| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.998563906175203, | |
| "eval_steps": 500, | |
| "global_step": 522, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0057443752991862135, | |
| "grad_norm": 2.7811572551727295, | |
| "learning_rate": 1.886792452830189e-07, | |
| "loss": 0.7552, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.011488750598372427, | |
| "grad_norm": 2.7668187618255615, | |
| "learning_rate": 3.773584905660378e-07, | |
| "loss": 0.7816, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.01723312589755864, | |
| "grad_norm": 2.714698314666748, | |
| "learning_rate": 5.660377358490567e-07, | |
| "loss": 0.8084, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.022977501196744854, | |
| "grad_norm": 2.6264641284942627, | |
| "learning_rate": 7.547169811320755e-07, | |
| "loss": 0.7774, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.028721876495931067, | |
| "grad_norm": 2.5013437271118164, | |
| "learning_rate": 9.433962264150944e-07, | |
| "loss": 0.7726, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.03446625179511728, | |
| "grad_norm": 2.7753398418426514, | |
| "learning_rate": 1.1320754716981133e-06, | |
| "loss": 0.7708, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.040210627094303494, | |
| "grad_norm": 2.3244540691375732, | |
| "learning_rate": 1.3207547169811322e-06, | |
| "loss": 0.7759, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.04595500239348971, | |
| "grad_norm": 1.9560602903366089, | |
| "learning_rate": 1.509433962264151e-06, | |
| "loss": 0.7493, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.05169937769267592, | |
| "grad_norm": 1.9135451316833496, | |
| "learning_rate": 1.6981132075471698e-06, | |
| "loss": 0.7216, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.057443752991862135, | |
| "grad_norm": 1.8730158805847168, | |
| "learning_rate": 1.8867924528301889e-06, | |
| "loss": 0.7732, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.06318812829104835, | |
| "grad_norm": 1.520505666732788, | |
| "learning_rate": 2.075471698113208e-06, | |
| "loss": 0.7432, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.06893250359023456, | |
| "grad_norm": 1.342366099357605, | |
| "learning_rate": 2.2641509433962266e-06, | |
| "loss": 0.6762, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.07467687888942078, | |
| "grad_norm": 1.3205207586288452, | |
| "learning_rate": 2.4528301886792453e-06, | |
| "loss": 0.7395, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.08042125418860699, | |
| "grad_norm": 1.3457564115524292, | |
| "learning_rate": 2.6415094339622644e-06, | |
| "loss": 0.7279, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.0861656294877932, | |
| "grad_norm": 1.1626056432724, | |
| "learning_rate": 2.830188679245283e-06, | |
| "loss": 0.7218, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.09191000478697942, | |
| "grad_norm": 1.1689456701278687, | |
| "learning_rate": 3.018867924528302e-06, | |
| "loss": 0.7202, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.09765438008616563, | |
| "grad_norm": 1.100240707397461, | |
| "learning_rate": 3.207547169811321e-06, | |
| "loss": 0.6957, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.10339875538535184, | |
| "grad_norm": 0.9969179630279541, | |
| "learning_rate": 3.3962264150943395e-06, | |
| "loss": 0.695, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.10914313068453806, | |
| "grad_norm": 0.8645837306976318, | |
| "learning_rate": 3.5849056603773586e-06, | |
| "loss": 0.6393, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.11488750598372427, | |
| "grad_norm": 0.726235568523407, | |
| "learning_rate": 3.7735849056603777e-06, | |
| "loss": 0.6471, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.12063188128291048, | |
| "grad_norm": 0.7744720578193665, | |
| "learning_rate": 3.962264150943396e-06, | |
| "loss": 0.6303, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.1263762565820967, | |
| "grad_norm": 0.8271118998527527, | |
| "learning_rate": 4.150943396226416e-06, | |
| "loss": 0.6215, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.13212063188128292, | |
| "grad_norm": 0.7437446713447571, | |
| "learning_rate": 4.339622641509435e-06, | |
| "loss": 0.5848, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.13786500718046912, | |
| "grad_norm": 0.6880975365638733, | |
| "learning_rate": 4.528301886792453e-06, | |
| "loss": 0.6208, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.14360938247965535, | |
| "grad_norm": 0.6708999872207642, | |
| "learning_rate": 4.716981132075472e-06, | |
| "loss": 0.6067, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.14935375777884155, | |
| "grad_norm": 0.5787381529808044, | |
| "learning_rate": 4.905660377358491e-06, | |
| "loss": 0.5981, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.15509813307802778, | |
| "grad_norm": 0.6275501847267151, | |
| "learning_rate": 5.09433962264151e-06, | |
| "loss": 0.6233, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.16084250837721398, | |
| "grad_norm": 0.6212226748466492, | |
| "learning_rate": 5.283018867924529e-06, | |
| "loss": 0.5856, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.1665868836764002, | |
| "grad_norm": 0.5922460556030273, | |
| "learning_rate": 5.4716981132075475e-06, | |
| "loss": 0.5698, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.1723312589755864, | |
| "grad_norm": 0.5896409749984741, | |
| "learning_rate": 5.660377358490566e-06, | |
| "loss": 0.5768, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.17807563427477263, | |
| "grad_norm": 0.5254840850830078, | |
| "learning_rate": 5.849056603773585e-06, | |
| "loss": 0.539, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.18382000957395883, | |
| "grad_norm": 0.5485825538635254, | |
| "learning_rate": 6.037735849056604e-06, | |
| "loss": 0.5176, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.18956438487314506, | |
| "grad_norm": 0.5224868655204773, | |
| "learning_rate": 6.226415094339623e-06, | |
| "loss": 0.5541, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.19530876017233126, | |
| "grad_norm": 0.521843433380127, | |
| "learning_rate": 6.415094339622642e-06, | |
| "loss": 0.5707, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.20105313547151749, | |
| "grad_norm": 0.5119784474372864, | |
| "learning_rate": 6.60377358490566e-06, | |
| "loss": 0.5567, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.20679751077070369, | |
| "grad_norm": 0.5173502564430237, | |
| "learning_rate": 6.792452830188679e-06, | |
| "loss": 0.5481, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.2125418860698899, | |
| "grad_norm": 0.41040942072868347, | |
| "learning_rate": 6.981132075471699e-06, | |
| "loss": 0.5361, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.2182862613690761, | |
| "grad_norm": 0.4279431104660034, | |
| "learning_rate": 7.169811320754717e-06, | |
| "loss": 0.5454, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.22403063666826234, | |
| "grad_norm": 0.4288845956325531, | |
| "learning_rate": 7.358490566037736e-06, | |
| "loss": 0.5459, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.22977501196744854, | |
