777909747761085e-05, | |
| "loss": 0.3713, | |
| "step": 429 | |
| }, | |
| { | |
| "epoch": 0.8174904942965779, | |
| "grad_norm": 0.7267684127187504, | |
| "learning_rate": 1.4748710029150296e-05, | |
| "loss": 0.4219, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.8193916349809885, | |
| "grad_norm": 0.6997797943607886, | |
| "learning_rate": 1.4719457939315468e-05, | |
| "loss": 0.4019, | |
| "step": 431 | |
| }, | |
| { | |
| "epoch": 0.8212927756653993, | |
| "grad_norm": 0.6777517270062602, | |
| "learning_rate": 1.4690153800863743e-05, | |
| "loss": 0.396, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 0.8231939163498099, | |
| "grad_norm": 0.7079807349494212, | |
| "learning_rate": 1.4660797936976278e-05, | |
| "loss": 0.3938, | |
| "step": 433 | |
| }, | |
| { | |
| "epoch": 0.8250950570342205, | |
| "grad_norm": 0.67605432336797, | |
| "learning_rate": 1.4631390671404682e-05, | |
| "loss": 0.3901, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 0.8269961977186312, | |
| "grad_norm": 0.6869663607923845, | |
| "learning_rate": 1.460193232846745e-05, | |
| "loss": 0.394, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.8288973384030418, | |
| "grad_norm": 0.7459830376378118, | |
| "learning_rate": 1.4572423233046386e-05, | |
| "loss": 0.4258, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 0.8307984790874525, | |
| "grad_norm": 0.6929292646826665, | |
| "learning_rate": 1.4542863710583022e-05, | |
| "loss": 0.3963, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 0.8326996197718631, | |
| "grad_norm": 0.6669410564430523, | |
| "learning_rate": 1.4513254087075015e-05, | |
| "loss": 0.404, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 0.8346007604562737, | |
| "grad_norm": 0.6433805267698699, | |
| "learning_rate": 1.4483594689072571e-05, | |
| "loss": 0.4029, | |
| "step": 439 | |
| }, | |
| { | |
| "epoch": 0.8365019011406845, | |
| "grad_norm": 0.7679120641620589, | |
| "learning_rate": 1.4453885843674837e-05, | |
| "loss": 0.3861, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.8384030418250951, | |
| "grad_norm": 0.6584529402067636, | |
| "learning_rate": 1.4424127878526278e-05, | |
| "loss": 0.3965, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 0.8403041825095057, | |
| "grad_norm": 0.6646344005085827, | |
| "learning_rate": 1.4394321121813093e-05, | |
| "loss": 0.403, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 0.8422053231939164, | |
| "grad_norm": 0.7738523501024654, | |
| "learning_rate": 1.436446590225957e-05, | |
| "loss": 0.4045, | |
| "step": 443 | |
| }, | |
| { | |
| "epoch": 0.844106463878327, | |
| "grad_norm": 0.6368791236149156, | |
| "learning_rate": 1.433456254912447e-05, | |
| "loss": 0.386, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 0.8460076045627376, | |
| "grad_norm": 0.7567894664805073, | |
| "learning_rate": 1.4304611392197399e-05, | |
| "loss": 0.3965, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.8479087452471483, | |
| "grad_norm": 0.7489215480662238, | |
| "learning_rate": 1.427461276179517e-05, | |
| "loss": 0.3981, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 0.8498098859315589, | |
| "grad_norm": 0.6740459651875343, | |
| "learning_rate": 1.4244566988758152e-05, | |
| "loss": 0.3925, | |
| "step": 447 | |
| }, | |
| { | |
| "epoch": 0.8517110266159695, | |
| "grad_norm": 0.8336979821371919, | |
| "learning_rate": 1.4214474404446633e-05, | |
| "loss": 0.399, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 0.8536121673003803, | |
| "grad_norm": 0.6929229974360013, | |
| "learning_rate": 1.4184335340737158e-05, | |
| "loss": 0.4008, | |
| "step": 449 | |
| }, | |
| { | |
| "epoch": 0.8555133079847909, | |
| "grad_norm": 0.6767557276485131, | |
| "learning_rate": 1.4154150130018867e-05, | |
| "loss": 0.3973, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.8574144486692015, | |
| "grad_norm": 0.7505386738174119, | |
| "learning_rate": 1.4123919105189836e-05, | |
| "loss": 0.4147, | |
| "step": 451 | |
| }, | |
| { | |
| "epoch": 0.8593155893536122, | |
| "grad_norm": 0.6859440702306527, | |
| "learning_rate": 1.4093642599653406e-05, | |
| "loss": 0.3919, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 0.8612167300380228, | |
| "grad_norm": 0.7033034156710803, | |
| "learning_rate": 1.40633209473145e-05, | |
| "loss": 0.4216, | |
| "step": 453 | |
| }, | |
| { | |
| "epoch": 0.8631178707224335, | |
| "grad_norm": 0.6821295760561467, | |
| "learning_rate": 1.4032954482575938e-05, | |
| "loss": 0.4089, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 0.8650190114068441, | |
| "grad_norm": 0.6683423377717309, | |
| "learning_rate": 1.4002543540334766e-05, | |
| "loss": 0.4094, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.8669201520912547, | |
| "grad_norm": 0.6903220567860381, | |
| "learning_rate": 1.3972088455978537e-05, | |
| "loss": 0.3936, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 0.8688212927756654, | |
| "grad_norm": 0.6907013390792703, | |
| "learning_rate": 1.3941589565381635e-05, | |
| "loss": 0.4001, | |
| "step": 457 | |
| }, | |
| { | |
| "epoch": 0.870722433460076, | |
| "grad_norm": 0.6839655626974248, | |
| "learning_rate": 1.391104720490156e-05, | |
| "loss": 0.4111, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 0.8726235741444867, | |
| "grad_norm": 0.6544263712253102, | |
| "learning_rate": 1.3880461711375224e-05, | |
| "loss": 0.3824, | |
| "step": 459 | |
| }, | |
| { | |
| "epoch": 0.8745247148288974, | |
| "grad_norm": 0.7142550347734252, | |
| "learning_rate": 1.3849833422115221e-05, | |
| "loss": 0.4116, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.876425855513308, | |
| "grad_norm": 0.6822476124154475, | |
| "learning_rate": 1.3819162674906134e-05, | |
| "loss": 0.3997, | |
| "step": 461 | |
| }, | |
| { | |
| "epoch": 0.8783269961977186, | |
| "grad_norm": 0.6760600385243434, | |
| "learning_rate": 1.378844980800078e-05, | |
| "loss": 0.3897, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 0.8802281368821293, | |
| "grad_norm": 0.6642030173159551, | |
| "learning_rate": 1.3757695160116502e-05, | |
| "loss": 0.409, | |
| "step": 463 | |
| }, | |
| { | |
| "epoch": 0.8821292775665399, | |
| "grad_norm": 0.6054639465794597, | |
| "learning_rate": 1.3726899070431423e-05, | |
| "loss": 0.3831, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 0.8840304182509505, | |
| "grad_norm": 0.7421063979070679, | |
| "learning_rate": 1.3696061878580707e-05, | |
| "loss": 0.4092, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.8859315589353612, | |
| "grad_norm": 0.6525677249013153, | |
| "learning_rate": 1.3665183924652817e-05, | |
| "loss": 0.4054, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 0.8878326996197718, | |
| "grad_norm": 0.7200090507001417, | |
| "learning_rate": 1.3634265549185755e-05, | |
| "loss": 0.3771, | |
| "step": 467 | |
| }, | |
| { | |
| "epoch": 0.8897338403041825, | |
| "grad_norm": 0.722655757477792, | |
| "learning_rate": 1.3603307093163319e-05, | |
| "loss": 0.4109, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 0.8916349809885932, | |
| "grad_norm": 0.8860974580937121, | |
| "learning_rate": 1.3572308898011328e-05, | |
| "loss": 0.4036, | |
| "step": 469 | |
| }, | |
| { | |
| "epoch": 0.8935361216730038, | |
| "grad_norm": 0.6741501986606298, | |
| "learning_rate": 1.3541271305593878e-05, | |
| "loss": 0.4019, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.8954372623574145, | |
| "grad_norm": 0.6544639047853891, | |
| "learning_rate": 1.3510194658209547e-05, | |
| "loss": 0.39, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 0.8973384030418251, | |
| "grad_norm": 0.6590564260402584, | |
| "learning_rate": 1.3479079298587634e-05, | |
| "loss": 0.3948, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 0.8992395437262357, | |
| "grad_norm": 0.6719760867784107, | |
| "learning_rate": 1.3447925569884374e-05, | |
| "loss": 0.4023, | |
| "step": 473 | |
| }, | |
| { | |
| "epoch": 0.9011406844106464, | |
| "grad_norm": 0.6283204245473711, | |
| "learning_rate": 1.3416733815679166e-05, | |
| "loss": 0.3958, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 0.903041825095057, | |
| "grad_norm": 0.6782863521959651, | |
| "learning_rate": 1.3385504379970764e-05, | |
| "loss": 0.4101, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.9049429657794676, | |
| "grad_norm": 0.6390247264182968, | |
| "learning_rate": 1.3354237607173494e-05, | |
| "loss": 0.3883, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 0.9068441064638784, | |
| "grad_norm": 0.6857592286932283, | |
| "learning_rate": 1.3322933842113457e-05, | |
| "loss": 0.4209, | |
| "step": 477 | |
| }, | |
| { | |
| "epoch": 0.908745247148289, | |
| "grad_norm": 0.6556807618245046, | |
| "learning_rate": 1.3291593430024727e-05, | |
| "loss": 0.3851, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 0.9106463878326996, | |
| "grad_norm": 0.6758286632475549, | |
| "learning_rate": 1.3260216716545534e-05, | |
| "loss": 0.4132, | |
| "step": 479 | |
| }, | |
| { | |
| "epoch": 0.9125475285171103, | |
| "grad_norm": 0.7327310692942914, | |
| "learning_rate": 1.3228804047714462e-05, | |
| "loss": 0.3882, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.9144486692015209, | |
| "grad_norm": 0.6642529628936523, | |
| "learning_rate": 1.319735576996663e-05, | |
| "loss": 0.4103, | |
| "step": 481 | |
| }, | |
| { | |
| "epoch": 0.9163498098859315, | |
| "grad_norm": 0.747909451177315, | |
| "learning_rate": 1.3165872230129869e-05, | |
| "loss": 0.4017, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 0.9182509505703422, | |
| "grad_norm": 0.7076227227993218, | |
| "learning_rate": 1.3134353775420895e-05, | |
| "loss": 0.3755, | |
| "step": 483 | |
| }, | |
| { | |
| "epoch": 0.9201520912547528, | |
| "grad_norm": 0.6053744954998052, | |
| "learning_rate": 1.3102800753441488e-05, | |
| "loss": 0.3961, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 0.9220532319391636, | |
| "grad_norm": 0.7038504395047525, | |
| "learning_rate": 1.3071213512174655e-05, | |
| "loss": 0.3731, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.9239543726235742, | |
| "grad_norm": 0.7952573572199146, | |
| "learning_rate": 1.3039592399980785e-05, | |
| "loss": 0.3803, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 0.9258555133079848, | |
| "grad_norm": 0.6661383993167328, | |
| "learning_rate": 1.3007937765593818e-05, | |
| "loss": 0.4036, | |
| "step": 487 | |
| }, | |
| { | |
| "epoch": 0.9277566539923955, | |
| "grad_norm": 0.7308645581031047, | |
| "learning_rate": 1.2976249958117395e-05, | |
| "loss": 0.4165, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 0.9296577946768061, | |
| "grad_norm": 0.8436231027138845, | |
| "learning_rate": 1.2944529327021002e-05, | |
| "loss": 0.396, | |
| "step": 489 | |
| }, | |
| { | |
| "epoch": 0.9315589353612167, | |
| "grad_norm": 0.6945522700032737, | |
| "learning_rate": 1.291277622213612e-05, | |
| "loss": 0.373, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.9334600760456274, | |
| "grad_norm": 0.8079168407857312, | |
| "learning_rate": 1.2880990993652379e-05, | |
| "loss": 0.3926, | |
| "step": 491 | |
| }, | |
| { | |
| "epoch": 0.935361216730038, | |
| "grad_norm": 0.7189701971602107, | |
| "learning_rate": 1.2849173992113669e-05, | |
| "loss": 0.3859, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 0.9372623574144486, | |
| "grad_norm": 0.6377694908819881, | |
| "learning_rate": 1.2817325568414299e-05, | |
| "loss": 0.3951, | |
| "step": 493 | |
| }, | |
| { | |
| "epoch": 0.9391634980988594, | |
| "grad_norm": 0.6927268080914422, | |
| "learning_rate": 1.2785446073795118e-05, | |
| "loss": 0.4011, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 0.94106463878327, | |
| "grad_norm": 0.7610325721389116, | |
| "learning_rate": 1.2753535859839638e-05, | |
| "loss": 0.4104, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.9429657794676806, | |
| "grad_norm": 0.6385274729333882, | |
| "learning_rate": 1.272159527847016e-05, | |
| "loss": 0.377, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 0.9448669201520913, | |
| "grad_norm": 0.6980938289198161, | |
| "learning_rate": 1.2689624681943897e-05, | |
| "loss": 0.4092, | |
| "step": 497 | |
| }, | |
| { | |
| "epoch": 0.9467680608365019, | |
| "grad_norm": 0.7643504112408278, | |
| "learning_rate": 1.2657624422849077e-05, | |
| "loss": 0.4111, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 0.9486692015209125, | |
| "grad_norm": 0.6780464226831658, | |
| "learning_rate": 1.2625594854101066e-05, | |
| "loss": 0.4073, | |
| "step": 499 | |
| }, | |
| { | |
| "epoch": 0.9505703422053232, | |
| "grad_norm": 0.7258945591192271, | |
| "learning_rate": 1.2593536328938471e-05, | |
| "loss": 0.3964, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.9524714828897338, | |
| "grad_norm": 0.7444601713007722, | |
| "learning_rate": 1.2561449200919253e-05, | |
| "loss": 0.3856, | |
| "step": 501 | |
| }, | |
| { | |
| "epoch": 0.9543726235741445, | |
| "grad_norm": 0.6375923970468063, | |
| "learning_rate": 1.2529333823916807e-05, | |
| "loss": 0.3812, | |
| "step": 502 | |
| }, | |
| { | |
| "epoch": 0.9562737642585551, | |
| "grad_norm": 0.6672359020474062, | |
| "learning_rate": 1.2497190552116082e-05, | |
| "loss": 0.4002, | |
| "step": 503 | |
| }, | |
| { | |
| "epoch": 0.9581749049429658, | |
| "grad_norm": 0.6619667152371748, | |
| "learning_rate": 1.2465019740009662e-05, | |
| "loss": 0.3875, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 0.9600760456273765, | |
| "grad_norm": 0.7223282379834804, | |
| "learning_rate": 1.2432821742393854e-05, | |
| "loss": 0.4086, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 0.9619771863117871, | |
| "grad_norm": 0.6658973107419177, | |
| "learning_rate": 1.2400596914364792e-05, | |
| "loss": 0.3907, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 0.9638783269961977, | |
| "grad_norm": 0.716121145062994, | |
| "learning_rate": 1.2368345611314508e-05, | |
| "loss": 0.3995, | |
| "step": 507 | |
| }, | |
| { | |
| "epoch": 0.9657794676806084, | |
| "grad_norm": 0.6112196272103922, | |
| "learning_rate": 1.2336068188927002e-05, | |
| "loss": 0.3657, | |
| "step": 508 | |
| }, | |
| { | |
| "epoch": 0.967680608365019, | |
| "grad_norm": 0.7226651795861709, | |
| "learning_rate": 1.2303765003174342e-05, | |
| "loss": 0.3921, | |
| "step": 509 | |
| }, | |
| { | |
| "epoch": 0.9695817490494296, | |
| "grad_norm": 0.6807883764264409, | |
| "learning_rate": 1.2271436410312727e-05, | |
| "loss": 0.4125, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.9714828897338403, | |
| "grad_norm": 0.7191570289015976, | |
| "learning_rate": 1.2239082766878557e-05, | |
| "loss": 0.4054, | |
| "step": 511 | |
| }, | |
| { | |
| "epoch": 0.973384030418251, | |
| "grad_norm": 0.7485232192165344, | |
| "learning_rate": 1.2206704429684504e-05, | |
| "loss": 0.3883, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 0.9752851711026616, | |
| "grad_norm": 0.6773083656419889, | |
| "learning_rate": 1.2174301755815572e-05, | |
| "loss": 0.403, | |
| "step": 513 | |
| }, | |
| { | |
| "epoch": 0.9771863117870723, | |
| "grad_norm": 0.6473072014675867, | |
| "learning_rate": 1.2141875102625166e-05, | |
| "loss": 0.3881, | |
| "step": 514 | |
| }, | |
| { | |
| "epoch": 0.9790874524714829, | |
| "grad_norm": 0.6664083412498389, | |
| "learning_rate": 1.2109424827731144e-05, | |
| "loss": 0.4091, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 0.9809885931558935, | |
| "grad_norm": 0.6477691026325739, | |
| "learning_rate": 1.2076951289011884e-05, | |
| "loss": 0.411, | |
| "step": 516 | |
| }, | |
| { | |
| "epoch": 0.9828897338403042, | |
| "grad_norm": 0.6610633008691374, | |
| "learning_rate": 1.204445484460232e-05, | |
| "loss": 0.3944, | |
| "step": 517 | |
| }, | |
| { | |
| "epoch": 0.9847908745247148, | |
| "grad_norm": 0.6649513930736037, | |
| "learning_rate": 1.2011935852890004e-05, | |
| "loss": 0.4042, | |
| "step": 518 | |
| }, | |
| { | |
| "epoch": 0.9866920152091255, | |
| "grad_norm": 0.6483409260137976, | |
| "learning_rate": 1.1979394672511156e-05, | |
| "loss": 0.372, | |
| "step": 519 | |
| }, | |
| { | |
| "epoch": 0.9885931558935361, | |
| "grad_norm": 0.6775340967183738, | |
| "learning_rate": 1.19468316623467e-05, | |
| "loss": 0.3948, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.9904942965779467, | |
| "grad_norm": 0.6829131001974992, | |
| "learning_rate": 1.1914247181518312e-05, | |
| "loss": 0.3909, | |
| "step": 521 | |
| }, | |
| { | |
| "epoch": 0.9923954372623575, | |
| "grad_norm": 0.6567660137946015, | |
| "learning_rate": 1.1881641589384456e-05, | |
| "loss": 0.3933, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 0.9942965779467681, | |
| "grad_norm": 0.6556715705566348, | |
| "learning_rate": 1.1849015245536424e-05, | |
| "loss": 0.3936, | |
| "step": 523 | |
| }, | |
| { | |
| "epoch": 0.9961977186311787, | |
| "grad_norm": 0.7580144697416081, | |
| "learning_rate": 1.1816368509794365e-05, | |
| "loss": 0.4012, | |
| "step": 524 | |
| }, | |
| { | |
| "epoch": 0.9980988593155894, | |
| "grad_norm": 0.6248428018449151, | |
| "learning_rate": 1.1783701742203326e-05, | |
| "loss": 0.3828, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 0.6825783060470704, | |
| "learning_rate": 1.1751015303029272e-05, | |
| "loss": 0.379, | |
| "step": 526 | |
| }, | |
| { | |
| "epoch": 1.0019011406844107, | |
| "grad_norm": 0.8142039115220996, | |
| "learning_rate": 1.1718309552755118e-05, | |
| "loss": 0.3051, | |
| "step": 527 | |
| }, | |
| { | |
| "epoch": 1.0038022813688212, | |
| "grad_norm": 0.7052627278804361, | |
| "learning_rate": 1.1685584852076746e-05, | |
| "loss": 0.3113, | |
| "step": 528 | |
| }, | |
| { | |
| "epoch": 1.005703422053232, | |
| "grad_norm": 0.6853584158841833, | |
| "learning_rate": 1.1652841561899042e-05, | |
| "loss": 0.2999, | |
| "step": 529 | |
| }, | |
| { | |
| "epoch": 1.0076045627376427, | |
| "grad_norm": 0.74623323857354, | |
| "learning_rate": 1.1620080043331901e-05, | |
| "loss": 0.3008, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 1.0095057034220531, | |
| "grad_norm": 0.7142932186373186, | |
| "learning_rate": 1.1587300657686254e-05, | |
| "loss": 0.3026, | |
| "step": 531 | |
| }, | |
| { | |
| "epoch": 1.0114068441064639, | |
| "grad_norm": 0.7889467800665719, | |
| "learning_rate": 1.1554503766470069e-05, | |
| "loss": 0.3026, | |
| "step": 532 | |
| }, | |
| { | |
| "epoch": 1.0133079847908746, | |
| "grad_norm": 0.820310870092898, | |
| "learning_rate": 1.1521689731384391e-05, | |
| "loss": 0.3035, | |
| "step": 533 | |
| }, | |
| { | |
| "epoch": 1.015209125475285, | |
| "grad_norm": 0.7895563861841615, | |
| "learning_rate": 1.1488858914319321e-05, | |
| "loss": 0.2988, | |
| "step": 534 | |
| }, | |
| { | |
| "epoch": 1.0171102661596958, | |
| "grad_norm": 0.7304844565424085, | |
| "learning_rate": 1.1456011677350052e-05, | |
| "loss": 0.2938, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 1.0190114068441065, | |
| "grad_norm": 0.7136911810627554, | |
| "learning_rate": 1.1423148382732854e-05, | |
| "loss": 0.302, | |
| "step": 536 | |
| }, | |
| { | |
| "epoch": 1.020912547528517, | |
| "grad_norm": 0.7502436210950613, | |
| "learning_rate": 1.1390269392901096e-05, | |
| "loss": 0.2921, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 1.0228136882129277, | |
| "grad_norm": 0.6922039043493811, | |
| "learning_rate": 1.1357375070461241e-05, | |
| "loss": 0.2918, | |
| "step": 538 | |
| }, | |
| { | |
| "epoch": 1.0247148288973384, | |
| "grad_norm": 0.6764290481853367, | |
| "learning_rate": 1.1324465778188846e-05, | |
| "loss": 0.2868, | |
| "step": 539 | |
| }, | |
| { | |
| "epoch": 1.026615969581749, | |
| "grad_norm": 0.6730430310082527, | |
| "learning_rate": 1.1291541879024568e-05, | |
| "loss": 0.2955, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 1.0285171102661597, | |
| "grad_norm": 0.6713050660186555, | |
| "learning_rate": 1.1258603736070145e-05, | |
| "loss": 0.2744, | |
| "step": 541 | |
| }, | |
| { | |
| "epoch": 1.0304182509505704, | |
| "grad_norm": 0.7061709902513252, | |
| "learning_rate": 1.1225651712584413e-05, | |
| "loss": 0.2947, | |
| "step": 542 | |
| }, | |
| { | |
| "epoch": 1.032319391634981, | |
| "grad_norm": 0.6662652645714827, | |
| "learning_rate": 1.1192686171979288e-05, | |
| "loss": 0.2846, | |
| "step": 543 | |
| }, | |
| { | |
| "epoch": 1.0342205323193916, | |
| "grad_norm": 0.6907579555022946, | |
| "learning_rate": 1.1159707477815756e-05, | |
| "loss": 0.2848, | |
| "step": 544 | |
| }, | |
| { | |
| "epoch": 1.0361216730038023, | |
| "grad_norm": 0.8088853789106621, | |
| "learning_rate": 1.1126715993799875e-05, | |
| "loss": 0.2922, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 1.038022813688213, | |
| "grad_norm": 0.7059137358047947, | |
| "learning_rate": 1.1093712083778748e-05, | |
| "loss": 0.2763, | |
| "step": 546 | |
| }, | |
| { | |
| "epoch": 1.0399239543726235, | |
| "grad_norm": 0.7863171740284177, | |
| "learning_rate": 1.1060696111736515e-05, | |
| "loss": 0.3069, | |
| "step": 547 | |
| }, | |
| { | |
| "epoch": 1.0418250950570342, | |
| "grad_norm": 0.8213727009731766, | |
| "learning_rate": 1.1027668441790358e-05, | |
| "loss": 0.2885, | |
| "step": 548 | |
| }, | |
| { | |
| "epoch": 1.043726235741445, | |
| "grad_norm": 0.7369533198500264, | |
| "learning_rate": 1.099462943818646e-05, | |
| "loss": 0.2953, | |
| "step": 549 | |
| }, | |
| { | |
| "epoch": 1.0456273764258555, | |
| "grad_norm": 0.6963474436924002, | |
| "learning_rate": 1.0961579465295987e-05, | |
| "loss": 0.2836, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.0475285171102662, | |
| "grad_norm": 0.7789074969416175, | |
| "learning_rate": 1.0928518887611099e-05, | |
| "loss": 0.2998, | |
| "step": 551 | |
| }, | |
| { | |
| "epoch": 1.049429657794677, | |
| "grad_norm": 0.6952934510193522, | |
| "learning_rate": 1.0895448069740902e-05, | |
| "loss": 0.2923, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 1.0513307984790874, | |
| "grad_norm": 0.6631333353766575, | |
| "learning_rate": 1.0862367376407433e-05, | |
| "loss": 0.2924, | |
| "step": 553 | |
| }, | |
| { | |
| "epoch": 1.053231939163498, | |
| "grad_norm": 0.7273790508474828, | |
| "learning_rate": 1.0829277172441648e-05, | |
| "loss": 0.301, | |
| "step": 554 | |
| }, | |
| { | |
| "epoch": 1.0551330798479088, | |
| "grad_norm": 0.7228822249000719, | |
| "learning_rate": 1.0796177822779384e-05, | |
| "loss": 0.2771, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 1.0570342205323193, | |
| "grad_norm": 0.7024688188551425, | |
| "learning_rate": 1.0763069692457346e-05, | |
| "loss": 0.2992, | |
| "step": 556 | |
| }, | |
| { | |
| "epoch": 1.05893536121673, | |
| "grad_norm": 0.7741871456149745, | |
| "learning_rate": 1.0729953146609076e-05, | |
| "loss": 0.287, | |
| "step": 557 | |
| }, | |
| { | |
| "epoch": 1.0608365019011408, | |
| "grad_norm": 0.6916471475548848, | |
| "learning_rate": 1.0696828550460928e-05, | |
| "loss": 0.2931, | |
| "step": 558 | |
| }, | |
| { | |
| "epoch": 1.0627376425855513, | |
| "grad_norm": 0.704498081398421, | |
| "learning_rate": 1.0663696269328034e-05, | |
| "loss": 0.2976, | |
| "step": 559 | |
| }, | |
| { | |
| "epoch": 1.064638783269962, | |
| "grad_norm": 0.6752559163176793, | |
| "learning_rate": 1.0630556668610286e-05, | |
| "loss": 0.2827, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 1.0665399239543727, | |
| "grad_norm": 0.681227310274368, | |
| "learning_rate": 1.059741011378829e-05, | |
| "loss": 0.2875, | |
| "step": 561 | |
| }, | |
| { | |
| "epoch": 1.0684410646387832, | |
| "grad_norm": 0.6996503058847626, | |
| "learning_rate": 1.0564256970419367e-05, | |
| "loss": 0.2892, | |
| "step": 562 | |
| }, | |
| { | |
| "epoch": 1.070342205323194, | |
| "grad_norm": 0.7310658183035069, | |
| "learning_rate": 1.0531097604133473e-05, | |
| "loss": 0.2912, | |
| "step": 563 | |
| }, | |
| { | |
| "epoch": 1.0722433460076046, | |
| "grad_norm": 0.7158324958421526, | |
| "learning_rate": 1.0497932380629207e-05, | |
| "loss": 0.2834, | |
| "step": 564 | |
| }, | |
| { | |
| "epoch": 1.0741444866920151, | |
| "grad_norm": 0.6695546446317904, | |
| "learning_rate": 1.0464761665669771e-05, | |
| "loss": 0.2871, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 1.0760456273764258, | |
| "grad_norm": 0.6762674706816263, | |
| "learning_rate": 1.0431585825078916e-05, | |
| "loss": 0.288, | |
| "step": 566 | |
| }, | |
| { | |
| "epoch": 1.0779467680608366, | |
| "grad_norm": 0.6884283039501352, | |
| "learning_rate": 1.0398405224736927e-05, | |
| "loss": 0.2943, | |
| "step": 567 | |
| }, | |
| { | |
| "epoch": 1.079847908745247, | |
| "grad_norm": 0.6896642264078586, | |
| "learning_rate": 1.0365220230576592e-05, | |
| "loss": 0.2801, | |
| "step": 568 | |
| }, | |
| { | |
| "epoch": 1.0817490494296578, | |
| "grad_norm": 0.7059097594323285, | |
| "learning_rate": 1.0332031208579133e-05, | |
| "loss": 0.2969, | |
| "step": 569 | |
| }, | |
| { | |
| "epoch": 1.0836501901140685, | |
| "grad_norm": 0.7015422714928794, | |
| "learning_rate": 1.0298838524770212e-05, | |
| "loss": 0.2918, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 1.085551330798479, | |
| "grad_norm": 0.7002472371821828, | |
| "learning_rate": 1.0265642545215872e-05, | |
| "loss": 0.2932, | |
| "step": 571 | |
| }, | |
| { | |
| "epoch": 1.0874524714828897, | |
| "grad_norm": 0.6998970716960781, | |
| "learning_rate": 1.0232443636018502e-05, | |
| "loss": 0.2901, | |
| "step": 572 | |
| }, | |
| { | |
| "epoch": 1.0893536121673004, | |
| "grad_norm": 0.699782723590395, | |
| "learning_rate": 1.0199242163312794e-05, | |
| "loss": 0.289, | |
| "step": 573 | |
| }, | |
| { | |
| "epoch": 1.091254752851711, | |
| "grad_norm": 0.6752711789954923, | |
| "learning_rate": 1.0166038493261723e-05, | |
| "loss": 0.2889, | |
| "step": 574 | |
| }, | |
| { | |
| "epoch": 1.0931558935361216, | |
| "grad_norm": 0.7372613806577799, | |
| "learning_rate": 1.013283299205249e-05, | |
| "loss": 0.2936, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 1.0950570342205324, | |
| "grad_norm": 0.7415117728534114, | |
| "learning_rate": 1.0099626025892491e-05, | |
| "loss": 0.2908, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 1.0969581749049429, | |
| "grad_norm": 0.7296946433771271, | |
| "learning_rate": 1.0066417961005283e-05, | |
| "loss": 0.2916, | |
| "step": 577 | |
| }, | |
| { | |
| "epoch": 1.0988593155893536, | |
| "grad_norm": 0.6976413259069777, | |
| "learning_rate": 1.0033209163626539e-05, | |
| "loss": 0.2837, | |
| "step": 578 | |
| }, | |
| { | |
| "epoch": 1.1007604562737643, | |
| "grad_norm": 0.6950355864447233, | |
| "learning_rate": 1e-05, | |
| "loss": 0.2789, | |
| "step": 579 | |
| }, | |
| { | |
| "epoch": 1.102661596958175, | |
| "grad_norm": 0.731405081003338, | |
| "learning_rate": 9.966790836373465e-06, | |
| "loss": 0.2852, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 1.1045627376425855, | |
| "grad_norm": 0.7278394825569937, | |
| "learning_rate": 9.933582038994719e-06, | |
| "loss": 0.2957, | |
| "step": 581 | |
| }, | |
| { | |
| "epoch": 1.1064638783269962, | |
| "grad_norm": 0.7558673539388312, | |
| "learning_rate": 9.90037397410751e-06, | |
| "loss": 0.2809, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 1.108365019011407, | |
| "grad_norm": 0.705775513584956, | |
| "learning_rate": 9.867167007947511e-06, | |
| "loss": 0.2895, | |
| "step": 583 | |
| }, | |
| { | |
| "epoch": 1.1102661596958174, | |
| "grad_norm": 0.7161014074136918, | |
| "learning_rate": 9.833961506738282e-06, | |
| "loss": 0.2955, | |
| "step": 584 | |
| }, | |
| { | |
| "epoch": 1.1121673003802282, | |
| "grad_norm": 0.7256782212121473, | |
| "learning_rate": 9.80075783668721e-06, | |