| "grad_norm": 0.4141804277896881, | |
| "learning_rate": 7.5471698113207555e-06, | |
| "loss": 0.5267, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.23551938726663477, | |
| "grad_norm": 4.088698387145996, | |
| "learning_rate": 7.735849056603775e-06, | |
| "loss": 0.5617, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.24126376256582097, | |
| "grad_norm": 0.40810075402259827, | |
| "learning_rate": 7.924528301886793e-06, | |
| "loss": 0.5066, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.2470081378650072, | |
| "grad_norm": 0.4773496687412262, | |
| "learning_rate": 8.113207547169812e-06, | |
| "loss": 0.5375, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.2527525131641934, | |
| "grad_norm": 0.3834821581840515, | |
| "learning_rate": 8.301886792452832e-06, | |
| "loss": 0.5502, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.2584968884633796, | |
| "grad_norm": 0.38557955622673035, | |
| "learning_rate": 8.49056603773585e-06, | |
| "loss": 0.5338, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.26424126376256585, | |
| "grad_norm": 0.3510483503341675, | |
| "learning_rate": 8.67924528301887e-06, | |
| "loss": 0.4924, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.26998563906175205, | |
| "grad_norm": 0.3511549234390259, | |
| "learning_rate": 8.867924528301887e-06, | |
| "loss": 0.5186, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.27573001436093825, | |
| "grad_norm": 0.41062384843826294, | |
| "learning_rate": 9.056603773584907e-06, | |
| "loss": 0.5021, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.28147438966012445, | |
| "grad_norm": 0.3798637092113495, | |
| "learning_rate": 9.245283018867926e-06, | |
| "loss": 0.5338, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.2872187649593107, | |
| "grad_norm": 0.35235193371772766, | |
| "learning_rate": 9.433962264150944e-06, | |
| "loss": 0.4922, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.2929631402584969, | |
| "grad_norm": 0.3465921878814697, | |
| "learning_rate": 9.622641509433963e-06, | |
| "loss": 0.4825, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.2987075155576831, | |
| "grad_norm": 0.36125242710113525, | |
| "learning_rate": 9.811320754716981e-06, | |
| "loss": 0.5178, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.3044518908568693, | |
| "grad_norm": 0.3282877504825592, | |
| "learning_rate": 1e-05, | |
| "loss": 0.494, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.31019626615605556, | |
| "grad_norm": 0.3224699795246124, | |
| "learning_rate": 9.999887825938495e-06, | |
| "loss": 0.4878, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.31594064145524176, | |
| "grad_norm": 0.36641696095466614, | |
| "learning_rate": 9.999551308787183e-06, | |
| "loss": 0.4905, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.32168501675442795, | |
| "grad_norm": 0.321207195520401, | |
| "learning_rate": 9.998990463645464e-06, | |
| "loss": 0.4956, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.32742939205361415, | |
| "grad_norm": 0.32767999172210693, | |
| "learning_rate": 9.998205315678248e-06, | |
| "loss": 0.5046, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.3331737673528004, | |
| "grad_norm": 0.3131692409515381, | |
| "learning_rate": 9.997195900114833e-06, | |
| "loss": 0.4961, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.3389181426519866, | |
| "grad_norm": 0.304837167263031, | |
| "learning_rate": 9.995962262247314e-06, | |
| "loss": 0.4712, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.3446625179511728, | |
| "grad_norm": 0.3071947395801544, | |
| "learning_rate": 9.994504457428557e-06, | |
| "loss": 0.4851, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.350406893250359, | |
| "grad_norm": 0.344100683927536, | |
| "learning_rate": 9.99282255106972e-06, | |
| "loss": 0.4947, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.35615126854954526, | |
| "grad_norm": 0.32712873816490173, | |
| "learning_rate": 9.99091661863731e-06, | |
| "loss": 0.502, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.36189564384873146, | |
| "grad_norm": 0.2983171045780182, | |
| "learning_rate": 9.988786745649798e-06, | |
| "loss": 0.4892, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.36764001914791766, | |
| "grad_norm": 0.45408669114112854, | |
| "learning_rate": 9.986433027673786e-06, | |
| "loss": 0.4738, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.37338439444710386, | |
| "grad_norm": 0.34534087777137756, | |
| "learning_rate": 9.983855570319716e-06, | |
| "loss": 0.4915, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.3791287697462901, | |
| "grad_norm": 0.32527977228164673, | |
| "learning_rate": 9.981054489237132e-06, | |
| "loss": 0.4849, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.3848731450454763, | |
| "grad_norm": 0.2880174219608307, | |
| "learning_rate": 9.978029910109491e-06, | |
| "loss": 0.4474, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.3906175203446625, | |
| "grad_norm": 0.3428208529949188, | |
| "learning_rate": 9.974781968648523e-06, | |
| "loss": 0.4731, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.3963618956438487, | |
| "grad_norm": 0.33169832825660706, | |
| "learning_rate": 9.971310810588141e-06, | |
| "loss": 0.4951, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.40210627094303497, | |
| "grad_norm": 0.3041759729385376, | |
| "learning_rate": 9.967616591677906e-06, | |
| "loss": 0.4278, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.40785064624222117, | |
| "grad_norm": 0.3041941225528717, | |
| "learning_rate": 9.963699477676031e-06, | |
| "loss": 0.4755, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.41359502154140737, | |
| "grad_norm": 0.30077141523361206, | |
| "learning_rate": 9.959559644341954e-06, | |
| "loss": 0.4572, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.41933939684059357, | |
| "grad_norm": 0.36518850922584534, | |
| "learning_rate": 9.95519727742844e-06, | |
| "loss": 0.4752, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.4250837721397798, | |
| "grad_norm": 0.2795243263244629, | |
| "learning_rate": 9.950612572673255e-06, | |
| "loss": 0.4759, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.430828147438966, | |
| "grad_norm": 0.30531245470046997, | |
| "learning_rate": 9.945805735790383e-06, | |
| "loss": 0.4897, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.4365725227381522, | |
| "grad_norm": 0.31547942757606506, | |
| "learning_rate": 9.940776982460787e-06, | |
| "loss": 0.466, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.4423168980373384, | |
| "grad_norm": 0.276068776845932, | |
| "learning_rate": 9.935526538322744e-06, | |
| "loss": 0.4882, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.4480612733365247, | |
| "grad_norm": 0.34043920040130615, | |
| "learning_rate": 9.930054638961709e-06, | |
| "loss": 0.4685, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.4538056486357109, | |
| "grad_norm": 0.2898786664009094, | |
| "learning_rate": 9.924361529899754e-06, | |
| "loss": 0.4756, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.4595500239348971, | |
| "grad_norm": 0.270378977060318, | |
| "learning_rate": 9.918447466584545e-06, | |
| "loss": 0.4599, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.4652943992340833, | |
| "grad_norm": 0.35345959663391113, | |
| "learning_rate": 9.91231271437788e-06, | |
| "loss": 0.4629, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.47103877453326953, | |
| "grad_norm": 0.3456457555294037, | |
| "learning_rate": 9.905957548543794e-06, | |
| "loss": 0.481, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.47678314983245573, | |
| "grad_norm": 0.2923058271408081, | |
| "learning_rate": 9.899382254236186e-06, | |
| "loss": 0.4679, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.48252752513164193, | |
| "grad_norm": 0.3531160354614258, | |
| "learning_rate": 9.892587126486046e-06, | |
| "loss": 0.4906, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.48827190043082813, | |
| "grad_norm": 0.36479318141937256, | |
| "learning_rate": 9.885572470188207e-06, | |
| "loss": 0.4569, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.4940162757300144, | |
| "grad_norm": 0.33483925461769104, | |
| "learning_rate": 9.878338600087658e-06, | |
| "loss": 0.4668, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.4997606510292006, | |
| "grad_norm": 0.30246540904045105, | |
| "learning_rate": 9.87088584076544e-06, | |
| "loss": 0.4847, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.5055050263283868, | |
| "grad_norm": 0.3419553339481354, | |
| "learning_rate": 9.863214526624065e-06, | |
| "loss": 0.4698, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.511249401627573, | |
| "grad_norm": 0.30689287185668945, | |
| "learning_rate": 9.85532500187252e-06, | |
| "loss": 0.4646, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.5169937769267592, | |
| "grad_norm": 0.36106929183006287, | |
| "learning_rate": 9.847217620510815e-06, | |
| "loss": 0.4746, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.5227381522259454, | |
| "grad_norm": 0.29505378007888794, | |
| "learning_rate": 9.83889274631411e-06, | |
| "loss": 0.4571, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.5284825275251317, | |
| "grad_norm": 0.29619649052619934, | |
| "learning_rate": 9.830350752816386e-06, | |
| "loss": 0.4578, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.5342269028243178, | |
| "grad_norm": 0.3188607692718506, | |
| "learning_rate": 9.821592023293686e-06, | |
| "loss": 0.4508, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.5399712781235041, | |
| "grad_norm": 0.2994546592235565, | |
| "learning_rate": 9.81261695074691e-06, | |
| "loss": 0.4774, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.5457156534226902, | |
| "grad_norm": 0.27736401557922363, | |
| "learning_rate": 9.803425937884202e-06, | |
| "loss": 0.4627, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.5514600287218765, | |
| "grad_norm": 0.313815176486969, | |
| "learning_rate": 9.794019397102852e-06, | |
| "loss": 0.4753, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.5572044040210627, | |
| "grad_norm": 0.3198351562023163, | |
| "learning_rate": 9.784397750470818e-06, | |
| "loss": 0.45, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.5629487793202489, | |
| "grad_norm": 0.3304007351398468, | |
| "learning_rate": 9.774561429707769e-06, | |
| "loss": 0.4879, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.5686931546194351, | |
| "grad_norm": 0.2905180752277374, | |
| "learning_rate": 9.764510876165727e-06, | |
| "loss": 0.448, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.5744375299186214, | |
| "grad_norm": 0.31816208362579346, | |
| "learning_rate": 9.754246540809257e-06, | |
| "loss": 0.4483, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.5801819052178075, | |
| "grad_norm": 0.3271932005882263, | |
| "learning_rate": 9.743768884195233e-06, | |
| "loss": 0.4772, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.5859262805169938, | |
| "grad_norm": 0.3400787115097046, | |
| "learning_rate": 9.733078376452172e-06, | |
| "loss": 0.4762, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.59167065581618, | |
| "grad_norm": 0.3152044415473938, | |
| "learning_rate": 9.722175497259145e-06, | |
| "loss": 0.4625, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.5974150311153662, | |
| "grad_norm": 0.29500487446784973, | |
| "learning_rate": 9.71106073582425e-06, | |
| "loss": 0.4513, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.6031594064145525, | |
| "grad_norm": 0.30806249380111694, | |
| "learning_rate": 9.699734590862655e-06, | |
| "loss": 0.4493, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.6089037817137386, | |
| "grad_norm": 0.32135871052742004, | |
| "learning_rate": 9.688197570574238e-06, | |
| "loss": 0.4687, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.6146481570129249, | |
| "grad_norm": 0.2976573705673218, | |
| "learning_rate": 9.676450192620767e-06, | |
| "loss": 0.464, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.6203925323121111, | |
| "grad_norm": 0.29863908886909485, | |
| "learning_rate": 9.66449298410268e-06, | |
| "loss": 0.4816, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.6261369076112973, | |
| "grad_norm": 0.31702297925949097, | |
| "learning_rate": 9.652326481535434e-06, | |
| "loss": 0.4278, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.6318812829104835, | |
| "grad_norm": 0.3526081442832947, | |
| "learning_rate": 9.639951230825433e-06, | |
| "loss": 0.4931, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.6376256582096697, | |
| "grad_norm": 0.33451569080352783, | |
| "learning_rate": 9.62736778724553e-06, | |
| "loss": 0.475, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.6433700335088559, | |
| "grad_norm": 0.33537557721138, | |
| "learning_rate": 9.614576715410116e-06, | |
| "loss": 0.4799, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.6491144088080422, | |
| "grad_norm": 0.3408398926258087, | |
| "learning_rate": 9.60157858924978e-06, | |
| "loss": 0.4483, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.6548587841072283, | |
| "grad_norm": 0.3780161440372467, | |
| "learning_rate": 9.588373991985566e-06, | |
| "loss": 0.4684, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.6606031594064146, | |
| "grad_norm": 0.35523924231529236, | |
| "learning_rate": 9.574963516102795e-06, | |
| "loss": 0.4783, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.6663475347056008, | |
| "grad_norm": 0.35043102502822876, | |
| "learning_rate": 9.561347763324484e-06, | |
| "loss": 0.4462, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.672091910004787, | |
| "grad_norm": 0.388933390378952, | |
| "learning_rate": 9.547527344584353e-06, | |
| "loss": 0.4617, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.6778362853039732, | |
| "grad_norm": 0.32936355471611023, | |
| "learning_rate": 9.533502879999398e-06, | |
| "loss": 0.48, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.6835806606031594, | |
| "grad_norm": 0.3353422284126282, | |
| "learning_rate": 9.519274998842084e-06, | |
| "loss": 0.4832, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.6893250359023456, | |
| "grad_norm": 0.36188051104545593, | |