| "loss": 0.2874, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 1.1140684410646389, | |
| "grad_norm": 0.6948571842788817, | |
| "learning_rate": 9.767556363981503e-06, | |
| "loss": 0.2865, | |
| "step": 586 | |
| }, | |
| { | |
| "epoch": 1.1159695817490494, | |
| "grad_norm": 0.6903518042233049, | |
| "learning_rate": 9.734357454784131e-06, | |
| "loss": 0.2789, | |
| "step": 587 | |
| }, | |
| { | |
| "epoch": 1.11787072243346, | |
| "grad_norm": 0.6869700695706998, | |
| "learning_rate": 9.701161475229791e-06, | |
| "loss": 0.2843, | |
| "step": 588 | |
| }, | |
| { | |
| "epoch": 1.1197718631178708, | |
| "grad_norm": 0.7154177823545638, | |
| "learning_rate": 9.66796879142087e-06, | |
| "loss": 0.2939, | |
| "step": 589 | |
| }, | |
| { | |
| "epoch": 1.1216730038022813, | |
| "grad_norm": 0.6716868513067762, | |
| "learning_rate": 9.634779769423412e-06, | |
| "loss": 0.2782, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 1.123574144486692, | |
| "grad_norm": 0.6779422452474894, | |
| "learning_rate": 9.601594775263073e-06, | |
| "loss": 0.2821, | |
| "step": 591 | |
| }, | |
| { | |
| "epoch": 1.1254752851711027, | |
| "grad_norm": 0.7064376225277209, | |
| "learning_rate": 9.568414174921085e-06, | |
| "loss": 0.3014, | |
| "step": 592 | |
| }, | |
| { | |
| "epoch": 1.1273764258555132, | |
| "grad_norm": 0.7177402417964964, | |
| "learning_rate": 9.535238334330234e-06, | |
| "loss": 0.2894, | |
| "step": 593 | |
| }, | |
| { | |
| "epoch": 1.129277566539924, | |
| "grad_norm": 0.7256776192675685, | |
| "learning_rate": 9.502067619370794e-06, | |
| "loss": 0.2895, | |
| "step": 594 | |
| }, | |
| { | |
| "epoch": 1.1311787072243347, | |
| "grad_norm": 0.693540284640893, | |
| "learning_rate": 9.468902395866532e-06, | |
| "loss": 0.2906, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 1.1330798479087452, | |
| "grad_norm": 0.6713420942423118, | |
| "learning_rate": 9.435743029580638e-06, | |
| "loss": 0.287, | |
| "step": 596 | |
| }, | |
| { | |
| "epoch": 1.1349809885931559, | |
| "grad_norm": 0.7516419136175736, | |
| "learning_rate": 9.402589886211711e-06, | |
| "loss": 0.2823, | |
| "step": 597 | |
| }, | |
| { | |
| "epoch": 1.1368821292775666, | |
| "grad_norm": 0.7128017406730036, | |
| "learning_rate": 9.369443331389718e-06, | |
| "loss": 0.2981, | |
| "step": 598 | |
| }, | |
| { | |
| "epoch": 1.138783269961977, | |
| "grad_norm": 0.7258209995845645, | |
| "learning_rate": 9.336303730671968e-06, | |
| "loss": 0.2974, | |
| "step": 599 | |
| }, | |
| { | |
| "epoch": 1.1406844106463878, | |
| "grad_norm": 0.8047675767917835, | |
| "learning_rate": 9.303171449539074e-06, | |
| "loss": 0.3003, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.1425855513307985, | |
| "grad_norm": 0.6915844131637288, | |
| "learning_rate": 9.270046853390924e-06, | |
| "loss": 0.2846, | |
| "step": 601 | |
| }, | |
| { | |
| "epoch": 1.144486692015209, | |
| "grad_norm": 0.7147267649899951, | |
| "learning_rate": 9.236930307542654e-06, | |
| "loss": 0.2865, | |
| "step": 602 | |
| }, | |
| { | |
| "epoch": 1.1463878326996197, | |
| "grad_norm": 0.7011448808134404, | |
| "learning_rate": 9.203822177220621e-06, | |
| "loss": 0.2749, | |
| "step": 603 | |
| }, | |
| { | |
| "epoch": 1.1482889733840305, | |
| "grad_norm": 0.8309576237198794, | |
| "learning_rate": 9.170722827558357e-06, | |
| "loss": 0.2885, | |
| "step": 604 | |
| }, | |
| { | |
| "epoch": 1.1501901140684412, | |
| "grad_norm": 0.7059951712193847, | |
| "learning_rate": 9.13763262359257e-06, | |
| "loss": 0.2943, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 1.1520912547528517, | |
| "grad_norm": 0.7197943003361735, | |
| "learning_rate": 9.104551930259101e-06, | |
| "loss": 0.2948, | |
| "step": 606 | |
| }, | |
| { | |
| "epoch": 1.1539923954372624, | |
| "grad_norm": 0.7741184893329084, | |
| "learning_rate": 9.071481112388905e-06, | |
| "loss": 0.2966, | |
| "step": 607 | |
| }, | |
| { | |
| "epoch": 1.1558935361216731, | |
| "grad_norm": 0.671170453211042, | |
| "learning_rate": 9.038420534704015e-06, | |
| "loss": 0.2812, | |
| "step": 608 | |
| }, | |
| { | |
| "epoch": 1.1577946768060836, | |
| "grad_norm": 0.6995105885355609, | |
| "learning_rate": 9.005370561813545e-06, | |
| "loss": 0.3021, | |
| "step": 609 | |
| }, | |
| { | |
| "epoch": 1.1596958174904943, | |
| "grad_norm": 0.6991215413273377, | |
| "learning_rate": 8.972331558209644e-06, | |
| "loss": 0.2968, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 1.161596958174905, | |
| "grad_norm": 0.6481022921361693, | |
| "learning_rate": 8.939303888263485e-06, | |
| "loss": 0.2936, | |
| "step": 611 | |
| }, | |
| { | |
| "epoch": 1.1634980988593155, | |
| "grad_norm": 0.7016092293253109, | |
| "learning_rate": 8.906287916221259e-06, | |
| "loss": 0.2892, | |
| "step": 612 | |
| }, | |
| { | |
| "epoch": 1.1653992395437263, | |
| "grad_norm": 0.6735053653347486, | |
| "learning_rate": 8.873284006200129e-06, | |
| "loss": 0.2812, | |
| "step": 613 | |
| }, | |
| { | |
| "epoch": 1.167300380228137, | |
| "grad_norm": 0.6943496029404549, | |
| "learning_rate": 8.840292522184247e-06, | |
| "loss": 0.2837, | |
| "step": 614 | |
| }, | |
| { | |
| "epoch": 1.1692015209125475, | |
| "grad_norm": 0.6716744616354126, | |
| "learning_rate": 8.807313828020715e-06, | |
| "loss": 0.2912, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 1.1711026615969582, | |
| "grad_norm": 0.7639192241457258, | |
| "learning_rate": 8.774348287415589e-06, | |
| "loss": 0.2996, | |
| "step": 616 | |
| }, | |
| { | |
| "epoch": 1.173003802281369, | |
| "grad_norm": 0.7328723007102226, | |
| "learning_rate": 8.74139626392986e-06, | |
| "loss": 0.295, | |
| "step": 617 | |
| }, | |
| { | |
| "epoch": 1.1749049429657794, | |
| "grad_norm": 0.7081289317051815, | |
| "learning_rate": 8.708458120975436e-06, | |
| "loss": 0.2864, | |
| "step": 618 | |
| }, | |
| { | |
| "epoch": 1.1768060836501901, | |
| "grad_norm": 0.6837874077882323, | |
| "learning_rate": 8.675534221811156e-06, | |
| "loss": 0.2832, | |
| "step": 619 | |
| }, | |
| { | |
| "epoch": 1.1787072243346008, | |
| "grad_norm": 0.7010387610477602, | |
| "learning_rate": 8.64262492953876e-06, | |
| "loss": 0.2842, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 1.1806083650190113, | |
| "grad_norm": 0.7479947764040538, | |
| "learning_rate": 8.60973060709891e-06, | |
| "loss": 0.3094, | |
| "step": 621 | |
| }, | |
| { | |
| "epoch": 1.182509505703422, | |
| "grad_norm": 0.7170186075953432, | |
| "learning_rate": 8.576851617267151e-06, | |
| "loss": 0.2903, | |
| "step": 622 | |
| }, | |
| { | |
| "epoch": 1.1844106463878328, | |
| "grad_norm": 0.7266952761403496, | |
| "learning_rate": 8.543988322649954e-06, | |
| "loss": 0.3084, | |
| "step": 623 | |
| }, | |
| { | |
| "epoch": 1.1863117870722433, | |
| "grad_norm": 0.6795779427148725, | |
| "learning_rate": 8.511141085680684e-06, | |
| "loss": 0.2882, | |
| "step": 624 | |
| }, | |
| { | |
| "epoch": 1.188212927756654, | |
| "grad_norm": 0.7776491024235828, | |
| "learning_rate": 8.478310268615612e-06, | |
| "loss": 0.305, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 1.1901140684410647, | |
| "grad_norm": 0.662764212238965, | |
| "learning_rate": 8.445496233529934e-06, | |
| "loss": 0.2867, | |
| "step": 626 | |
| }, | |
| { | |
| "epoch": 1.1920152091254752, | |
| "grad_norm": 0.7035333574027661, | |
| "learning_rate": 8.41269934231375e-06, | |
| "loss": 0.3021, | |
| "step": 627 | |
| }, | |
| { | |
| "epoch": 1.193916349809886, | |
| "grad_norm": 0.6799324578798722, | |
| "learning_rate": 8.3799199566681e-06, | |
| "loss": 0.2761, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 1.1958174904942966, | |
| "grad_norm": 0.6817765744269734, | |
| "learning_rate": 8.34715843810096e-06, | |
| "loss": 0.2786, | |
| "step": 629 | |
| }, | |
| { | |
| "epoch": 1.1977186311787071, | |
| "grad_norm": 0.7075562237878374, | |
| "learning_rate": 8.314415147923254e-06, | |
| "loss": 0.2924, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 1.1996197718631179, | |
| "grad_norm": 0.7338756872874812, | |
| "learning_rate": 8.281690447244887e-06, | |
| "loss": 0.2792, | |
| "step": 631 | |
| }, | |
| { | |
| "epoch": 1.2015209125475286, | |
| "grad_norm": 0.6946767048089274, | |
| "learning_rate": 8.248984696970732e-06, | |
| "loss": 0.2885, | |
| "step": 632 | |
| }, | |
| { | |
| "epoch": 1.203422053231939, | |
| "grad_norm": 0.7422398591315209, | |
| "learning_rate": 8.216298257796677e-06, | |
| "loss": 0.302, | |
| "step": 633 | |
| }, | |
| { | |
| "epoch": 1.2053231939163498, | |
| "grad_norm": 0.7118666463806281, | |
| "learning_rate": 8.183631490205636e-06, | |
| "loss": 0.2834, | |
| "step": 634 | |
| }, | |
| { | |
| "epoch": 1.2072243346007605, | |
| "grad_norm": 0.7246815563048843, | |
| "learning_rate": 8.150984754463578e-06, | |
| "loss": 0.2857, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 1.209125475285171, | |
| "grad_norm": 0.7042138889215468, | |
| "learning_rate": 8.118358410615545e-06, | |
| "loss": 0.2792, | |
| "step": 636 | |
| }, | |
| { | |
| "epoch": 1.2110266159695817, | |
| "grad_norm": 0.7435747882066673, | |
| "learning_rate": 8.08575281848169e-06, | |
| "loss": 0.2782, | |
| "step": 637 | |
| }, | |
| { | |
| "epoch": 1.2129277566539924, | |
| "grad_norm": 0.7696459061749992, | |
| "learning_rate": 8.0531683376533e-06, | |
| "loss": 0.2779, | |
| "step": 638 | |
| }, | |
| { | |
| "epoch": 1.214828897338403, | |
| "grad_norm": 0.7175049818230963, | |
| "learning_rate": 8.020605327488846e-06, | |
| "loss": 0.2926, | |
| "step": 639 | |
| }, | |
| { | |
| "epoch": 1.2167300380228137, | |
| "grad_norm": 0.7389663952755043, | |
| "learning_rate": 7.988064147110001e-06, | |
| "loss": 0.2998, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 1.2186311787072244, | |
| "grad_norm": 0.7039834488507702, | |
| "learning_rate": 7.955545155397684e-06, | |
| "loss": 0.2914, | |
| "step": 641 | |
| }, | |
| { | |
| "epoch": 1.2205323193916349, | |
| "grad_norm": 0.6893202348058041, | |
| "learning_rate": 7.923048710988119e-06, | |
| "loss": 0.2844, | |
| "step": 642 | |
| }, | |
| { | |
| "epoch": 1.2224334600760456, | |
| "grad_norm": 0.718292926181725, | |
| "learning_rate": 7.890575172268858e-06, | |
| "loss": 0.2963, | |
| "step": 643 | |
| }, | |
| { | |
| "epoch": 1.2243346007604563, | |
| "grad_norm": 0.7020714849172128, | |
| "learning_rate": 7.858124897374837e-06, | |
| "loss": 0.2847, | |
| "step": 644 | |
| }, | |
| { | |
| "epoch": 1.2262357414448668, | |
| "grad_norm": 0.7200909047326488, | |
| "learning_rate": 7.825698244184432e-06, | |
| "loss": 0.2883, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 1.2281368821292775, | |
| "grad_norm": 0.7167461912462018, | |
| "learning_rate": 7.7932955703155e-06, | |
| "loss": 0.2805, | |
| "step": 646 | |
| }, | |
| { | |
| "epoch": 1.2300380228136882, | |
| "grad_norm": 0.733315549182106, | |
| "learning_rate": 7.760917233121443e-06, | |
| "loss": 0.3014, | |
| "step": 647 | |
| }, | |
| { | |
| "epoch": 1.231939163498099, | |
| "grad_norm": 0.7144343626049009, | |
| "learning_rate": 7.728563589687275e-06, | |
| "loss": 0.2772, | |
| "step": 648 | |
| }, | |
| { | |
| "epoch": 1.2338403041825095, | |
| "grad_norm": 0.7190483047925709, | |
| "learning_rate": 7.696234996825663e-06, | |
| "loss": 0.286, | |
| "step": 649 | |
| }, | |
| { | |
| "epoch": 1.2357414448669202, | |
| "grad_norm": 0.6715209363326557, | |
| "learning_rate": 7.663931811073003e-06, | |
| "loss": 0.2831, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.2376425855513309, | |
| "grad_norm": 0.7383230955252225, | |
| "learning_rate": 7.631654388685496e-06, | |
| "loss": 0.2869, | |
| "step": 651 | |
| }, | |
| { | |
| "epoch": 1.2395437262357414, | |
| "grad_norm": 0.7605432701879106, | |
| "learning_rate": 7.599403085635208e-06, | |
| "loss": 0.2852, | |
| "step": 652 | |
| }, | |
| { | |
| "epoch": 1.241444866920152, | |
| "grad_norm": 0.7282610678445604, | |
| "learning_rate": 7.567178257606147e-06, | |
| "loss": 0.2941, | |
| "step": 653 | |
| }, | |
| { | |
| "epoch": 1.2433460076045628, | |
| "grad_norm": 0.6658903175737286, | |
| "learning_rate": 7.534980259990341e-06, | |
| "loss": 0.2796, | |
| "step": 654 | |
| }, | |
| { | |
| "epoch": 1.2452471482889733, | |
| "grad_norm": 0.6991792371264595, | |
| "learning_rate": 7.50280944788392e-06, | |
| "loss": 0.2843, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 1.247148288973384, | |
| "grad_norm": 0.705458667520552, | |
| "learning_rate": 7.470666176083193e-06, | |
| "loss": 0.2968, | |
| "step": 656 | |
| }, | |
| { | |
| "epoch": 1.2490494296577948, | |
| "grad_norm": 0.7631742673654526, | |
| "learning_rate": 7.438550799080746e-06, | |
| "loss": 0.2901, | |
| "step": 657 | |
| }, | |
| { | |
| "epoch": 1.2509505703422052, | |
| "grad_norm": 0.7063031532301915, | |
| "learning_rate": 7.40646367106153e-06, | |
| "loss": 0.2932, | |
| "step": 658 | |
| }, | |
| { | |
| "epoch": 1.252851711026616, | |
| "grad_norm": 0.6582108683771557, | |
| "learning_rate": 7.3744051458989395e-06, | |
| "loss": 0.2788, | |
| "step": 659 | |
| }, | |
| { | |
| "epoch": 1.2547528517110267, | |
| "grad_norm": 0.6799561864162934, | |
| "learning_rate": 7.342375577150928e-06, | |
| "loss": 0.2813, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 1.2566539923954372, | |
| "grad_norm": 0.7125150659723062, | |
| "learning_rate": 7.310375318056107e-06, | |
| "loss": 0.2963, | |
| "step": 661 | |
| }, | |
| { | |
| "epoch": 1.258555133079848, | |
| "grad_norm": 0.7088244150642155, | |
| "learning_rate": 7.278404721529843e-06, | |
| "loss": 0.279, | |
| "step": 662 | |
| }, | |
| { | |
| "epoch": 1.2604562737642586, | |
| "grad_norm": 0.6795116665311195, | |
| "learning_rate": 7.246464140160365e-06, | |
| "loss": 0.2761, | |
| "step": 663 | |
| }, | |
| { | |
| "epoch": 1.2623574144486693, | |
| "grad_norm": 0.7516701361616173, | |
| "learning_rate": 7.214553926204884e-06, | |
| "loss": 0.286, | |
| "step": 664 | |
| }, | |
| { | |
| "epoch": 1.2642585551330798, | |
| "grad_norm": 0.710218839301933, | |
| "learning_rate": 7.182674431585703e-06, | |
| "loss": 0.2913, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 1.2661596958174905, | |
| "grad_norm": 0.7841079248748619, | |
| "learning_rate": 7.150826007886334e-06, | |
| "loss": 0.2938, | |
| "step": 666 | |
| }, | |
| { | |
| "epoch": 1.2680608365019013, | |
| "grad_norm": 0.6976286182717685, | |
| "learning_rate": 7.119009006347625e-06, | |
| "loss": 0.2758, | |
| "step": 667 | |
| }, | |
| { | |
| "epoch": 1.2699619771863118, | |
| "grad_norm": 0.7228585207032258, | |
| "learning_rate": 7.087223777863883e-06, | |
| "loss": 0.2922, | |
| "step": 668 | |
| }, | |
| { | |
| "epoch": 1.2718631178707225, | |
| "grad_norm": 0.7505774782539952, | |
| "learning_rate": 7.055470672979003e-06, | |
| "loss": 0.2938, | |
| "step": 669 | |
| }, | |
| { | |
| "epoch": 1.2737642585551332, | |
| "grad_norm": 0.7107274614813491, | |
| "learning_rate": 7.023750041882609e-06, | |
| "loss": 0.2891, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 1.2756653992395437, | |
| "grad_norm": 0.7097285083695535, | |
| "learning_rate": 6.992062234406185e-06, | |
| "loss": 0.2849, | |
| "step": 671 | |
| }, | |
| { | |
| "epoch": 1.2775665399239544, | |
| "grad_norm": 0.7261080301802124, | |
| "learning_rate": 6.960407600019217e-06, | |
| "loss": 0.2852, | |
| "step": 672 | |
| }, | |
| { | |
| "epoch": 1.2794676806083651, | |
| "grad_norm": 0.8115850076337936, | |
| "learning_rate": 6.9287864878253475e-06, | |
| "loss": 0.2929, | |
| "step": 673 | |
| }, | |
| { | |
| "epoch": 1.2813688212927756, | |
| "grad_norm": 0.7090460414373848, | |
| "learning_rate": 6.897199246558515e-06, | |
| "loss": 0.2777, | |
| "step": 674 | |
| }, | |
| { | |
| "epoch": 1.2832699619771863, | |
| "grad_norm": 0.681077900206575, | |
| "learning_rate": 6.865646224579108e-06, | |
| "loss": 0.2843, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 1.285171102661597, | |
| "grad_norm": 0.7694853595792261, | |
| "learning_rate": 6.834127769870134e-06, | |
| "loss": 0.2815, | |
| "step": 676 | |
| }, | |
| { | |
| "epoch": 1.2870722433460076, | |
| "grad_norm": 0.7405583138237021, | |
| "learning_rate": 6.802644230033373e-06, | |
| "loss": 0.2979, | |
| "step": 677 | |
| }, | |
| { | |
| "epoch": 1.2889733840304183, | |
| "grad_norm": 0.7207719626138162, | |
| "learning_rate": 6.771195952285541e-06, | |
| "loss": 0.29, | |
| "step": 678 | |
| }, | |
| { | |
| "epoch": 1.290874524714829, | |
| "grad_norm": 0.7458723865519009, | |
| "learning_rate": 6.739783283454469e-06, | |
| "loss": 0.2883, | |
| "step": 679 | |
| }, | |
| { | |
| "epoch": 1.2927756653992395, | |
| "grad_norm": 0.6699649154881325, | |
| "learning_rate": 6.708406569975274e-06, | |
| "loss": 0.2734, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 1.2946768060836502, | |
| "grad_norm": 0.7212774551389382, | |
| "learning_rate": 6.6770661578865444e-06, | |
| "loss": 0.2828, | |
| "step": 681 | |
| }, | |
| { | |
| "epoch": 1.296577946768061, | |
| "grad_norm": 0.7099141504068575, | |
| "learning_rate": 6.645762392826509e-06, | |
| "loss": 0.2933, | |
| "step": 682 | |
| }, | |
| { | |
| "epoch": 1.2984790874524714, | |
| "grad_norm": 0.7081684664666053, | |
| "learning_rate": 6.614495620029238e-06, | |
| "loss": 0.2925, | |
| "step": 683 | |
| }, | |
| { | |
| "epoch": 1.3003802281368821, | |
| "grad_norm": 0.7160316648624385, | |
| "learning_rate": 6.583266184320836e-06, | |
| "loss": 0.2842, | |
| "step": 684 | |
| }, | |
| { | |
| "epoch": 1.3022813688212929, | |
| "grad_norm": 0.7129388916783157, | |
| "learning_rate": 6.552074430115624e-06, | |
| "loss": 0.2976, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 1.3041825095057034, | |
| "grad_norm": 0.6832912099044737, | |
| "learning_rate": 6.520920701412371e-06, | |
| "loss": 0.2739, | |
| "step": 686 | |
| }, | |
| { | |
| "epoch": 1.306083650190114, | |
| "grad_norm": 0.669171647769967, | |
| "learning_rate": 6.489805341790456e-06, | |
| "loss": 0.2772, | |
| "step": 687 | |
| }, | |
| { | |
| "epoch": 1.3079847908745248, | |
| "grad_norm": 0.7321466366283711, | |
| "learning_rate": 6.458728694406124e-06, | |
| "loss": 0.2959, | |
| "step": 688 | |
| }, | |
| { | |
| "epoch": 1.3098859315589353, | |
| "grad_norm": 0.7039098126424078, | |
| "learning_rate": 6.427691101988673e-06, | |
| "loss": 0.2844, | |
| "step": 689 | |
| }, | |
| { | |
| "epoch": 1.311787072243346, | |
| "grad_norm": 0.7109080068601229, | |
| "learning_rate": 6.396692906836686e-06, | |
| "loss": 0.2879, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 1.3136882129277567, | |
| "grad_norm": 0.6944384101495022, | |
| "learning_rate": 6.3657344508142495e-06, | |
| "loss": 0.277, | |
| "step": 691 | |
| }, | |
| { | |
| "epoch": 1.3155893536121672, | |
| "grad_norm": 0.7348778482117885, | |
| "learning_rate": 6.334816075347185e-06, | |
| "loss": 0.2841, | |
| "step": 692 | |
| }, | |
| { | |
| "epoch": 1.317490494296578, | |
| "grad_norm": 0.6865016893129265, | |
| "learning_rate": 6.303938121419295e-06, | |
| "loss": 0.2768, | |
| "step": 693 | |
| }, | |
| { | |
| "epoch": 1.3193916349809887, | |
| "grad_norm": 0.7015844893848481, | |
| "learning_rate": 6.273100929568579e-06, | |
| "loss": 0.2842, | |
| "step": 694 | |
| }, | |
| { | |
| "epoch": 1.3212927756653992, | |
| "grad_norm": 0.7015620581412725, | |
| "learning_rate": 6.242304839883502e-06, | |
| "loss": 0.2897, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 1.3231939163498099, | |
| "grad_norm": 0.6753625751541211, | |
| "learning_rate": 6.211550191999223e-06, | |
| "loss": 0.2906, | |
| "step": 696 | |
| }, | |
| { | |
| "epoch": 1.3250950570342206, | |
| "grad_norm": 0.7486556594868576, | |
| "learning_rate": 6.18083732509387e-06, | |
| "loss": 0.2886, | |
| "step": 697 | |
| }, | |
| { | |
| "epoch": 1.326996197718631, | |
| "grad_norm": 0.711790663346095, | |
| "learning_rate": 6.150166577884781e-06, | |
| "loss": 0.2815, | |
| "step": 698 | |
| }, | |
| { | |
| "epoch": 1.3288973384030418, | |
| "grad_norm": 0.7378444993556201, | |
| "learning_rate": 6.119538288624778e-06, | |
| "loss": 0.2808, | |
| "step": 699 | |
| }, | |
| { | |
| "epoch": 1.3307984790874525, | |
| "grad_norm": 0.7147095322172046, | |
| "learning_rate": 6.088952795098442e-06, | |
| "loss": 0.2848, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.332699619771863, | |
| "grad_norm": 0.7362222846764638, | |
| "learning_rate": 6.058410434618367e-06, | |
| "loss": 0.2851, | |
| "step": 701 | |
| }, | |
| { | |
| "epoch": 1.3346007604562737, | |
| "grad_norm": 0.6885396529320125, | |
| "learning_rate": 6.027911544021465e-06, | |
| "loss": 0.2786, | |
| "step": 702 | |
| }, | |
| { | |
| "epoch": 1.3365019011406845, | |
| "grad_norm": 0.6740508744631317, | |
| "learning_rate": 5.997456459665237e-06, | |
| "loss": 0.2858, | |
| "step": 703 | |
| }, | |
| { | |
| "epoch": 1.338403041825095, | |
| "grad_norm": 0.6840152622376718, | |
| "learning_rate": 5.967045517424062e-06, | |
| "loss": 0.2827, | |
| "step": 704 | |
| }, | |
| { | |
| "epoch": 1.3403041825095057, | |
| "grad_norm": 0.7105718149869477, | |
| "learning_rate": 5.936679052685505e-06, | |
| "loss": 0.2859, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 1.3422053231939164, | |
| "grad_norm": 0.7277095025134493, | |
| "learning_rate": 5.906357400346596e-06, | |
| "loss": 0.2876, | |
| "step": 706 | |
| }, | |
| { | |
| "epoch": 1.3441064638783269, | |
| "grad_norm": 0.7016522584825272, | |
| "learning_rate": 5.876080894810167e-06, | |
| "loss": 0.2849, | |
| "step": 707 | |
| }, | |
| { | |
| "epoch": 1.3460076045627376, | |
| "grad_norm": 0.6962181083405024, | |
| "learning_rate": 5.845849869981137e-06, | |
| "loss": 0.2823, | |
| "step": 708 | |
| }, | |
| { | |
| "epoch": 1.3479087452471483, | |
| "grad_norm": 0.6640308587499374, | |
| "learning_rate": 5.815664659262845e-06, | |
| "loss": 0.289, | |
| "step": 709 | |
| }, | |
| { | |
| "epoch": 1.3498098859315588, | |
| "grad_norm": 0.7294346552636364, | |
| "learning_rate": 5.78552559555337e-06, | |
| "loss": 0.2824, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 1.3517110266159695, | |
| "grad_norm": 0.6967819680270614, | |
| "learning_rate": 5.755433011241851e-06, | |
| "loss": 0.2704, | |
| "step": 711 | |
| }, | |
| { | |
| "epoch": 1.3536121673003803, | |
| "grad_norm": 0.6991570465817818, | |
| "learning_rate": 5.725387238204831e-06, | |
| "loss": 0.2957, | |
| "step": 712 | |
| }, | |
| { | |
| "epoch": 1.3555133079847907, | |
| "grad_norm": 0.7167183897811851, | |
| "learning_rate": 5.695388607802603e-06, | |
| "loss": 0.2924, | |
| "step": 713 | |
| }, | |
| { | |
| "epoch": 1.3574144486692015, | |
| "grad_norm": 0.6779523722642533, | |
| "learning_rate": 5.665437450875534e-06, | |
| "loss": 0.281, | |
| "step": 714 | |
| }, | |
| { | |
| "epoch": 1.3593155893536122, | |
| "grad_norm": 0.7112601375822984, | |
| "learning_rate": 5.635534097740435e-06, | |
| "loss": 0.2818, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 1.3612167300380227, | |
| "grad_norm": 0.7394398538763441, | |
| "learning_rate": 5.605678878186911e-06, | |
| "loss": 0.2864, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 1.3631178707224334, | |
| "grad_norm": 0.687430001800992, | |
| "learning_rate": 5.575872121473722e-06, | |
| "loss": 0.2772, | |
| "step": 717 | |
| }, | |
| { | |
| "epoch": 1.3650190114068441, | |
| "grad_norm": 0.7388488885668707, | |
| "learning_rate": 5.546114156325166e-06, | |
| "loss": 0.2891, | |
| "step": 718 | |
| }, | |
| { | |
| "epoch": 1.3669201520912546, | |
| "grad_norm": 0.6770076945366967, | |
| "learning_rate": 5.516405310927431e-06, | |
| "loss": 0.279, | |
| "step": 719 | |
| }, | |
| { | |
| "epoch": 1.3688212927756653, | |
| "grad_norm": 0.7015960524841336, | |
| "learning_rate": 5.4867459129249846e-06, | |
| "loss": 0.2728, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 1.370722433460076, | |
| "grad_norm": 0.6971539269894002, | |
| "learning_rate": 5.4571362894169795e-06, | |
| "loss": 0.2866, | |
| "step": 721 | |
| }, | |
| { | |
| "epoch": 1.3726235741444868, | |
| "grad_norm": 0.7137906839009186, | |
| "learning_rate": 5.427576766953615e-06, | |
| "loss": 0.2794, | |
| "step": 722 | |
| }, | |
| { | |
| "epoch": 1.3745247148288973, | |
| "grad_norm": 0.6867045815992675, | |
| "learning_rate": 5.398067671532554e-06, | |
| "loss": 0.2797, | |
| "step": 723 | |
| }, | |
| { | |
| "epoch": 1.376425855513308, | |
| "grad_norm": 0.6751411917697044, | |
| "learning_rate": 5.368609328595323e-06, | |
| "loss": 0.277, | |
| "step": 724 | |
| }, | |
| { | |
| "epoch": 1.3783269961977187, | |
| "grad_norm": 0.7076275318989775, | |
| "learning_rate": 5.339202063023727e-06, | |
| "loss": 0.2763, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 1.3802281368821292, | |
| "grad_norm": 0.7083451557421777, | |
| "learning_rate": 5.309846199136258e-06, | |
| "loss": 0.2711, | |
| "step": 726 | |
| }, | |
| { | |
| "epoch": 1.38212927756654, | |
| "grad_norm": 0.6934315812499172, | |
| "learning_rate": 5.280542060684535e-06, | |
| "loss": 0.2784, | |
| "step": 727 | |
| }, | |
| { | |
| "epoch": 1.3840304182509506, | |
| "grad_norm": 0.7174469786210449, | |
| "learning_rate": 5.2512899708497086e-06, | |
| "loss": 0.2796, | |
| "step": 728 | |
| }, | |
| { | |
| "epoch": 1.3859315589353614, | |
| "grad_norm": 0.6662380596814552, | |
| "learning_rate": 5.222090252238916e-06, | |
| "loss": 0.2722, | |
| "step": 729 | |
| }, | |
| { | |
| "epoch": 1.3878326996197718, | |
| "grad_norm": 0.7159778195181844, | |
| "learning_rate": 5.192943226881724e-06, | |
| "loss": 0.2684, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 1.3897338403041826, | |
| "grad_norm": 0.7233890566671783, | |
| "learning_rate": 5.163849216226562e-06, | |
| "loss": 0.2951, | |
| "step": 731 | |
| }, | |
| { | |
| "epoch": 1.3916349809885933, | |
| "grad_norm": 0.6784428160543446, | |
| "learning_rate": 5.134808541137183e-06, | |
| "loss": 0.2805, | |
| "step": 732 | |
| }, | |
| { | |
| "epoch": 1.3935361216730038, | |
| "grad_norm": 0.7245723000944231, | |
| "learning_rate": 5.105821521889147e-06, | |
| "loss": 0.295, | |
| "step": 733 | |
| }, | |
| { | |
| "epoch": 1.3954372623574145, | |
| "grad_norm": 0.7152594998315716, | |
| "learning_rate": 5.076888478166247e-06, | |
| "loss": 0.2883, | |
| "step": 734 | |
| }, | |
| { | |
| "epoch": 1.3973384030418252, | |
| "grad_norm": 0.6971380497757804, | |
| "learning_rate": 5.048009729057012e-06, | |
| "loss": 0.2801, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 1.3992395437262357, | |
| "grad_norm": 0.7156239750237565, | |
| "learning_rate": 5.0191855930511946e-06, | |
| "loss": 0.2917, | |
| "step": 736 | |
| }, | |
| { | |
| "epoch": 1.4011406844106464, | |
| "grad_norm": 0.6557339679621987, | |
| "learning_rate": 4.990416388036233e-06, | |
| "loss": 0.2715, | |
| "step": 737 | |
| }, | |
| { | |
| "epoch": 1.4030418250950571, | |
| "grad_norm": 0.7200397015319556, | |
| "learning_rate": 4.961702431293759e-06, | |
| "loss": 0.2928, | |
| "step": 738 | |
| }, | |
| { | |
| "epoch": 1.4049429657794676, | |
| "grad_norm": 0.7011678331571025, | |
| "learning_rate": 4.933044039496107e-06, | |
| "loss": 0.2771, | |
| "step": 739 | |
| }, | |
| { | |
| "epoch": 1.4068441064638784, | |
| "grad_norm": 0.6711267315969884, | |
| "learning_rate": 4.904441528702806e-06, | |
| "loss": 0.2714, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 1.408745247148289, | |
| "grad_norm": 0.7577637501858279, | |
| "learning_rate": 4.875895214357093e-06, | |
| "loss": 0.2892, | |
| "step": 741 | |
| }, | |
| { | |
| "epoch": 1.4106463878326996, | |
| "grad_norm": 0.7266803720621844, | |
| "learning_rate": 4.847405411282462e-06, | |