| "learning_rate": 9.504844339512096e-06, | |
| "loss": 0.4527, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.6950694112015319, | |
| "grad_norm": 0.30944687128067017, | |
| "learning_rate": 9.490211549507701e-06, | |
| "loss": 0.4712, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.700813786500718, | |
| "grad_norm": 0.3265434801578522, | |
| "learning_rate": 9.475377285396692e-06, | |
| "loss": 0.4505, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.7065581617999043, | |
| "grad_norm": 0.3855360150337219, | |
| "learning_rate": 9.460342212786933e-06, | |
| "loss": 0.4417, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.7123025370990905, | |
| "grad_norm": 0.32236477732658386, | |
| "learning_rate": 9.445107006296488e-06, | |
| "loss": 0.4765, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.7180469123982767, | |
| "grad_norm": 0.32656142115592957, | |
| "learning_rate": 9.42967234952335e-06, | |
| "loss": 0.4829, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.7237912876974629, | |
| "grad_norm": 0.3501734435558319, | |
| "learning_rate": 9.414038935014777e-06, | |
| "loss": 0.4471, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.7295356629966491, | |
| "grad_norm": 0.32487544417381287, | |
| "learning_rate": 9.398207464236209e-06, | |
| "loss": 0.4483, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.7352800382958353, | |
| "grad_norm": 0.31272241473197937, | |
| "learning_rate": 9.382178647539794e-06, | |
| "loss": 0.4545, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.7410244135950216, | |
| "grad_norm": 0.3410162031650543, | |
| "learning_rate": 9.365953204132526e-06, | |
| "loss": 0.4642, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.7467687888942077, | |
| "grad_norm": 0.327289342880249, | |
| "learning_rate": 9.349531862043952e-06, | |
| "loss": 0.4794, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.752513164193394, | |
| "grad_norm": 0.3137280344963074, | |
| "learning_rate": 9.332915358093532e-06, | |
| "loss": 0.4453, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.7582575394925802, | |
| "grad_norm": 0.3067407011985779, | |
| "learning_rate": 9.316104437857561e-06, | |
| "loss": 0.4506, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.7640019147917664, | |
| "grad_norm": 0.3284437656402588, | |
| "learning_rate": 9.299099855635716e-06, | |
| "loss": 0.4443, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.7697462900909526, | |
| "grad_norm": 0.4092399477958679, | |
| "learning_rate": 9.28190237441722e-06, | |
| "loss": 0.4351, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.7754906653901388, | |
| "grad_norm": 0.3092726469039917, | |
| "learning_rate": 9.2645127658466e-06, | |
| "loss": 0.4532, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.781235040689325, | |
| "grad_norm": 0.34482482075691223, | |
| "learning_rate": 9.246931810189061e-06, | |
| "loss": 0.4626, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.7869794159885113, | |
| "grad_norm": 0.3547237813472748, | |
| "learning_rate": 9.229160296295488e-06, | |
| "loss": 0.4305, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.7927237912876974, | |
| "grad_norm": 0.32667168974876404, | |
| "learning_rate": 9.211199021567034e-06, | |
| "loss": 0.4675, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.7984681665868837, | |
| "grad_norm": 0.3909010589122772, | |
| "learning_rate": 9.193048791919357e-06, | |
| "loss": 0.4743, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.8042125418860699, | |
| "grad_norm": 0.31883516907691956, | |
| "learning_rate": 9.174710421746445e-06, | |
| "loss": 0.4824, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.8099569171852561, | |
| "grad_norm": 0.36612772941589355, | |
| "learning_rate": 9.156184733884084e-06, | |
| "loss": 0.4475, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.8157012924844423, | |
| "grad_norm": 0.32956618070602417, | |
| "learning_rate": 9.137472559572935e-06, | |
| "loss": 0.4762, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.8214456677836285, | |
| "grad_norm": 0.3318566679954529, | |
| "learning_rate": 9.118574738421236e-06, | |
| "loss": 0.4381, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.8271900430828147, | |
| "grad_norm": 0.34344008564949036, | |
| "learning_rate": 9.099492118367123e-06, | |
| "loss": 0.4693, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.832934418382001, | |
| "grad_norm": 0.2847115695476532, | |
| "learning_rate": 9.080225555640601e-06, | |
| "loss": 0.4552, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.8386787936811871, | |
| "grad_norm": 0.31437164545059204, | |
| "learning_rate": 9.0607759147251e-06, | |
| "loss": 0.457, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.8444231689803734, | |
| "grad_norm": 0.2915739417076111, | |
| "learning_rate": 9.04114406831871e-06, | |
| "loss": 0.4318, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.8501675442795597, | |
| "grad_norm": 0.32622477412223816, | |
| "learning_rate": 9.021330897295011e-06, | |
| "loss": 0.456, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.8559119195787458, | |
| "grad_norm": 0.3252268135547638, | |
| "learning_rate": 9.001337290663548e-06, | |
| "loss": 0.4397, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.861656294877932, | |
| "grad_norm": 0.3453269898891449, | |
| "learning_rate": 8.981164145529943e-06, | |
| "loss": 0.4612, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.8674006701771182, | |
| "grad_norm": 0.28802141547203064, | |
| "learning_rate": 8.960812367055646e-06, | |
| "loss": 0.473, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.8731450454763044, | |
| "grad_norm": 0.27404195070266724, | |
| "learning_rate": 8.940282868417321e-06, | |
| "loss": 0.4133, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.8788894207754907, | |
| "grad_norm": 0.3235178589820862, | |
| "learning_rate": 8.91957657076586e-06, | |
| "loss": 0.4658, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.8846337960746768, | |
| "grad_norm": 0.29341384768486023, | |
| "learning_rate": 8.898694403185066e-06, | |
| "loss": 0.4287, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.8903781713738631, | |
| "grad_norm": 0.335021436214447, | |
| "learning_rate": 8.877637302649962e-06, | |
| "loss": 0.4424, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.8961225466730494, | |
| "grad_norm": 0.2819330096244812, | |
| "learning_rate": 8.856406213984743e-06, | |
| "loss": 0.4579, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.9018669219722355, | |
| "grad_norm": 0.3036133646965027, | |
| "learning_rate": 8.835002089820387e-06, | |
| "loss": 0.4468, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.9076112972714218, | |
| "grad_norm": 0.30535823106765747, | |
| "learning_rate": 8.81342589055191e-06, | |
| "loss": 0.4735, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.9133556725706079, | |
| "grad_norm": 0.2687799036502838, | |
| "learning_rate": 8.791678584295276e-06, | |
| "loss": 0.4381, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.9191000478697942, | |
| "grad_norm": 0.3096264898777008, | |
| "learning_rate": 8.76976114684395e-06, | |