| "loss": 0.2945, | |
| "step": 742 | |
| }, | |
| { | |
| "epoch": 1.4125475285171103, | |
| "grad_norm": 0.6805573614551326, | |
| "learning_rate": 4.818972433679145e-06, | |
| "loss": 0.2812, | |
| "step": 743 | |
| }, | |
| { | |
| "epoch": 1.414448669201521, | |
| "grad_norm": 0.7240617623195237, | |
| "learning_rate": 4.790596595120699e-06, | |
| "loss": 0.2744, | |
| "step": 744 | |
| }, | |
| { | |
| "epoch": 1.4163498098859315, | |
| "grad_norm": 0.6402584452230183, | |
| "learning_rate": 4.762278208550505e-06, | |
| "loss": 0.2663, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 1.4182509505703422, | |
| "grad_norm": 0.6759375205142505, | |
| "learning_rate": 4.734017586278337e-06, | |
| "loss": 0.2815, | |
| "step": 746 | |
| }, | |
| { | |
| "epoch": 1.420152091254753, | |
| "grad_norm": 0.6846596111104619, | |
| "learning_rate": 4.7058150399769245e-06, | |
| "loss": 0.2841, | |
| "step": 747 | |
| }, | |
| { | |
| "epoch": 1.4220532319391634, | |
| "grad_norm": 0.7077060690881396, | |
| "learning_rate": 4.677670880678493e-06, | |
| "loss": 0.2843, | |
| "step": 748 | |
| }, | |
| { | |
| "epoch": 1.4239543726235742, | |
| "grad_norm": 0.6707134976304687, | |
| "learning_rate": 4.649585418771348e-06, | |
| "loss": 0.2756, | |
| "step": 749 | |
| }, | |
| { | |
| "epoch": 1.4258555133079849, | |
| "grad_norm": 0.7024687190868353, | |
| "learning_rate": 4.621558963996458e-06, | |
| "loss": 0.2825, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.4277566539923954, | |
| "grad_norm": 0.7033820945287464, | |
| "learning_rate": 4.593591825444028e-06, | |
| "loss": 0.2789, | |
| "step": 751 | |
| }, | |
| { | |
| "epoch": 1.429657794676806, | |
| "grad_norm": 0.7147583548491873, | |
| "learning_rate": 4.565684311550077e-06, | |
| "loss": 0.2754, | |
| "step": 752 | |
| }, | |
| { | |
| "epoch": 1.4315589353612168, | |
| "grad_norm": 0.7354507952492936, | |
| "learning_rate": 4.537836730093077e-06, | |
| "loss": 0.2917, | |
| "step": 753 | |
| }, | |
| { | |
| "epoch": 1.4334600760456273, | |
| "grad_norm": 0.6893377484207849, | |
| "learning_rate": 4.510049388190518e-06, | |
| "loss": 0.2793, | |
| "step": 754 | |
| }, | |
| { | |
| "epoch": 1.435361216730038, | |
| "grad_norm": 0.7053788455661335, | |
| "learning_rate": 4.482322592295541e-06, | |
| "loss": 0.2796, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 1.4372623574144487, | |
| "grad_norm": 0.6887312402078235, | |
| "learning_rate": 4.454656648193559e-06, | |
| "loss": 0.2713, | |
| "step": 756 | |
| }, | |
| { | |
| "epoch": 1.4391634980988592, | |
| "grad_norm": 0.6596062463828095, | |
| "learning_rate": 4.427051860998877e-06, | |
| "loss": 0.2795, | |
| "step": 757 | |
| }, | |
| { | |
| "epoch": 1.44106463878327, | |
| "grad_norm": 0.6624406930358201, | |
| "learning_rate": 4.399508535151321e-06, | |
| "loss": 0.2673, | |
| "step": 758 | |
| }, | |
| { | |
| "epoch": 1.4429657794676807, | |
| "grad_norm": 0.7052253288648378, | |
| "learning_rate": 4.372026974412907e-06, | |
| "loss": 0.2817, | |
| "step": 759 | |
| }, | |
| { | |
| "epoch": 1.4448669201520912, | |
| "grad_norm": 0.7241934382245591, | |
| "learning_rate": 4.344607481864466e-06, | |
| "loss": 0.2852, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 1.446768060836502, | |
| "grad_norm": 0.684060619515388, | |
| "learning_rate": 4.317250359902295e-06, | |
| "loss": 0.2832, | |
| "step": 761 | |
| }, | |
| { | |
| "epoch": 1.4486692015209126, | |
| "grad_norm": 0.7097665716011491, | |
| "learning_rate": 4.2899559102348585e-06, | |
| "loss": 0.2832, | |
| "step": 762 | |
| }, | |
| { | |
| "epoch": 1.450570342205323, | |
| "grad_norm": 0.6857324527858434, | |
| "learning_rate": 4.262724433879427e-06, | |
| "loss": 0.2774, | |
| "step": 763 | |
| }, | |
| { | |
| "epoch": 1.4524714828897338, | |
| "grad_norm": 0.7070680247313532, | |
| "learning_rate": 4.235556231158765e-06, | |
| "loss": 0.2829, | |
| "step": 764 | |
| }, | |
| { | |
| "epoch": 1.4543726235741445, | |
| "grad_norm": 0.6773739566625588, | |
| "learning_rate": 4.208451601697836e-06, | |
| "loss": 0.2734, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 1.456273764258555, | |
| "grad_norm": 0.66095059829, | |
| "learning_rate": 4.181410844420473e-06, | |
| "loss": 0.2777, | |
| "step": 766 | |
| }, | |
| { | |
| "epoch": 1.4581749049429658, | |
| "grad_norm": 0.7257464910986734, | |
| "learning_rate": 4.154434257546095e-06, | |
| "loss": 0.2775, | |
| "step": 767 | |
| }, | |
| { | |
| "epoch": 1.4600760456273765, | |
| "grad_norm": 0.6638586785050444, | |
| "learning_rate": 4.127522138586424e-06, | |
| "loss": 0.2736, | |
| "step": 768 | |
| }, | |
| { | |
| "epoch": 1.461977186311787, | |
| "grad_norm": 0.6878241734465779, | |
| "learning_rate": 4.10067478434219e-06, | |
| "loss": 0.2764, | |
| "step": 769 | |
| }, | |
| { | |
| "epoch": 1.4638783269961977, | |
| "grad_norm": 0.689012234887939, | |
| "learning_rate": 4.073892490899865e-06, | |
| "loss": 0.2791, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 1.4657794676806084, | |
| "grad_norm": 0.6801449191041793, | |
| "learning_rate": 4.047175553628397e-06, | |
| "loss": 0.2809, | |
| "step": 771 | |
| }, | |
| { | |
| "epoch": 1.467680608365019, | |
| "grad_norm": 0.7029380116992813, | |
| "learning_rate": 4.020524267175954e-06, | |
| "loss": 0.2822, | |
| "step": 772 | |
| }, | |
| { | |
| "epoch": 1.4695817490494296, | |
| "grad_norm": 0.6991470277525647, | |
| "learning_rate": 3.993938925466674e-06, | |
| "loss": 0.2775, | |
| "step": 773 | |
| }, | |
| { | |
| "epoch": 1.4714828897338403, | |
| "grad_norm": 0.6785062525518484, | |
| "learning_rate": 3.96741982169742e-06, | |
| "loss": 0.2792, | |
| "step": 774 | |
| }, | |
| { | |
| "epoch": 1.4733840304182508, | |
| "grad_norm": 0.7123967643545539, | |
| "learning_rate": 3.9409672483345465e-06, | |
| "loss": 0.2785, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 1.4752851711026616, | |
| "grad_norm": 0.7502634084478487, | |
| "learning_rate": 3.914581497110684e-06, | |
| "loss": 0.2785, | |
| "step": 776 | |
| }, | |
| { | |
| "epoch": 1.4771863117870723, | |
| "grad_norm": 0.706155272751926, | |
| "learning_rate": 3.888262859021508e-06, | |
| "loss": 0.2765, | |
| "step": 777 | |
| }, | |
| { | |
| "epoch": 1.4790874524714828, | |
| "grad_norm": 0.7055901876192786, | |
| "learning_rate": 3.862011624322534e-06, | |
| "loss": 0.2724, | |
| "step": 778 | |
| }, | |
| { | |
| "epoch": 1.4809885931558935, | |
| "grad_norm": 0.7065045367483449, | |
| "learning_rate": 3.835828082525925e-06, | |
| "loss": 0.2865, | |
| "step": 779 | |
| }, | |
| { | |
| "epoch": 1.4828897338403042, | |
| "grad_norm": 0.6989364362375882, | |
| "learning_rate": 3.8097125223972864e-06, | |
| "loss": 0.2836, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 1.4847908745247147, | |
| "grad_norm": 0.7082736407526695, | |
| "learning_rate": 3.7836652319524835e-06, | |
| "loss": 0.282, | |
| "step": 781 | |
| }, | |
| { | |
| "epoch": 1.4866920152091254, | |
| "grad_norm": 0.739340856217501, | |
| "learning_rate": 3.7576864984544814e-06, | |
| "loss": 0.2899, | |
| "step": 782 | |
| }, | |
| { | |
| "epoch": 1.4885931558935361, | |
| "grad_norm": 0.7379971693415569, | |
| "learning_rate": 3.73177660841015e-06, | |
| "loss": 0.2963, | |
| "step": 783 | |
| }, | |
| { | |
| "epoch": 1.4904942965779466, | |
| "grad_norm": 0.6959560196532841, | |
| "learning_rate": 3.7059358475671225e-06, | |
| "loss": 0.2728, | |
| "step": 784 | |
| }, | |
| { | |
| "epoch": 1.4923954372623573, | |
| "grad_norm": 0.7025413831801317, | |
| "learning_rate": 3.680164500910646e-06, | |
| "loss": 0.2743, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 1.494296577946768, | |
| "grad_norm": 0.7096999351118269, | |
| "learning_rate": 3.654462852660423e-06, | |
| "loss": 0.282, | |
| "step": 786 | |
| }, | |
| { | |
| "epoch": 1.4961977186311788, | |
| "grad_norm": 0.6784469536845382, | |
| "learning_rate": 3.6288311862674885e-06, | |
| "loss": 0.2781, | |
| "step": 787 | |
| }, | |
| { | |
| "epoch": 1.4980988593155893, | |
| "grad_norm": 0.7067332921876521, | |
| "learning_rate": 3.6032697844110896e-06, | |
| "loss": 0.2953, | |
| "step": 788 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "grad_norm": 0.6794762738486263, | |
| "learning_rate": 3.5777789289955454e-06, | |
| "loss": 0.2722, | |
| "step": 789 | |
| }, | |
| { | |
| "epoch": 1.5019011406844105, | |
| "grad_norm": 0.6617943411881202, | |
| "learning_rate": 3.5523589011471592e-06, | |
| "loss": 0.2616, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 1.5038022813688214, | |
| "grad_norm": 0.7607357039080292, | |
| "learning_rate": 3.527009981211119e-06, | |
| "loss": 0.2862, | |
| "step": 791 | |
| }, | |
| { | |
| "epoch": 1.505703422053232, | |
| "grad_norm": 0.6676416660269762, | |
| "learning_rate": 3.5017324487483873e-06, | |
| "loss": 0.2754, | |
| "step": 792 | |
| }, | |
| { | |
| "epoch": 1.5076045627376424, | |
| "grad_norm": 0.7058231015708121, | |
| "learning_rate": 3.47652658253263e-06, | |
| "loss": 0.2829, | |
| "step": 793 | |
| }, | |
| { | |
| "epoch": 1.5095057034220534, | |
| "grad_norm": 0.7354768100700626, | |
| "learning_rate": 3.4513926605471504e-06, | |
| "loss": 0.2818, | |
| "step": 794 | |
| }, | |
| { | |
| "epoch": 1.5114068441064639, | |
| "grad_norm": 0.6894320399876503, | |
| "learning_rate": 3.4263309599818017e-06, | |
| "loss": 0.2691, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 1.5133079847908744, | |
| "grad_norm": 0.6857681326475394, | |
| "learning_rate": 3.4013417572299446e-06, | |
| "loss": 0.2734, | |
| "step": 796 | |
| }, | |
| { | |
| "epoch": 1.5152091254752853, | |
| "grad_norm": 0.6876950796507441, | |
| "learning_rate": 3.37642532788541e-06, | |
| "loss": 0.2652, | |
| "step": 797 | |
| }, | |
| { | |
| "epoch": 1.5171102661596958, | |
| "grad_norm": 0.7057871059292105, | |
| "learning_rate": 3.3515819467394184e-06, | |
| "loss": 0.2874, | |
| "step": 798 | |
| }, | |
| { | |
| "epoch": 1.5190114068441065, | |
| "grad_norm": 0.6862854720082124, | |
| "learning_rate": 3.326811887777607e-06, | |
| "loss": 0.2877, | |
| "step": 799 | |
| }, | |
| { | |
| "epoch": 1.5209125475285172, | |
| "grad_norm": 0.6993151684816991, | |
| "learning_rate": 3.3021154241769606e-06, | |
| "loss": 0.2805, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.5228136882129277, | |
| "grad_norm": 0.7439062231529465, | |
| "learning_rate": 3.2774928283028153e-06, | |
| "loss": 0.2816, | |
| "step": 801 | |
| }, | |
| { | |
| "epoch": 1.5247148288973384, | |
| "grad_norm": 0.7237727019958492, | |
| "learning_rate": 3.2529443717058693e-06, | |
| "loss": 0.2855, | |
| "step": 802 | |
| }, | |
| { | |
| "epoch": 1.5266159695817492, | |
| "grad_norm": 0.6800250840824873, | |
| "learning_rate": 3.228470325119164e-06, | |
| "loss": 0.2781, | |
| "step": 803 | |
| }, | |
| { | |
| "epoch": 1.5285171102661597, | |
| "grad_norm": 0.7110160968530334, | |
| "learning_rate": 3.20407095845511e-06, | |
| "loss": 0.2712, | |
| "step": 804 | |
| }, | |
| { | |
| "epoch": 1.5304182509505704, | |
| "grad_norm": 0.7725984461597548, | |
| "learning_rate": 3.179746540802506e-06, | |
| "loss": 0.2875, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 1.532319391634981, | |
| "grad_norm": 0.6714832734269215, | |
| "learning_rate": 3.155497340423588e-06, | |
| "loss": 0.2721, | |
| "step": 806 | |
| }, | |
| { | |
| "epoch": 1.5342205323193916, | |
| "grad_norm": 0.7147986794264566, | |
| "learning_rate": 3.1313236247510414e-06, | |
| "loss": 0.2863, | |
| "step": 807 | |
| }, | |
| { | |
| "epoch": 1.5361216730038023, | |
| "grad_norm": 0.8276470179241081, | |
| "learning_rate": 3.107225660385077e-06, | |
| "loss": 0.2999, | |
| "step": 808 | |
| }, | |
| { | |
| "epoch": 1.538022813688213, | |
| "grad_norm": 0.7297871686491798, | |
| "learning_rate": 3.0832037130904748e-06, | |
| "loss": 0.2821, | |
| "step": 809 | |
| }, | |
| { | |
| "epoch": 1.5399239543726235, | |
| "grad_norm": 0.7180724346503052, | |
| "learning_rate": 3.0592580477936606e-06, | |
| "loss": 0.2764, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 1.5418250950570342, | |
| "grad_norm": 0.6882836210270292, | |
| "learning_rate": 3.035388928579792e-06, | |
| "loss": 0.2811, | |
| "step": 811 | |
| }, | |
| { | |
| "epoch": 1.543726235741445, | |
| "grad_norm": 0.7095905102283543, | |
| "learning_rate": 3.011596618689825e-06, | |
| "loss": 0.2787, | |
| "step": 812 | |
| }, | |
| { | |
| "epoch": 1.5456273764258555, | |
| "grad_norm": 0.7077167552163011, | |
| "learning_rate": 2.9878813805176252e-06, | |
| "loss": 0.2661, | |
| "step": 813 | |
| }, | |
| { | |
| "epoch": 1.5475285171102662, | |
| "grad_norm": 0.7213686037703009, | |
| "learning_rate": 2.9642434756070793e-06, | |
| "loss": 0.2768, | |
| "step": 814 | |
| }, | |
| { | |
| "epoch": 1.549429657794677, | |
| "grad_norm": 0.6900286441180529, | |
| "learning_rate": 2.940683164649194e-06, | |
| "loss": 0.2759, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 1.5513307984790874, | |
| "grad_norm": 0.6765423062952957, | |
| "learning_rate": 2.9172007074792342e-06, | |
| "loss": 0.2739, | |
| "step": 816 | |
| }, | |
| { | |
| "epoch": 1.553231939163498, | |
| "grad_norm": 0.6535819672507082, | |
| "learning_rate": 2.8937963630738517e-06, | |
| "loss": 0.2623, | |
| "step": 817 | |
| }, | |
| { | |
| "epoch": 1.5551330798479088, | |
| "grad_norm": 0.6883817446952744, | |
| "learning_rate": 2.87047038954823e-06, | |
| "loss": 0.2766, | |
| "step": 818 | |
| }, | |
| { | |
| "epoch": 1.5570342205323193, | |
| "grad_norm": 0.718795255532413, | |
| "learning_rate": 2.8472230441532365e-06, | |
| "loss": 0.2706, | |
| "step": 819 | |
| }, | |
| { | |
| "epoch": 1.55893536121673, | |
| "grad_norm": 0.689746560147919, | |
| "learning_rate": 2.8240545832725963e-06, | |
| "loss": 0.2688, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 1.5608365019011408, | |
| "grad_norm": 0.6873985651638447, | |
| "learning_rate": 2.8009652624200436e-06, | |
| "loss": 0.2768, | |
| "step": 821 | |
| }, | |
| { | |
| "epoch": 1.5627376425855513, | |
| "grad_norm": 0.6453609615358062, | |
| "learning_rate": 2.7779553362365184e-06, | |
| "loss": 0.2599, | |
| "step": 822 | |
| }, | |
| { | |
| "epoch": 1.564638783269962, | |
| "grad_norm": 0.7432795627788588, | |
| "learning_rate": 2.755025058487364e-06, | |
| "loss": 0.287, | |
| "step": 823 | |
| }, | |
| { | |
| "epoch": 1.5665399239543727, | |
| "grad_norm": 0.707463057165011, | |
| "learning_rate": 2.7321746820595084e-06, | |
| "loss": 0.2768, | |
| "step": 824 | |
| }, | |
| { | |
| "epoch": 1.5684410646387832, | |
| "grad_norm": 0.7427407497603649, | |
| "learning_rate": 2.709404458958693e-06, | |
| "loss": 0.2898, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 1.570342205323194, | |
| "grad_norm": 0.6688849244886582, | |
| "learning_rate": 2.6867146403066833e-06, | |
| "loss": 0.2738, | |
| "step": 826 | |
| }, | |
| { | |
| "epoch": 1.5722433460076046, | |
| "grad_norm": 0.6703835294681526, | |
| "learning_rate": 2.6641054763385044e-06, | |
| "loss": 0.2622, | |
| "step": 827 | |
| }, | |
| { | |
| "epoch": 1.5741444866920151, | |
| "grad_norm": 0.7059337434360533, | |
| "learning_rate": 2.6415772163996845e-06, | |
| "loss": 0.2803, | |
| "step": 828 | |
| }, | |
| { | |
| "epoch": 1.5760456273764258, | |
| "grad_norm": 0.7352099927513953, | |
| "learning_rate": 2.619130108943494e-06, | |
| "loss": 0.2957, | |
| "step": 829 | |
| }, | |
| { | |
| "epoch": 1.5779467680608366, | |
| "grad_norm": 0.6623034031870155, | |
| "learning_rate": 2.5967644015282146e-06, | |
| "loss": 0.2563, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 1.579847908745247, | |
| "grad_norm": 0.6961628682819226, | |
| "learning_rate": 2.5744803408144026e-06, | |
| "loss": 0.2801, | |
| "step": 831 | |
| }, | |
| { | |
| "epoch": 1.5817490494296578, | |
| "grad_norm": 0.7203937899809977, | |
| "learning_rate": 2.5522781725621814e-06, | |
| "loss": 0.2917, | |
| "step": 832 | |
| }, | |
| { | |
| "epoch": 1.5836501901140685, | |
| "grad_norm": 0.6576215484604234, | |
| "learning_rate": 2.530158141628515e-06, | |
| "loss": 0.2658, | |
| "step": 833 | |
| }, | |
| { | |
| "epoch": 1.585551330798479, | |
| "grad_norm": 0.6711814097342703, | |
| "learning_rate": 2.508120491964512e-06, | |
| "loss": 0.2813, | |
| "step": 834 | |
| }, | |
| { | |
| "epoch": 1.5874524714828897, | |
| "grad_norm": 0.6952922674899649, | |
| "learning_rate": 2.486165466612751e-06, | |
| "loss": 0.2805, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 1.5893536121673004, | |
| "grad_norm": 0.6859550441606425, | |
| "learning_rate": 2.464293307704566e-06, | |
| "loss": 0.2692, | |
| "step": 836 | |
| }, | |
| { | |
| "epoch": 1.591254752851711, | |
| "grad_norm": 0.7819265965241207, | |
| "learning_rate": 2.4425042564574186e-06, | |
| "loss": 0.2768, | |
| "step": 837 | |
| }, | |
| { | |
| "epoch": 1.5931558935361216, | |
| "grad_norm": 0.666978352503959, | |
| "learning_rate": 2.4207985531722034e-06, | |
| "loss": 0.2719, | |
| "step": 838 | |
| }, | |
| { | |
| "epoch": 1.5950570342205324, | |
| "grad_norm": 0.7116942086459851, | |
| "learning_rate": 2.3991764372306113e-06, | |
| "loss": 0.2798, | |
| "step": 839 | |
| }, | |
| { | |
| "epoch": 1.5969581749049429, | |
| "grad_norm": 0.6313785007176683, | |
| "learning_rate": 2.377638147092497e-06, | |
| "loss": 0.2646, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 1.5988593155893536, | |
| "grad_norm": 0.688399500450701, | |
| "learning_rate": 2.3561839202932344e-06, | |
| "loss": 0.2743, | |
| "step": 841 | |
| }, | |
| { | |
| "epoch": 1.6007604562737643, | |
| "grad_norm": 0.6718823098844975, | |
| "learning_rate": 2.3348139934411008e-06, | |
| "loss": 0.275, | |
| "step": 842 | |
| }, | |
| { | |
| "epoch": 1.6026615969581748, | |
| "grad_norm": 0.6782964652352013, | |
| "learning_rate": 2.3135286022146785e-06, | |
| "loss": 0.2795, | |
| "step": 843 | |
| }, | |
| { | |
| "epoch": 1.6045627376425855, | |
| "grad_norm": 0.6174284999376568, | |
| "learning_rate": 2.292327981360245e-06, | |
| "loss": 0.251, | |
| "step": 844 | |
| }, | |
| { | |
| "epoch": 1.6064638783269962, | |
| "grad_norm": 0.7061117409230782, | |
| "learning_rate": 2.271212364689176e-06, | |
| "loss": 0.2725, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 1.6083650190114067, | |
| "grad_norm": 0.7258581064459089, | |
| "learning_rate": 2.2501819850753925e-06, | |
| "loss": 0.2828, | |
| "step": 846 | |
| }, | |
| { | |
| "epoch": 1.6102661596958177, | |
| "grad_norm": 0.7131222889098762, | |
| "learning_rate": 2.229237074452768e-06, | |
| "loss": 0.2805, | |
| "step": 847 | |
| }, | |
| { | |
| "epoch": 1.6121673003802282, | |
| "grad_norm": 0.6624053321723906, | |
| "learning_rate": 2.2083778638125796e-06, | |
| "loss": 0.2712, | |
| "step": 848 | |
| }, | |
| { | |
| "epoch": 1.6140684410646386, | |
| "grad_norm": 0.7010456337209631, | |
| "learning_rate": 2.1876045832009694e-06, | |
| "loss": 0.2605, | |
| "step": 849 | |
| }, | |
| { | |
| "epoch": 1.6159695817490496, | |
| "grad_norm": 0.7006105522115751, | |
| "learning_rate": 2.16691746171639e-06, | |
| "loss": 0.2845, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.61787072243346, | |
| "grad_norm": 0.6800519489510497, | |
| "learning_rate": 2.1463167275070863e-06, | |
| "loss": 0.2727, | |
| "step": 851 | |
| }, | |
| { | |
| "epoch": 1.6197718631178706, | |
| "grad_norm": 0.6693943096023052, | |
| "learning_rate": 2.125802607768588e-06, | |
| "loss": 0.268, | |
| "step": 852 | |
| }, | |
| { | |
| "epoch": 1.6216730038022815, | |
| "grad_norm": 0.7069575724524712, | |
| "learning_rate": 2.1053753287411895e-06, | |
| "loss": 0.2763, | |
| "step": 853 | |
| }, | |
| { | |
| "epoch": 1.623574144486692, | |
| "grad_norm": 0.6967188099421118, | |
| "learning_rate": 2.08503511570746e-06, | |
| "loss": 0.2772, | |
| "step": 854 | |
| }, | |
| { | |
| "epoch": 1.6254752851711025, | |
| "grad_norm": 0.6846467543440032, | |
| "learning_rate": 2.064782192989765e-06, | |
| "loss": 0.2632, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 1.6273764258555135, | |
| "grad_norm": 0.6833193040570177, | |
| "learning_rate": 2.0446167839477815e-06, | |
| "loss": 0.2682, | |
| "step": 856 | |
| }, | |
| { | |
| "epoch": 1.629277566539924, | |
| "grad_norm": 0.6948859005605714, | |
| "learning_rate": 2.0245391109760437e-06, | |
| "loss": 0.2826, | |
| "step": 857 | |
| }, | |
| { | |
| "epoch": 1.6311787072243344, | |
| "grad_norm": 0.7060100347376956, | |
| "learning_rate": 2.0045493955014915e-06, | |
| "loss": 0.268, | |
| "step": 858 | |
| }, | |
| { | |
| "epoch": 1.6330798479087454, | |
| "grad_norm": 0.6860672948720846, | |
| "learning_rate": 1.984647857981017e-06, | |
| "loss": 0.2809, | |
| "step": 859 | |
| }, | |
| { | |
| "epoch": 1.6349809885931559, | |
| "grad_norm": 0.7241686568183838, | |
| "learning_rate": 1.96483471789904e-06, | |
| "loss": 0.2853, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 1.6368821292775664, | |
| "grad_norm": 0.6679582016182153, | |
| "learning_rate": 1.9451101937650963e-06, | |
| "loss": 0.2755, | |
| "step": 861 | |
| }, | |
| { | |
| "epoch": 1.6387832699619773, | |
| "grad_norm": 0.6930404054883282, | |
| "learning_rate": 1.925474503111412e-06, | |
| "loss": 0.286, | |
| "step": 862 | |
| }, | |
| { | |
| "epoch": 1.6406844106463878, | |
| "grad_norm": 0.7142557583128137, | |
| "learning_rate": 1.905927862490512e-06, | |
| "loss": 0.2922, | |
| "step": 863 | |
| }, | |
| { | |
| "epoch": 1.6425855513307985, | |
| "grad_norm": 0.684147652430801, | |
| "learning_rate": 1.8864704874728346e-06, | |
| "loss": 0.2726, | |
| "step": 864 | |
| }, | |
| { | |
| "epoch": 1.6444866920152093, | |
| "grad_norm": 0.7181774180256446, | |
| "learning_rate": 1.8671025926443464e-06, | |
| "loss": 0.2809, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 1.6463878326996197, | |
| "grad_norm": 0.651008360784234, | |
| "learning_rate": 1.8478243916041882e-06, | |
| "loss": 0.2671, | |
| "step": 866 | |
| }, | |
| { | |
| "epoch": 1.6482889733840305, | |
| "grad_norm": 0.7300339753947043, | |
| "learning_rate": 1.828636096962304e-06, | |
| "loss": 0.2644, | |
| "step": 867 | |
| }, | |
| { | |
| "epoch": 1.6501901140684412, | |
| "grad_norm": 0.6956886570453094, | |
| "learning_rate": 1.8095379203371044e-06, | |
| "loss": 0.277, | |
| "step": 868 | |
| }, | |
| { | |
| "epoch": 1.6520912547528517, | |
| "grad_norm": 0.7269029992030692, | |
| "learning_rate": 1.7905300723531393e-06, | |
| "loss": 0.2753, | |
| "step": 869 | |
| }, | |
| { | |
| "epoch": 1.6539923954372624, | |
| "grad_norm": 0.6808357640605561, | |
| "learning_rate": 1.771612762638758e-06, | |
| "loss": 0.2688, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 1.6558935361216731, | |
| "grad_norm": 0.7230408613199361, | |
| "learning_rate": 1.7527861998238094e-06, | |
| "loss": 0.2784, | |
| "step": 871 | |
| }, | |
| { | |
| "epoch": 1.6577946768060836, | |
| "grad_norm": 0.6765387897380366, | |
| "learning_rate": 1.7340505915373495e-06, | |
| "loss": 0.2613, | |
| "step": 872 | |
| }, | |
| { | |
| "epoch": 1.6596958174904943, | |
| "grad_norm": 0.6934284795385596, | |
| "learning_rate": 1.7154061444053239e-06, | |
| "loss": 0.2796, | |
| "step": 873 | |
| }, | |
| { | |
| "epoch": 1.661596958174905, | |
| "grad_norm": 0.7249130776158051, | |
| "learning_rate": 1.6968530640483126e-06, | |
| "loss": 0.2758, | |
| "step": 874 | |
| }, | |
| { | |
| "epoch": 1.6634980988593155, | |
| "grad_norm": 0.7086669050540687, | |
| "learning_rate": 1.6783915550792652e-06, | |
| "loss": 0.2753, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 1.6653992395437263, | |
| "grad_norm": 0.7213797921734035, | |
| "learning_rate": 1.660021821101222e-06, | |
| "loss": 0.2853, | |
| "step": 876 | |
| }, | |
| { | |
| "epoch": 1.667300380228137, | |
| "grad_norm": 0.6811305103992509, | |
| "learning_rate": 1.6417440647050853e-06, | |
| "loss": 0.2704, | |
| "step": 877 | |
| }, | |
| { | |
| "epoch": 1.6692015209125475, | |
| "grad_norm": 0.6880318867185441, | |
| "learning_rate": 1.6235584874673848e-06, | |
| "loss": 0.2682, | |
| "step": 878 | |
| }, | |
| { | |
| "epoch": 1.6711026615969582, | |
| "grad_norm": 0.6711846721701289, | |
| "learning_rate": 1.6054652899480472e-06, | |
| "loss": 0.2602, | |
| "step": 879 | |
| }, | |
| { | |
| "epoch": 1.673003802281369, | |
| "grad_norm": 0.6256805181033526, | |
| "learning_rate": 1.587464671688187e-06, | |
| "loss": 0.2525, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 1.6749049429657794, | |
| "grad_norm": 0.6979584936812298, | |
| "learning_rate": 1.5695568312079156e-06, | |
| "loss": 0.2719, | |
| "step": 881 | |
| }, | |
| { | |
| "epoch": 1.6768060836501901, | |
| "grad_norm": 0.7223874643620409, | |
| "learning_rate": 1.5517419660041277e-06, | |
| "loss": 0.2655, | |
| "step": 882 | |
| }, | |
| { | |
| "epoch": 1.6787072243346008, | |
| "grad_norm": 0.7371134326771643, | |
| "learning_rate": 1.534020272548349e-06, | |
| "loss": 0.2825, | |
| "step": 883 | |
| }, | |
| { | |
| "epoch": 1.6806083650190113, | |
| "grad_norm": 0.727973375404682, | |
| "learning_rate": 1.5163919462845622e-06, | |
| "loss": 0.2836, | |
| "step": 884 | |
| }, | |
| { | |
| "epoch": 1.682509505703422, | |
| "grad_norm": 0.6995880442993847, | |
| "learning_rate": 1.4988571816270402e-06, | |
| "loss": 0.2813, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 1.6844106463878328, | |
| "grad_norm": 0.7329860431747285, | |
| "learning_rate": 1.4814161719582132e-06, | |
| "loss": 0.2768, | |
| "step": 886 | |
| }, | |
| { | |
| "epoch": 1.6863117870722433, | |
| "grad_norm": 0.7202315241044009, | |
| "learning_rate": 1.4640691096265358e-06, | |
| "loss": 0.2753, | |
| "step": 887 | |
| }, | |
| { | |
| "epoch": 1.688212927756654, | |
| "grad_norm": 0.6675099716861841, | |
| "learning_rate": 1.4468161859443609e-06, | |
| "loss": 0.2747, | |
| "step": 888 | |
| }, | |
| { | |
| "epoch": 1.6901140684410647, | |
| "grad_norm": 0.6835779954344514, | |
| "learning_rate": 1.4296575911858268e-06, | |
| "loss": 0.2599, | |
| "step": 889 | |
| }, | |
| { | |
| "epoch": 1.6920152091254752, | |
| "grad_norm": 0.7290396644687813, | |
| "learning_rate": 1.412593514584777e-06, | |
| "loss": 0.2752, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 1.693916349809886, | |
| "grad_norm": 0.7309918341615435, | |
| "learning_rate": 1.3956241443326423e-06, | |
| "loss": 0.271, | |
| "step": 891 | |
| }, | |
| { | |
| "epoch": 1.6958174904942966, | |
| "grad_norm": 0.7084062279964929, | |
| "learning_rate": 1.378749667576399e-06, | |
| "loss": 0.2841, | |
| "step": 892 | |
| }, | |
| { | |
| "epoch": 1.6977186311787071, | |
| "grad_norm": 0.6653801712801473, | |
| "learning_rate": 1.3619702704164783e-06, | |
| "loss": 0.2674, | |
| "step": 893 | |
| }, | |
| { | |
| "epoch": 1.6996197718631179, | |
| "grad_norm": 0.6864021885843341, | |
| "learning_rate": 1.3452861379047289e-06, | |
| "loss": 0.2757, | |
| "step": 894 | |
| }, | |
| { | |
| "epoch": 1.7015209125475286, | |
| "grad_norm": 0.6839104447710375, | |
| "learning_rate": 1.3286974540423747e-06, | |
| "loss": 0.2706, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 1.703422053231939, | |
| "grad_norm": 0.6871631247798478, | |
| "learning_rate": 1.3122044017779768e-06, | |
| "loss": 0.2812, | |
| "step": 896 | |
| }, | |
| { | |
| "epoch": 1.7053231939163498, | |
| "grad_norm": 0.6795857571366394, | |
| "learning_rate": 1.2958071630054214e-06, | |
| "loss": 0.271, | |
| "step": 897 | |
| }, | |
| { | |
| "epoch": 1.7072243346007605, | |
| "grad_norm": 0.7224198393402886, | |
| "learning_rate": 1.279505918561923e-06, | |
| "loss": 0.2886, | |
| "step": 898 | |
| }, | |
| { | |
| "epoch": 1.709125475285171, | |