| "loss": 0.4467, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.9248444231689804, | |
| "grad_norm": 0.3003120422363281, | |
| "learning_rate": 8.747674561625121e-06, | |
| "loss": 0.4633, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.9305887984681666, | |
| "grad_norm": 0.2911892533302307, | |
| "learning_rate": 8.725419819655582e-06, | |
| "loss": 0.4093, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.9363331737673528, | |
| "grad_norm": 0.28195834159851074, | |
| "learning_rate": 8.702997919497247e-06, | |
| "loss": 0.4535, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.9420775490665391, | |
| "grad_norm": 0.28224724531173706, | |
| "learning_rate": 8.680409867212359e-06, | |
| "loss": 0.4726, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.9478219243657252, | |
| "grad_norm": 0.31931445002555847, | |
| "learning_rate": 8.657656676318346e-06, | |
| "loss": 0.467, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.9535662996649115, | |
| "grad_norm": 0.3119199872016907, | |
| "learning_rate": 8.634739367742341e-06, | |
| "loss": 0.4248, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.9593106749640976, | |
| "grad_norm": 0.2851564586162567, | |
| "learning_rate": 8.611658969775378e-06, | |
| "loss": 0.4421, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.9650550502632839, | |
| "grad_norm": 0.287062406539917, | |
| "learning_rate": 8.588416518026248e-06, | |
| "loss": 0.4379, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.9707994255624701, | |
| "grad_norm": 0.29976505041122437, | |
| "learning_rate": 8.565013055375035e-06, | |
| "loss": 0.453, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.9765438008616563, | |
| "grad_norm": 0.3060532212257385, | |
| "learning_rate": 8.541449631926325e-06, | |
| "loss": 0.4484, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.9822881761608425, | |
| "grad_norm": 0.29888424277305603, | |
| "learning_rate": 8.51772730496208e-06, | |
| "loss": 0.4728, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.9880325514600288, | |
| "grad_norm": 0.31030842661857605, | |
| "learning_rate": 8.49384713889421e-06, | |
| "loss": 0.4441, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.9937769267592149, | |
| "grad_norm": 0.28488606214523315, | |
| "learning_rate": 8.469810205216795e-06, | |
| "loss": 0.4784, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.9995213020584012, | |
| "grad_norm": 0.2805924713611603, | |
| "learning_rate": 8.445617582458033e-06, | |
| "loss": 0.4643, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 1.0052656773575874, | |
| "grad_norm": 0.9323565363883972, | |
| "learning_rate": 8.42127035613182e-06, | |
| "loss": 0.7769, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 1.0110100526567736, | |
| "grad_norm": 0.28666186332702637, | |
| "learning_rate": 8.396769618689064e-06, | |
| "loss": 0.4213, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 1.0167544279559597, | |
| "grad_norm": 0.30998125672340393, | |
| "learning_rate": 8.372116469468654e-06, | |
| "loss": 0.4429, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 1.022498803255146, | |
| "grad_norm": 0.37547767162323, | |
| "learning_rate": 8.347312014648144e-06, | |
| "loss": 0.3927, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 1.0282431785543322, | |
| "grad_norm": 0.31034451723098755, | |
| "learning_rate": 8.32235736719411e-06, | |
| "loss": 0.3735, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 1.0339875538535184, | |
| "grad_norm": 0.4133894443511963, | |
| "learning_rate": 8.297253646812213e-06, | |
| "loss": 0.4508, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 1.0397319291527047, | |
| "grad_norm": 0.3431154787540436, | |
| "learning_rate": 8.272001979896962e-06, | |
| "loss": 0.4064, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 1.0454763044518909, | |
| "grad_norm": 0.31550031900405884, | |
| "learning_rate": 8.246603499481177e-06, | |
| "loss": 0.4075, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 1.051220679751077, | |
| "grad_norm": 0.3756781816482544, | |
| "learning_rate": 8.221059345185136e-06, | |
| "loss": 0.4108, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 1.0569650550502634, | |
| "grad_norm": 0.34178560972213745, | |
| "learning_rate": 8.195370663165455e-06, | |
| "loss": 0.4057, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 1.0627094303494495, | |
| "grad_norm": 0.31101906299591064, | |
| "learning_rate": 8.169538606063647e-06, | |
| "loss": 0.4237, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 1.0684538056486357, | |
| "grad_norm": 0.3273501694202423, | |
| "learning_rate": 8.143564332954426e-06, | |
| "loss": 0.4287, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 1.0741981809478218, | |
| "grad_norm": 0.34117910265922546, | |
| "learning_rate": 8.117449009293668e-06, | |
| "loss": 0.4279, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 1.0799425562470082, | |
| "grad_norm": 0.28906843066215515, | |
| "learning_rate": 8.091193806866147e-06, | |
| "loss": 0.4271, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 1.0856869315461943, | |
| "grad_norm": 0.2958456575870514, | |
| "learning_rate": 8.064799903732936e-06, | |
| "loss": 0.3967, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 1.0914313068453805, | |
| "grad_norm": 0.2881177067756653, | |
| "learning_rate": 8.038268484178566e-06, | |
| "loss": 0.4176, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 1.0971756821445668, | |
| "grad_norm": 0.32246285676956177, | |
| "learning_rate": 8.011600738657865e-06, | |
| "loss": 0.4222, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 1.102920057443753, | |
| "grad_norm": 0.3277246952056885, | |
| "learning_rate": 7.98479786374257e-06, | |
| "loss": 0.3853, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 1.1086644327429391, | |
| "grad_norm": 0.3409973382949829, | |
| "learning_rate": 7.957861062067614e-06, | |
| "loss": 0.4156, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 1.1144088080421255, | |
| "grad_norm": 0.2889862060546875, | |
| "learning_rate": 7.930791542277175e-06, | |
| "loss": 0.4143, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 1.1201531833413116, | |
| "grad_norm": 0.3289618492126465, | |
| "learning_rate": 7.903590518970445e-06, | |
| "loss": 0.4363, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 1.1258975586404978, | |
| "grad_norm": 0.2657482624053955, | |
| "learning_rate": 7.876259212647129e-06, | |
| "loss": 0.3729, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 1.1316419339396842, | |
| "grad_norm": 0.31660833954811096, | |
| "learning_rate": 7.848798849652684e-06, | |
| "loss": 0.4256, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 1.1373863092388703, | |
| "grad_norm": 0.2567324936389923, | |
| "learning_rate": 7.821210662123284e-06, | |
| "loss": 0.3998, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 1.1431306845380564, | |
| "grad_norm": 0.41787901520729065, | |
| "learning_rate": 7.793495887930551e-06, | |
| "loss": 0.3784, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 1.1488750598372426, | |
| "grad_norm": 0.2855132222175598, | |
| "learning_rate": 7.765655770625997e-06, | |
| "loss": 0.3947, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.154619435136429, | |
| "grad_norm": 0.3121788799762726, | |
| "learning_rate": 7.737691559385237e-06, | |
| "loss": 0.4236, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 1.160363810435615, | |
| "grad_norm": 0.3103139102458954, | |
| "learning_rate": 7.709604508951927e-06, | |
| "loss": 0.4296, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 1.1661081857348012, | |
| "grad_norm": 0.29311925172805786, | |
| "learning_rate": 7.68139587958148e-06, | |
| "loss": 0.3973, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 1.1718525610339876, | |
| "grad_norm": 0.2649705111980438, | |
| "learning_rate": 7.653066936984504e-06, | |
| "loss": 0.4285, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 1.1775969363331737, | |
| "grad_norm": 0.3288572430610657, | |
| "learning_rate": 7.6246189522700205e-06, | |
| "loss": 0.4415, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 1.18334131163236, | |
| "grad_norm": 0.3235880732536316, | |
| "learning_rate": 7.596053201888425e-06, | |
| "loss": 0.3963, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 1.1890856869315463, | |
| "grad_norm": 0.26849544048309326, | |
| "learning_rate": 7.56737096757421e-06, | |
| "loss": 0.4009, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 1.1948300622307324, | |
| "grad_norm": 0.34631481766700745, | |
| "learning_rate": 7.538573536288466e-06, | |
| "loss": 0.4067, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 1.2005744375299185, | |
| "grad_norm": 0.2917556166648865, | |
| "learning_rate": 7.509662200161122e-06, | |
| "loss": 0.3852, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 1.206318812829105, | |
| "grad_norm": 0.33460524678230286, | |
| "learning_rate": 7.480638256432977e-06, | |
| "loss": 0.4346, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 1.212063188128291, | |
| "grad_norm": 0.31476372480392456, | |
| "learning_rate": 7.4515030073974915e-06, | |
| "loss": 0.4224, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 1.2178075634274772, | |
| "grad_norm": 0.2829968333244324, | |
| "learning_rate": 7.422257760342351e-06, | |
| "loss": 0.416, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 1.2235519387266636, | |
| "grad_norm": 0.2671918272972107, | |
| "learning_rate": 7.392903827490814e-06, | |
| "loss": 0.3893, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 1.2292963140258497, | |
| "grad_norm": 0.3298075497150421, | |
| "learning_rate": 7.363442525942827e-06, | |
| "loss": 0.3805, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 1.2350406893250359, | |
| "grad_norm": 0.26619213819503784, | |
| "learning_rate": 7.333875177615931e-06, | |
| "loss": 0.4179, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 1.2407850646242222, | |
| "grad_norm": 0.26025766134262085, | |
| "learning_rate": 7.304203109185947e-06, | |
| "loss": 0.425, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 1.2465294399234084, | |
| "grad_norm": 0.2681398093700409, | |
| "learning_rate": 7.274427652027444e-06, | |
| "loss": 0.3768, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 1.2522738152225945, | |
| "grad_norm": 0.30282825231552124, | |
| "learning_rate": 7.244550142154009e-06, | |
| "loss": 0.4041, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 1.2580181905217809, | |
| "grad_norm": 0.24776627123355865, | |
| "learning_rate": 7.214571920158293e-06, | |
| "loss": 0.4106, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 1.263762565820967, | |
| "grad_norm": 0.25247007608413696, | |
| "learning_rate": 7.1844943311518665e-06, | |
| "loss": 0.444, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 1.2695069411201532, | |
| "grad_norm": 0.27782031893730164, | |
| "learning_rate": 7.1543187247048525e-06, | |
| "loss": 0.43, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 1.2752513164193395, | |
| "grad_norm": 0.2578721046447754, | |
| "learning_rate": 7.124046454785387e-06, | |
| "loss": 0.3989, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 1.2809956917185257, | |
| "grad_norm": 0.30569854378700256, | |
| "learning_rate": 7.093678879698858e-06, | |
| "loss": 0.4405, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 1.2867400670177118, | |
| "grad_norm": 0.2797735333442688, | |
| "learning_rate": 7.063217362026957e-06, | |
| "loss": 0.3981, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 1.292484442316898, | |
| "grad_norm": 0.2695583403110504, | |
| "learning_rate": 7.032663268566547e-06, | |
| "loss": 0.3935, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 1.2982288176160843, | |
| "grad_norm": 0.294919490814209, | |
| "learning_rate": 7.002017970268336e-06, | |
| "loss": 0.3952, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 1.3039731929152705, | |
| "grad_norm": 0.2755994200706482, | |
| "learning_rate": 6.97128284217535e-06, | |
| "loss": 0.4327, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 1.3097175682144566, | |
| "grad_norm": 0.2905407249927521, | |
| "learning_rate": 6.9404592633612486e-06, | |
| "loss": 0.4658, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 1.3154619435136428, | |
| "grad_norm": 0.2658722400665283, | |
| "learning_rate": 6.909548616868444e-06, | |
| "loss": 0.4237, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 1.3212063188128291, | |
| "grad_norm": 0.2581554651260376, | |
| "learning_rate": 6.878552289646041e-06, | |
| "loss": 0.41, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 1.3269506941120153, | |
| "grad_norm": 0.2688734233379364, | |
| "learning_rate": 6.847471672487607e-06, | |
| "loss": 0.3662, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 1.3326950694112014, | |
| "grad_norm": 0.30700597167015076, | |
| "learning_rate": 6.816308159968761e-06, | |
| "loss": 0.4595, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 1.3384394447103878, | |
| "grad_norm": 0.271698921918869, | |
| "learning_rate": 6.7850631503846165e-06, | |
| "loss": 0.4073, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 1.344183820009574, | |
| "grad_norm": 0.24926996231079102, | |
| "learning_rate": 6.753738045687021e-06, | |
| "loss": 0.42, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 1.34992819530876, | |
| "grad_norm": 0.2579510807991028, | |
| "learning_rate": 6.722334251421665e-06, | |
| "loss": 0.3975, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 1.3556725706079464, | |
| "grad_norm": 0.2766447067260742, | |
| "learning_rate": 6.690853176665007e-06, | |
| "loss": 0.4539, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 1.3614169459071326, | |
| "grad_norm": 0.2743930220603943, | |
| "learning_rate": 6.659296233961055e-06, | |
| "loss": 0.3857, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 1.3671613212063187, | |
| "grad_norm": 0.2682178318500519, | |
| "learning_rate": 6.627664839257979e-06, | |
| "loss": 0.4127, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 1.372905696505505, | |
| "grad_norm": 0.26600515842437744, | |
| "learning_rate": 6.595960411844589e-06, | |
| "loss": 0.4309, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 1.3786500718046912, | |