| "grad_norm": 0.7665282228173506, | |
| "learning_rate": 1.2633008482260146e-06, | |
| "loss": 0.278, | |
| "step": 899 | |
| }, | |
| { | |
| "epoch": 1.7110266159695817, | |
| "grad_norm": 0.6960404148658614, | |
| "learning_rate": 1.2471921307155655e-06, | |
| "loss": 0.2681, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.7129277566539924, | |
| "grad_norm": 0.6801223747320676, | |
| "learning_rate": 1.2311799436858275e-06, | |
| "loss": 0.2786, | |
| "step": 901 | |
| }, | |
| { | |
| "epoch": 1.714828897338403, | |
| "grad_norm": 0.690284664721884, | |
| "learning_rate": 1.2152644637274603e-06, | |
| "loss": 0.2624, | |
| "step": 902 | |
| }, | |
| { | |
| "epoch": 1.7167300380228137, | |
| "grad_norm": 0.6662740407127687, | |
| "learning_rate": 1.1994458663645836e-06, | |
| "loss": 0.2731, | |
| "step": 903 | |
| }, | |
| { | |
| "epoch": 1.7186311787072244, | |
| "grad_norm": 0.7226182807040702, | |
| "learning_rate": 1.1837243260528542e-06, | |
| "loss": 0.2697, | |
| "step": 904 | |
| }, | |
| { | |
| "epoch": 1.7205323193916349, | |
| "grad_norm": 0.6678810398720528, | |
| "learning_rate": 1.168100016177528e-06, | |
| "loss": 0.2688, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 1.7224334600760456, | |
| "grad_norm": 0.6830780833569307, | |
| "learning_rate": 1.1525731090515536e-06, | |
| "loss": 0.2698, | |
| "step": 906 | |
| }, | |
| { | |
| "epoch": 1.7243346007604563, | |
| "grad_norm": 0.6921873948451761, | |
| "learning_rate": 1.137143775913675e-06, | |
| "loss": 0.2797, | |
| "step": 907 | |
| }, | |
| { | |
| "epoch": 1.7262357414448668, | |
| "grad_norm": 0.696637658225887, | |
| "learning_rate": 1.1218121869265365e-06, | |
| "loss": 0.2739, | |
| "step": 908 | |
| }, | |
| { | |
| "epoch": 1.7281368821292775, | |
| "grad_norm": 0.6342205072175778, | |
| "learning_rate": 1.1065785111748117e-06, | |
| "loss": 0.2577, | |
| "step": 909 | |
| }, | |
| { | |
| "epoch": 1.7300380228136882, | |
| "grad_norm": 0.68328843506268, | |
| "learning_rate": 1.0914429166633355e-06, | |
| "loss": 0.2684, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 1.7319391634980987, | |
| "grad_norm": 0.6990756077185901, | |
| "learning_rate": 1.076405570315252e-06, | |
| "loss": 0.2807, | |
| "step": 911 | |
| }, | |
| { | |
| "epoch": 1.7338403041825095, | |
| "grad_norm": 0.7297533428619984, | |
| "learning_rate": 1.0614666379701732e-06, | |
| "loss": 0.2689, | |
| "step": 912 | |
| }, | |
| { | |
| "epoch": 1.7357414448669202, | |
| "grad_norm": 0.685990621577933, | |
| "learning_rate": 1.046626284382356e-06, | |
| "loss": 0.2719, | |
| "step": 913 | |
| }, | |
| { | |
| "epoch": 1.7376425855513307, | |
| "grad_norm": 0.69197451759612, | |
| "learning_rate": 1.0318846732188737e-06, | |
| "loss": 0.2687, | |
| "step": 914 | |
| }, | |
| { | |
| "epoch": 1.7395437262357416, | |
| "grad_norm": 0.7078391946545975, | |
| "learning_rate": 1.017241967057816e-06, | |
| "loss": 0.27, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 1.741444866920152, | |
| "grad_norm": 0.6658586434356862, | |
| "learning_rate": 1.0026983273865055e-06, | |
| "loss": 0.2688, | |
| "step": 916 | |
| }, | |
| { | |
| "epoch": 1.7433460076045626, | |
| "grad_norm": 0.6684565129945329, | |
| "learning_rate": 9.882539145997027e-07, | |
| "loss": 0.2729, | |
| "step": 917 | |
| }, | |
| { | |
| "epoch": 1.7452471482889735, | |
| "grad_norm": 0.6802871499392203, | |
| "learning_rate": 9.739088879978409e-07, | |
| "loss": 0.2741, | |
| "step": 918 | |
| }, | |
| { | |
| "epoch": 1.747148288973384, | |
| "grad_norm": 0.6598796644097581, | |
| "learning_rate": 9.59663405785277e-07, | |
| "loss": 0.2564, | |
| "step": 919 | |
| }, | |
| { | |
| "epoch": 1.7490494296577945, | |
| "grad_norm": 0.7034173418020995, | |
| "learning_rate": 9.455176250685338e-07, | |
| "loss": 0.2737, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 1.7509505703422055, | |
| "grad_norm": 0.6732706850666013, | |
| "learning_rate": 9.314717018545838e-07, | |
| "loss": 0.2637, | |
| "step": 921 | |
| }, | |
| { | |
| "epoch": 1.752851711026616, | |
| "grad_norm": 0.6704945807396309, | |
| "learning_rate": 9.17525791049112e-07, | |
| "loss": 0.2551, | |
| "step": 922 | |
| }, | |
| { | |
| "epoch": 1.7547528517110265, | |
| "grad_norm": 0.6569516299908645, | |
| "learning_rate": 9.036800464548157e-07, | |
| "loss": 0.2618, | |
| "step": 923 | |
| }, | |
| { | |
| "epoch": 1.7566539923954374, | |
| "grad_norm": 0.7141387744751692, | |
| "learning_rate": 8.899346207697135e-07, | |
| "loss": 0.274, | |
| "step": 924 | |
| }, | |
| { | |
| "epoch": 1.758555133079848, | |
| "grad_norm": 0.7262920610713236, | |
| "learning_rate": 8.762896655854481e-07, | |
| "loss": 0.286, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 1.7604562737642584, | |
| "grad_norm": 0.7157782461356356, | |
| "learning_rate": 8.627453313856249e-07, | |
| "loss": 0.2683, | |
| "step": 926 | |
| }, | |
| { | |
| "epoch": 1.7623574144486693, | |
| "grad_norm": 0.6866047406619347, | |
| "learning_rate": 8.493017675441495e-07, | |
| "loss": 0.274, | |
| "step": 927 | |
| }, | |
| { | |
| "epoch": 1.7642585551330798, | |
| "grad_norm": 0.7398178361947803, | |
| "learning_rate": 8.359591223235785e-07, | |
| "loss": 0.2804, | |
| "step": 928 | |
| }, | |
| { | |
| "epoch": 1.7661596958174905, | |
| "grad_norm": 0.7412860435178176, | |
| "learning_rate": 8.227175428734868e-07, | |
| "loss": 0.2853, | |
| "step": 929 | |
| }, | |
| { | |
| "epoch": 1.7680608365019013, | |
| "grad_norm": 0.6865704141699449, | |
| "learning_rate": 8.095771752288451e-07, | |
| "loss": 0.2653, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 1.7699619771863118, | |
| "grad_norm": 0.6929724387549249, | |
| "learning_rate": 7.965381643084069e-07, | |
| "loss": 0.2676, | |
| "step": 931 | |
| }, | |
| { | |
| "epoch": 1.7718631178707225, | |
| "grad_norm": 0.7082536516916701, | |
| "learning_rate": 7.83600653913108e-07, | |
| "loss": 0.2854, | |
| "step": 932 | |
| }, | |
| { | |
| "epoch": 1.7737642585551332, | |
| "grad_norm": 0.6593484085037514, | |
| "learning_rate": 7.707647867244927e-07, | |
| "loss": 0.2693, | |
| "step": 933 | |
| }, | |
| { | |
| "epoch": 1.7756653992395437, | |
| "grad_norm": 0.7237364106951545, | |
| "learning_rate": 7.580307043031232e-07, | |
| "loss": 0.2725, | |
| "step": 934 | |
| }, | |
| { | |
| "epoch": 1.7775665399239544, | |
| "grad_norm": 0.7050552887848592, | |
| "learning_rate": 7.453985470870284e-07, | |
| "loss": 0.273, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 1.7794676806083651, | |
| "grad_norm": 0.6902777845270339, | |
| "learning_rate": 7.328684543901598e-07, | |
| "loss": 0.2717, | |
| "step": 936 | |
| }, | |
| { | |
| "epoch": 1.7813688212927756, | |
| "grad_norm": 0.6669367613540208, | |
| "learning_rate": 7.204405644008416e-07, | |
| "loss": 0.258, | |
| "step": 937 | |
| }, | |
| { | |
| "epoch": 1.7832699619771863, | |
| "grad_norm": 0.6784795181478801, | |
| "learning_rate": 7.081150141802518e-07, | |
| "loss": 0.2817, | |
| "step": 938 | |
| }, | |
| { | |
| "epoch": 1.785171102661597, | |
| "grad_norm": 0.6416014082322432, | |
| "learning_rate": 6.958919396609231e-07, | |
| "loss": 0.2677, | |
| "step": 939 | |
| }, | |
| { | |
| "epoch": 1.7870722433460076, | |
| "grad_norm": 0.7082020212204242, | |
| "learning_rate": 6.837714756452241e-07, | |
| "loss": 0.2699, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 1.7889733840304183, | |
| "grad_norm": 0.691360513084028, | |
| "learning_rate": 6.717537558038845e-07, | |
| "loss": 0.2752, | |
| "step": 941 | |
| }, | |
| { | |
| "epoch": 1.790874524714829, | |
| "grad_norm": 0.6952937643747015, | |
| "learning_rate": 6.598389126745209e-07, | |
| "loss": 0.2897, | |
| "step": 942 | |
| }, | |
| { | |
| "epoch": 1.7927756653992395, | |
| "grad_norm": 0.6888292256142322, | |
| "learning_rate": 6.480270776601682e-07, | |
| "loss": 0.2744, | |
| "step": 943 | |
| }, | |
| { | |
| "epoch": 1.7946768060836502, | |
| "grad_norm": 0.6645556232090724, | |
| "learning_rate": 6.36318381027835e-07, | |
| "loss": 0.2667, | |
| "step": 944 | |
| }, | |
| { | |
| "epoch": 1.796577946768061, | |
| "grad_norm": 0.6333609181113602, | |
| "learning_rate": 6.247129519070728e-07, | |
| "loss": 0.2643, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 1.7984790874524714, | |
| "grad_norm": 0.6889962949787356, | |
| "learning_rate": 6.132109182885382e-07, | |
| "loss": 0.2774, | |
| "step": 946 | |
| }, | |
| { | |
| "epoch": 1.8003802281368821, | |
| "grad_norm": 0.6885058593726265, | |
| "learning_rate": 6.018124070225928e-07, | |
| "loss": 0.2675, | |
| "step": 947 | |
| }, | |
| { | |
| "epoch": 1.8022813688212929, | |
| "grad_norm": 0.651209620369779, | |
| "learning_rate": 5.905175438178979e-07, | |
| "loss": 0.2705, | |
| "step": 948 | |
| }, | |
| { | |
| "epoch": 1.8041825095057034, | |
| "grad_norm": 0.6526836229204668, | |
| "learning_rate": 5.793264532400311e-07, | |
| "loss": 0.2694, | |
| "step": 949 | |
| }, | |
| { | |
| "epoch": 1.806083650190114, | |
| "grad_norm": 0.6665075425953116, | |
| "learning_rate": 5.68239258710116e-07, | |
| "loss": 0.2656, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.8079847908745248, | |
| "grad_norm": 0.6662079336068244, | |
| "learning_rate": 5.572560825034523e-07, | |
| "loss": 0.2716, | |
| "step": 951 | |
| }, | |
| { | |
| "epoch": 1.8098859315589353, | |
| "grad_norm": 0.7192838775374292, | |
| "learning_rate": 5.463770457481732e-07, | |
| "loss": 0.2826, | |
| "step": 952 | |
| }, | |
| { | |
| "epoch": 1.811787072243346, | |
| "grad_norm": 0.6613718297914037, | |
| "learning_rate": 5.35602268423906e-07, | |
| "loss": 0.2727, | |
| "step": 953 | |
| }, | |
| { | |
| "epoch": 1.8136882129277567, | |
| "grad_norm": 0.743728145638565, | |
| "learning_rate": 5.249318693604577e-07, | |
| "loss": 0.2835, | |
| "step": 954 | |
| }, | |
| { | |
| "epoch": 1.8155893536121672, | |
| "grad_norm": 0.730285321055072, | |
| "learning_rate": 5.143659662364931e-07, | |
| "loss": 0.2939, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 1.817490494296578, | |
| "grad_norm": 0.7384696029650311, | |
| "learning_rate": 5.039046755782417e-07, | |
| "loss": 0.2719, | |
| "step": 956 | |
| }, | |
| { | |
| "epoch": 1.8193916349809887, | |
| "grad_norm": 0.6622185073496315, | |
| "learning_rate": 4.935481127582131e-07, | |
| "loss": 0.255, | |
| "step": 957 | |
| }, | |
| { | |
| "epoch": 1.8212927756653992, | |
| "grad_norm": 0.7318926689035536, | |
| "learning_rate": 4.83296391993926e-07, | |
| "loss": 0.2907, | |
| "step": 958 | |
| }, | |
| { | |
| "epoch": 1.8231939163498099, | |
| "grad_norm": 0.6807368127689508, | |
| "learning_rate": 4.7314962634664616e-07, | |
| "loss": 0.2638, | |
| "step": 959 | |
| }, | |
| { | |
| "epoch": 1.8250950570342206, | |
| "grad_norm": 0.6517029769206409, | |
| "learning_rate": 4.631079277201389e-07, | |
| "loss": 0.2655, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 1.826996197718631, | |
| "grad_norm": 0.6747136764661026, | |
| "learning_rate": 4.5317140685943726e-07, | |
| "loss": 0.2785, | |
| "step": 961 | |
| }, | |
| { | |
| "epoch": 1.8288973384030418, | |
| "grad_norm": 0.6726537213543585, | |
| "learning_rate": 4.433401733496201e-07, | |
| "loss": 0.27, | |
| "step": 962 | |
| }, | |
| { | |
| "epoch": 1.8307984790874525, | |
| "grad_norm": 0.6935907127605666, | |
| "learning_rate": 4.3361433561460274e-07, | |
| "loss": 0.2798, | |
| "step": 963 | |
| }, | |
| { | |
| "epoch": 1.832699619771863, | |
| "grad_norm": 0.6947733683041682, | |
| "learning_rate": 4.2399400091594154e-07, | |
| "loss": 0.2732, | |
| "step": 964 | |
| }, | |
| { | |
| "epoch": 1.8346007604562737, | |
| "grad_norm": 0.6474799932073511, | |
| "learning_rate": 4.14479275351648e-07, | |
| "loss": 0.2593, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 1.8365019011406845, | |
| "grad_norm": 0.6660020887610706, | |
| "learning_rate": 4.0507026385502747e-07, | |
| "loss": 0.266, | |
| "step": 966 | |
| }, | |
| { | |
| "epoch": 1.838403041825095, | |
| "grad_norm": 0.6935244267938072, | |
| "learning_rate": 3.9576707019350903e-07, | |
| "loss": 0.2722, | |
| "step": 967 | |
| }, | |
| { | |
| "epoch": 1.8403041825095057, | |
| "grad_norm": 0.7330273768222302, | |
| "learning_rate": 3.865697969675164e-07, | |
| "loss": 0.2787, | |
| "step": 968 | |
| }, | |
| { | |
| "epoch": 1.8422053231939164, | |
| "grad_norm": 0.6848234341001256, | |
| "learning_rate": 3.7747854560931996e-07, | |
| "loss": 0.271, | |
| "step": 969 | |
| }, | |
| { | |
| "epoch": 1.8441064638783269, | |
| "grad_norm": 0.6985333108921234, | |
| "learning_rate": 3.684934163819309e-07, | |
| "loss": 0.2747, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 1.8460076045627376, | |
| "grad_norm": 0.6540811744472368, | |
| "learning_rate": 3.596145083779912e-07, | |
| "loss": 0.2629, | |
| "step": 971 | |
| }, | |
| { | |
| "epoch": 1.8479087452471483, | |
| "grad_norm": 0.6641726185041076, | |
| "learning_rate": 3.508419195186774e-07, | |
| "loss": 0.2548, | |
| "step": 972 | |
| }, | |
| { | |
| "epoch": 1.8498098859315588, | |
| "grad_norm": 0.6969951477015527, | |
| "learning_rate": 3.421757465526243e-07, | |
| "loss": 0.2719, | |
| "step": 973 | |
| }, | |
| { | |
| "epoch": 1.8517110266159695, | |
| "grad_norm": 0.7015463879347003, | |
| "learning_rate": 3.33616085054862e-07, | |
| "loss": 0.269, | |
| "step": 974 | |
| }, | |
| { | |
| "epoch": 1.8536121673003803, | |
| "grad_norm": 0.7048284146090317, | |
| "learning_rate": 3.2516302942574794e-07, | |
| "loss": 0.2753, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 1.8555133079847907, | |
| "grad_norm": 0.7203585003365163, | |
| "learning_rate": 3.1681667288994353e-07, | |
| "loss": 0.2734, | |
| "step": 976 | |
| }, | |
| { | |
| "epoch": 1.8574144486692015, | |
| "grad_norm": 0.6795822153876112, | |
| "learning_rate": 3.0857710749537585e-07, | |