| "grad_norm": 0.2679958939552307, | |
| "learning_rate": 6.564184374286636e-06, | |
| "loss": 0.4276, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 1.3843944471038774, | |
| "grad_norm": 0.32891181111335754, | |
| "learning_rate": 6.532338152363001e-06, | |
| "loss": 0.3908, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 1.3901388224030637, | |
| "grad_norm": 0.2673075199127197, | |
| "learning_rate": 6.500423175001705e-06, | |
| "loss": 0.3662, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 1.39588319770225, | |
| "grad_norm": 0.29596373438835144, | |
| "learning_rate": 6.468440874215801e-06, | |
| "loss": 0.4268, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 1.401627573001436, | |
| "grad_norm": 0.2759062349796295, | |
| "learning_rate": 6.43639268503912e-06, | |
| "loss": 0.4216, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 1.4073719483006224, | |
| "grad_norm": 0.2963887155056, | |
| "learning_rate": 6.40428004546188e-06, | |
| "loss": 0.4303, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 1.4131163235998085, | |
| "grad_norm": 0.26301926374435425, | |
| "learning_rate": 6.372104396366162e-06, | |
| "loss": 0.3999, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 1.4188606988989947, | |
| "grad_norm": 0.26165592670440674, | |
| "learning_rate": 6.339867181461265e-06, | |
| "loss": 0.4315, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 1.424605074198181, | |
| "grad_norm": 0.2649396061897278, | |
| "learning_rate": 6.307569847218917e-06, | |
| "loss": 0.3932, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 1.4303494494973672, | |
| "grad_norm": 0.26852184534072876, | |
| "learning_rate": 6.275213842808383e-06, | |
| "loss": 0.4079, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 1.4360938247965533, | |
| "grad_norm": 0.2582874894142151, | |
| "learning_rate": 6.242800620031434e-06, | |
| "loss": 0.4011, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 1.4418382000957397, | |
| "grad_norm": 0.27596113085746765, | |
| "learning_rate": 6.2103316332572095e-06, | |
| "loss": 0.4273, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 1.4475825753949259, | |
| "grad_norm": 0.2603662610054016, | |
| "learning_rate": 6.177808339356954e-06, | |
| "loss": 0.4457, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 1.453326950694112, | |
| "grad_norm": 0.2617853283882141, | |
| "learning_rate": 6.14523219763866e-06, | |
| "loss": 0.3843, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 1.4590713259932984, | |
| "grad_norm": 0.2927669882774353, | |
| "learning_rate": 6.112604669781572e-06, | |
| "loss": 0.409, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 1.4648157012924845, | |
| "grad_norm": 0.2545565962791443, | |
| "learning_rate": 6.079927219770623e-06, | |
| "loss": 0.4197, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 1.4705600765916707, | |
| "grad_norm": 0.25056391954421997, | |
| "learning_rate": 6.047201313830724e-06, | |
| "loss": 0.412, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 1.4763044518908568, | |
| "grad_norm": 0.26942598819732666, | |
| "learning_rate": 6.014428420360987e-06, | |
| "loss": 0.4377, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 1.4820488271900432, | |
| "grad_norm": 0.2583950459957123, | |
| "learning_rate": 5.9816100098688456e-06, | |
| "loss": 0.3939, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 1.4877932024892293, | |
| "grad_norm": 0.286765992641449, | |
| "learning_rate": 5.948747554904054e-06, | |
| "loss": 0.4441, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 1.4935375777884154, | |
| "grad_norm": 0.2546658515930176, | |
| "learning_rate": 5.915842529992632e-06, | |
| "loss": 0.4084, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 1.4992819530876016, | |
| "grad_norm": 0.24491065740585327, | |
| "learning_rate": 5.8828964115706925e-06, | |
| "loss": 0.3642, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 1.505026328386788, | |
| "grad_norm": 0.2614350914955139, | |
| "learning_rate": 5.849910677918205e-06, | |
| "loss": 0.385, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 1.510770703685974, | |
| "grad_norm": 0.263822466135025, | |
| "learning_rate": 5.816886809092651e-06, | |
| "loss": 0.3977, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 1.5165150789851602, | |
| "grad_norm": 0.2711222469806671, | |
| "learning_rate": 5.783826286862631e-06, | |
| "loss": 0.4336, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 1.5222594542843466, | |
| "grad_norm": 0.23413866758346558, | |
| "learning_rate": 5.750730594641367e-06, | |
| "loss": 0.391, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 1.5280038295835328, | |
| "grad_norm": 0.26713666319847107, | |
| "learning_rate": 5.717601217420143e-06, | |
| "loss": 0.3995, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 1.533748204882719, | |
| "grad_norm": 0.2639370262622833, | |
| "learning_rate": 5.68443964170168e-06, | |
| "loss": 0.4074, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 1.5394925801819053, | |
| "grad_norm": 0.2865756154060364, | |
| "learning_rate": 5.6512473554334294e-06, | |
| "loss": 0.4146, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 1.5452369554810914, | |
| "grad_norm": 0.256339430809021, | |
| "learning_rate": 5.618025847940817e-06, | |
| "loss": 0.4143, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 1.5509813307802776, | |
| "grad_norm": 0.24432285130023956, | |
| "learning_rate": 5.584776609860414e-06, | |
| "loss": 0.3997, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 1.556725706079464, | |
| "grad_norm": 0.2671893835067749, | |
| "learning_rate": 5.551501133073048e-06, | |
| "loss": 0.4625, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 1.56247008137865, | |
| "grad_norm": 0.2614923417568207, | |
| "learning_rate": 5.518200910636875e-06, | |
| "loss": 0.3983, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 1.5682144566778362, | |
| "grad_norm": 0.2648710608482361, | |
| "learning_rate": 5.4848774367203715e-06, | |
| "loss": 0.4264, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 1.5739588319770226, | |
| "grad_norm": 0.25889015197753906, | |
| "learning_rate": 5.451532206535306e-06, | |
| "loss": 0.4327, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 1.5797032072762087, | |
| "grad_norm": 0.28398048877716064, | |
| "learning_rate": 5.418166716269636e-06, | |
| "loss": 0.4216, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 1.5854475825753949, | |
| "grad_norm": 0.2506558299064636, | |
| "learning_rate": 5.384782463020385e-06, | |
| "loss": 0.4183, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 1.5911919578745812, | |
| "grad_norm": 0.2716221511363983, | |
| "learning_rate": 5.351380944726465e-06, | |
| "loss": 0.4131, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 1.5969363331737674, | |
| "grad_norm": 0.2257940024137497, | |
| "learning_rate": 5.317963660101464e-06, | |
| "loss": 0.3597, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 1.6026807084729535, | |
| "grad_norm": 0.28616511821746826, | |
| "learning_rate": 5.284532108566396e-06, | |
| "loss": 0.4147, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 1.6084250837721399, | |
| "grad_norm": 0.26668059825897217, | |
| "learning_rate": 5.251087790182428e-06, | |
| "loss": 0.4133, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 1.614169459071326, | |
| "grad_norm": 0.28087204694747925, | |
| "learning_rate": 5.217632205583574e-06, | |
| "loss": 0.4326, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 1.6199138343705122, | |
| "grad_norm": 0.2532707452774048, | |
| "learning_rate": 5.184166855909355e-06, | |
| "loss": 0.4132, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 1.6256582096696985, | |
| "grad_norm": 0.2916858196258545, | |
| "learning_rate": 5.150693242737444e-06, | |
| "loss": 0.4304, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 1.6314025849688847, | |
| "grad_norm": 0.2610633969306946, | |
| "learning_rate": 5.117212868016303e-06, | |
| "loss": 0.4486, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 1.6371469602680708, | |
| "grad_norm": 0.27046632766723633, | |
| "learning_rate": 5.083727233997775e-06, | |
| "loss": 0.3898, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 1.6428913355672572, | |
| "grad_norm": 0.2789868116378784, | |
| "learning_rate": 5.05023784316969e-06, | |
| "loss": 0.4064, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 1.6486357108664431, | |
| "grad_norm": 0.31009382009506226, | |
| "learning_rate": 5.016746198188439e-06, | |
| "loss": 0.44, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 1.6543800861656295, | |
| "grad_norm": 0.24213601648807526, | |
| "learning_rate": 4.983253801811562e-06, | |
| "loss": 0.38, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 1.6601244614648158, | |
| "grad_norm": 0.2574230432510376, | |
| "learning_rate": 4.949762156830312e-06, | |
| "loss": 0.4279, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 1.6658688367640018, | |
| "grad_norm": 0.3645157814025879, | |
| "learning_rate": 4.916272766002227e-06, | |
| "loss": 0.4149, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 1.6716132120631881, | |
| "grad_norm": 0.25819137692451477, | |
| "learning_rate": 4.882787131983698e-06, | |
| "loss": 0.3851, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 1.6773575873623745, | |
| "grad_norm": 0.2552003264427185, | |
| "learning_rate": 4.849306757262558e-06, | |
| "loss": 0.3952, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 1.6831019626615604, | |
| "grad_norm": 0.31992608308792114, | |
| "learning_rate": 4.8158331440906466e-06, | |
| "loss": 0.4331, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 1.6888463379607468, | |
| "grad_norm": 0.276024729013443, | |
| "learning_rate": 4.7823677944164285e-06, | |
| "loss": 0.4551, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 1.694590713259933, | |
| "grad_norm": 0.2532188892364502, | |
| "learning_rate": 4.748912209817572e-06, | |
| "loss": 0.4066, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 1.700335088559119, | |
| "grad_norm": 0.24477799236774445, | |
| "learning_rate": 4.715467891433607e-06, | |
| "loss": 0.3797, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 1.7060794638583054, | |
| "grad_norm": 0.2809993624687195, | |
| "learning_rate": 4.682036339898537e-06, | |
| "loss": 0.4005, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 1.7118238391574916, | |
| "grad_norm": 0.2791358530521393, | |
| "learning_rate": 4.6486190552735375e-06, | |
| "loss": 0.4353, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 1.7175682144566777, | |
| "grad_norm": 0.2509957253932953, | |
| "learning_rate": 4.615217536979616e-06, | |
| "loss": 0.4061, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 1.723312589755864, | |
| "grad_norm": 0.24833880364894867, | |
| "learning_rate": 4.581833283730367e-06, | |
| "loss": 0.3808, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 1.7290569650550502, | |
| "grad_norm": 0.30278247594833374, | |
| "learning_rate": 4.548467793464696e-06, | |
| "loss": 0.4447, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 1.7348013403542364, | |
| "grad_norm": 0.2316392958164215, | |
| "learning_rate": 4.515122563279631e-06, | |
| "loss": 0.3859, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 1.7405457156534228, | |
| "grad_norm": 0.25202158093452454, | |
| "learning_rate": 4.481799089363127e-06, | |
| "loss": 0.3879, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 1.746290090952609, | |
| "grad_norm": 0.2888227701187134, | |
| "learning_rate": 4.448498866926952e-06, | |
| "loss": 0.4187, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 1.752034466251795, | |
| "grad_norm": 0.2678024470806122, | |
| "learning_rate": 4.415223390139588e-06, | |
| "loss": 0.4348, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 1.7577788415509814, | |
| "grad_norm": 0.25489702820777893, | |
| "learning_rate": 4.381974152059184e-06, | |
| "loss": 0.4216, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 1.7635232168501676, | |
| "grad_norm": 0.2304321825504303, | |
| "learning_rate": 4.348752644566573e-06, | |
| "loss": 0.3878, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 1.7692675921493537, | |
| "grad_norm": 0.2631133198738098, | |
| "learning_rate": 4.315560358298321e-06, | |
| "loss": 0.4249, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 1.77501196744854, | |
| "grad_norm": 0.25828343629837036, | |
| "learning_rate": 4.2823987825798575e-06, | |
| "loss": 0.4005, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 1.7807563427477262, | |
| "grad_norm": 0.26033782958984375, | |
| "learning_rate": 4.249269405358634e-06, | |
| "loss": 0.3837, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 1.7865007180469124, | |
| "grad_norm": 0.26372194290161133, | |
| "learning_rate": 4.2161737131373695e-06, | |
| "loss": 0.4337, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 1.7922450933460987, | |
| "grad_norm": 0.24711427092552185, | |
| "learning_rate": 4.183113190907349e-06, | |
| "loss": 0.4053, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 1.7979894686452849, | |
| "grad_norm": 0.2691591680049896, | |
| "learning_rate": 4.150089322081797e-06, | |
| "loss": 0.3914, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 1.803733843944471, | |
| "grad_norm": 0.24242590367794037, | |
| "learning_rate": 4.1171035884293075e-06, | |
| "loss": 0.3837, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 1.8094782192436574, | |
| "grad_norm": 0.24888436496257782, | |
| "learning_rate": 4.084157470007371e-06, | |
| "loss": 0.411, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 1.8152225945428435, | |
| "grad_norm": 0.26371899247169495, | |
| "learning_rate": 4.051252445095946e-06, | |
| "loss": 0.4466, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 1.8209669698420297, | |
| "grad_norm": 0.2363133430480957, | |
| "learning_rate": 4.018389990131156e-06, | |
| "loss": 0.4095, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 1.826711345141216, | |
| "grad_norm": 0.23559501767158508, | |
| "learning_rate": 3.985571579639013e-06, | |
| "loss": 0.3997, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 1.832455720440402, | |
| "grad_norm": 0.23539544641971588, | |
| "learning_rate": 3.952798686169