| "loss": 0.2729, | |
| "step": 977 | |
| }, | |
| { | |
| "epoch": 1.8593155893536122, | |
| "grad_norm": 0.6524288908360266, | |
| "learning_rate": 3.0044442411222066e-07, | |
| "loss": 0.2613, | |
| "step": 978 | |
| }, | |
| { | |
| "epoch": 1.8612167300380227, | |
| "grad_norm": 0.6822377821072468, | |
| "learning_rate": 2.9241871243190555e-07, | |
| "loss": 0.2721, | |
| "step": 979 | |
| }, | |
| { | |
| "epoch": 1.8631178707224336, | |
| "grad_norm": 0.6745040881714924, | |
| "learning_rate": 2.845000609661208e-07, | |
| "loss": 0.2724, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 1.8650190114068441, | |
| "grad_norm": 0.6830792325920209, | |
| "learning_rate": 2.7668855704583997e-07, | |
| "loss": 0.2671, | |
| "step": 981 | |
| }, | |
| { | |
| "epoch": 1.8669201520912546, | |
| "grad_norm": 0.6734393863249917, | |
| "learning_rate": 2.689842868203563e-07, | |
| "loss": 0.2637, | |
| "step": 982 | |
| }, | |
| { | |
| "epoch": 1.8688212927756656, | |
| "grad_norm": 0.6876591909426577, | |
| "learning_rate": 2.6138733525633896e-07, | |
| "loss": 0.268, | |
| "step": 983 | |
| }, | |
| { | |
| "epoch": 1.870722433460076, | |
| "grad_norm": 0.7126401039046718, | |
| "learning_rate": 2.5389778613688744e-07, | |
| "loss": 0.2738, | |
| "step": 984 | |
| }, | |
| { | |
| "epoch": 1.8726235741444865, | |
| "grad_norm": 0.658656916628553, | |
| "learning_rate": 2.46515722060614e-07, | |
| "loss": 0.2707, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 1.8745247148288975, | |
| "grad_norm": 0.668443449569823, | |
| "learning_rate": 2.392412244407294e-07, | |
| "loss": 0.2526, | |
| "step": 986 | |
| }, | |
| { | |
| "epoch": 1.876425855513308, | |
| "grad_norm": 0.6542198600822328, | |
| "learning_rate": 2.3207437350414418e-07, | |
| "loss": 0.266, | |
| "step": 987 | |
| }, | |
| { | |
| "epoch": 1.8783269961977185, | |
| "grad_norm": 0.6889222120803568, | |
| "learning_rate": 2.2501524829059208e-07, | |
| "loss": 0.2735, | |
| "step": 988 | |
| }, | |
| { | |
| "epoch": 1.8802281368821294, | |
| "grad_norm": 0.7098450145309124, | |
| "learning_rate": 2.180639266517448e-07, | |
| "loss": 0.2714, | |
| "step": 989 | |
| }, | |
| { | |
| "epoch": 1.88212927756654, | |
| "grad_norm": 0.7030838160994648, | |
| "learning_rate": 2.1122048525036409e-07, | |
| "loss": 0.2737, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 1.8840304182509504, | |
| "grad_norm": 0.6988050633571449, | |
| "learning_rate": 2.0448499955945223e-07, | |
| "loss": 0.2749, | |
| "step": 991 | |
| }, | |
| { | |
| "epoch": 1.8859315589353614, | |
| "grad_norm": 0.6929504289567172, | |
| "learning_rate": 1.9785754386142164e-07, | |
| "loss": 0.2775, | |
| "step": 992 | |
| }, | |
| { | |
| "epoch": 1.8878326996197718, | |
| "grad_norm": 0.6742315122403425, | |
| "learning_rate": 1.9133819124727003e-07, | |
| "loss": 0.2732, | |
| "step": 993 | |
| }, | |
| { | |
| "epoch": 1.8897338403041823, | |
| "grad_norm": 0.7242558609050879, | |
| "learning_rate": 1.8492701361578326e-07, | |
| "loss": 0.272, | |
| "step": 994 | |
| }, | |
| { | |
| "epoch": 1.8916349809885933, | |
| "grad_norm": 0.6818034850543364, | |
| "learning_rate": 1.7862408167273472e-07, | |
| "loss": 0.2855, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 1.8935361216730038, | |
| "grad_norm": 0.6873083816269854, | |
| "learning_rate": 1.724294649301095e-07, | |
| "loss": 0.2761, | |
| "step": 996 | |
| }, | |
| { | |
| "epoch": 1.8954372623574145, | |
| "grad_norm": 0.7026496158030875, | |
| "learning_rate": 1.6634323170533928e-07, | |
| "loss": 0.2728, | |
| "step": 997 | |
| }, | |
| { | |
| "epoch": 1.8973384030418252, | |
| "grad_norm": 0.6625972688159788, | |
| "learning_rate": 1.6036544912054087e-07, | |
| "loss": 0.2627, | |
| "step": 998 | |
| }, | |
| { | |
| "epoch": 1.8992395437262357, | |
| "grad_norm": 0.6981551682588837, | |
| "learning_rate": 1.544961831017855e-07, | |
| "loss": 0.2785, | |
| "step": 999 | |
| }, | |
| { | |
| "epoch": 1.9011406844106464, | |
| "grad_norm": 0.703308913859999, | |
| "learning_rate": 1.487354983783673e-07, | |
| "loss": 0.268, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.9030418250950571, | |
| "grad_norm": 0.6769611901573057, | |
| "learning_rate": 1.430834584820895e-07, | |
| "loss": 0.2648, | |
| "step": 1001 | |
| }, | |
| { | |
| "epoch": 1.9049429657794676, | |
| "grad_norm": 0.684996304643279, | |
| "learning_rate": 1.375401257465625e-07, | |
| "loss": 0.2737, | |
| "step": 1002 | |
| }, | |
| { | |
| "epoch": 1.9068441064638784, | |
| "grad_norm": 0.6717778579358412, | |
| "learning_rate": 1.3210556130652031e-07, | |
| "loss": 0.2731, | |
| "step": 1003 | |
| }, | |
| { | |
| "epoch": 1.908745247148289, | |
| "grad_norm": 0.6754980475256993, | |
| "learning_rate": 1.2677982509714415e-07, | |
| "loss": 0.2614, | |
| "step": 1004 | |
| }, | |
| { | |
| "epoch": 1.9106463878326996, | |
| "grad_norm": 0.6913769040246529, | |
| "learning_rate": 1.2156297585339872e-07, | |
| "loss": 0.2655, | |
| "step": 1005 | |
| }, | |
| { | |
| "epoch": 1.9125475285171103, | |
| "grad_norm": 0.6917909571309567, | |
| "learning_rate": 1.1645507110938925e-07, | |
| "loss": 0.269, | |
| "step": 1006 | |
| }, | |
| { | |
| "epoch": 1.914448669201521, | |
| "grad_norm": 0.6812651911233605, | |
| "learning_rate": 1.1145616719772545e-07, | |
| "loss": 0.2731, | |
| "step": 1007 | |
| }, | |
| { | |
| "epoch": 1.9163498098859315, | |
| "grad_norm": 0.6792436836974647, | |
| "learning_rate": 1.0656631924889749e-07, | |
| "loss": 0.2736, | |
| "step": 1008 | |
| }, | |
| { | |
| "epoch": 1.9182509505703422, | |
| "grad_norm": 0.6905703763943478, | |
| "learning_rate": 1.0178558119067316e-07, | |
| "loss": 0.2744, | |
| "step": 1009 | |
| }, | |
| { | |
| "epoch": 1.920152091254753, | |
| "grad_norm": 0.6845029278713911, | |
| "learning_rate": 9.711400574749507e-08, | |
| "loss": 0.2641, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 1.9220532319391634, | |
| "grad_norm": 0.6622831183533047, | |
| "learning_rate": 9.255164443990994e-08, | |
| "loss": 0.2715, | |
| "step": 1011 | |
| }, | |
| { | |
| "epoch": 1.9239543726235742, | |
| "grad_norm": 0.6945608026730091, | |
| "learning_rate": 8.809854758399017e-08, | |
| "loss": 0.2661, | |
| "step": 1012 | |
| }, | |
| { | |
| "epoch": 1.9258555133079849, | |
| "grad_norm": 0.6635770331279389, | |
| "learning_rate": 8.375476429078543e-08, | |
| "loss": 0.2655, | |
| "step": 1013 | |
| }, | |
| { | |
| "epoch": 1.9277566539923954, | |
| "grad_norm": 0.6955787984366238, | |
| "learning_rate": 7.952034246577977e-08, | |
| "loss": 0.271, | |
| "step": 1014 | |
| }, | |
| { | |
| "epoch": 1.929657794676806, | |
| "grad_norm": 0.6795875745021076, | |
| "learning_rate": 7.539532880836087e-08, | |
| "loss": 0.2696, | |
| "step": 1015 | |
| }, | |
| { | |
| "epoch": 1.9315589353612168, | |
| "grad_norm": 0.6370112632067916, | |
| "learning_rate": 7.137976881130826e-08, | |
| "loss": 0.2621, | |
| "step": 1016 | |
| }, | |
| { | |
| "epoch": 1.9334600760456273, | |
| "grad_norm": 0.6946651821074374, | |
| "learning_rate": 6.747370676028819e-08, | |
| "loss": 0.2708, | |
| "step": 1017 | |
| }, | |
| { | |
| "epoch": 1.935361216730038, | |
| "grad_norm": 0.7018166576391898, | |
| "learning_rate": 6.367718573336845e-08, | |
| "loss": 0.2731, | |
| "step": 1018 | |
| }, | |
| { | |
| "epoch": 1.9372623574144487, | |
| "grad_norm": 0.718014631323638, | |
| "learning_rate": 5.999024760054095e-08, | |
| "loss": 0.2667, | |
| "step": 1019 | |
| }, | |
| { | |
| "epoch": 1.9391634980988592, | |
| "grad_norm": 0.6657768591273471, | |
| "learning_rate": 5.641293302326323e-08, | |
| "loss": 0.2611, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 1.94106463878327, | |
| "grad_norm": 0.6737387216231969, | |
| "learning_rate": 5.2945281454003236e-08, | |
| "loss": 0.2689, | |
| "step": 1021 | |
| }, | |
| { | |
| "epoch": 1.9429657794676807, | |
| "grad_norm": 0.6790679081811256, | |
| "learning_rate": 4.958733113581415e-08, | |
| "loss": 0.2717, | |
| "step": 1022 | |
| }, | |
| { | |
| "epoch": 1.9448669201520912, | |
| "grad_norm": 0.7172923641541281, | |
| "learning_rate": 4.6339119101902475e-08, | |
| "loss": 0.2655, | |
| "step": 1023 | |
| }, | |
| { | |
| "epoch": 1.946768060836502, | |
| "grad_norm": 0.7094728977446307, | |
| "learning_rate": 4.320068117522835e-08, | |
| "loss": 0.2789, | |
| "step": 1024 | |
| }, | |
| { | |
| "epoch": 1.9486692015209126, | |
| "grad_norm": 0.6849762793320817, | |
| "learning_rate": 4.0172051968101474e-08, | |
| "loss": 0.2719, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 1.950570342205323, | |
| "grad_norm": 0.6505325493943667, | |
| "learning_rate": 3.7253264881809137e-08, | |
| "loss": 0.2634, | |
| "step": 1026 | |
| }, | |
| { | |
| "epoch": 1.9524714828897338, | |
| "grad_norm": 0.6778001650205526, | |
| "learning_rate": 3.4444352106242086e-08, | |
| "loss": 0.2669, | |
| "step": 1027 | |
| }, | |
| { | |
| "epoch": 1.9543726235741445, | |
| "grad_norm": 0.6681791224630136, | |
| "learning_rate": 3.174534461953593e-08, | |
| "loss": 0.2669, | |
| "step": 1028 | |
| }, | |
| { | |
| "epoch": 1.956273764258555, | |
| "grad_norm": 0.6789406485342497, | |
| "learning_rate": 2.915627218774142e-08, | |
| "loss": 0.2689, | |
| "step": 1029 | |
| }, | |
| { | |
| "epoch": 1.9581749049429658, | |
| "grad_norm": 0.7087378560635103, | |
| "learning_rate": 2.667716336448356e-08, | |
| "loss": 0.2776, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 1.9600760456273765, | |
| "grad_norm": 0.6462195686809685, | |
| "learning_rate": 2.430804549065302e-08, | |
| "loss": 0.2665, | |
| "step": 1031 | |
| }, | |
| { | |
| "epoch": 1.961977186311787, | |
| "grad_norm": 0.6894488274293517, | |
| "learning_rate": 2.2048944694104123e-08, | |
| "loss": 0.2652, | |
| "step": 1032 | |
| }, | |
| { | |
| "epoch": 1.9638783269961977, | |
| "grad_norm": 0.7092138019551788, | |
| "learning_rate": 1.989988588936509e-08, | |
| "loss": 0.2856, | |
| "step": 1033 | |
| }, | |
| { | |
| "epoch": 1.9657794676806084, | |
| "grad_norm": 0.6716340764626926, | |
| "learning_rate": 1.7860892777367133e-08, | |
| "loss": 0.264, | |
| "step": 1034 | |
| }, | |
| { | |
| "epoch": 1.967680608365019, | |
| "grad_norm": 0.6761036665802346, | |
| "learning_rate": 1.5931987845176912e-08, | |
| "loss": 0.2633, | |
| "step": 1035 | |
| }, | |
| { | |
| "epoch": 1.9695817490494296, | |
| "grad_norm": 0.6584647463348391, | |
| "learning_rate": 1.411319236575337e-08, | |
| "loss": 0.2699, | |
| "step": 1036 | |
| }, | |
| { | |
| "epoch": 1.9714828897338403, | |
| "grad_norm": 0.7351907237798092, | |
| "learning_rate": 1.2404526397711281e-08, | |
| "loss": 0.2833, | |
| "step": 1037 | |
| }, | |
| { | |
| "epoch": 1.9733840304182508, | |
| "grad_norm": 0.6645293108499867, | |
| "learning_rate": 1.0806008785100297e-08, | |
| "loss": 0.2607, | |
| "step": 1038 | |
| }, | |
| { | |
| "epoch": 1.9752851711026616, | |
| "grad_norm": 0.7251892020513389, | |
| "learning_rate": 9.317657157197347e-09, | |
| "loss": 0.2828, | |
| "step": 1039 | |
| }, | |
| { | |
| "epoch": 1.9771863117870723, | |
| "grad_norm": 0.7308540818805154, | |
| "learning_rate": 7.93948792831234e-09, | |
| "loss": 0.2798, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 1.9790874524714828, | |
| "grad_norm": 0.6878381821178471, | |
| "learning_rate": 6.671516297606095e-09, | |
| "loss": 0.2764, | |
| "step": 1041 | |
| }, | |
| { | |
| "epoch": 1.9809885931558935, | |
| "grad_norm": 0.7087513629127422, | |
| "learning_rate": 5.513756248924917e-09, | |
| "loss": 0.2704, | |
| "step": 1042 | |
| }, | |
| { | |
| "epoch": 1.9828897338403042, | |
| "grad_norm": 0.6926135256320923, | |
| "learning_rate": 4.466220550641831e-09, | |
| "loss": 0.2615, | |
| "step": 1043 | |
| }, | |
| { | |
| "epoch": 1.9847908745247147, | |
| "grad_norm": 0.6973776066035858, | |
| "learning_rate": 3.528920755523357e-09, | |
| "loss": 0.2765, | |
| "step": 1044 | |
| }, | |
| { | |
| "epoch": 1.9866920152091256, | |
| "grad_norm": 0.704517663600957, | |
| "learning_rate": 2.701867200592956e-09, | |
| "loss": 0.273, | |
| "step": 1045 | |
| }, | |
| { | |
| "epoch": 1.9885931558935361, | |
| "grad_norm": 0.6639585086446169, | |
| "learning_rate": 1.9850690070266633e-09, | |
| "loss": 0.2727, | |
| "step": 1046 | |
| }, | |
| { | |
| "epoch": 1.9904942965779466, | |
| "grad_norm": 0.6772753301297887, | |
| "learning_rate": 1.378534080042071e-09, | |
| "loss": 0.2719, | |
| "step": 1047 | |
| }, | |
| { | |
| "epoch": 1.9923954372623576, | |
| "grad_norm": 0.6967246165493637, | |
| "learning_rate": 8.822691088195001e-10, | |
| "loss": 0.2666, | |
| "step": 1048 | |
| }, | |
| { | |
| "epoch": 1.994296577946768, | |
| "grad_norm": 0.6903412472660575, | |
| "learning_rate": 4.962795664265052e-10, | |
| "loss": 0.2694, | |
| "step": 1049 | |
| }, | |
| { | |
| "epoch": 1.9961977186311786, | |
| "grad_norm": 0.695432186970267, | |
| "learning_rate": 2.2056970975459223e-10, | |
| "loss": 0.2642, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.9980988593155895, | |
| "grad_norm": 0.7176049552636734, | |
| "learning_rate": 5.514257947369928e-11, | |
| "loss": 0.291, | |
| "step": 1051 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "grad_norm": 0.6471020809475667, | |
| "learning_rate": 0.0, | |
| "loss": 0.2671, | |
| "step": 1052 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "step": 1052, | |
| "total_flos": 96805854412800.0, | |
| "train_loss": 0.35639558661686604, | |
| "train_runtime": 3414.7262, | |
| "train_samples_per_second": 39.434, | |
| "train_steps_per_second": 0.308 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 1052, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 2, | |
| "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": 96805854412800.0, | |
| "train_batch_size": 2, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |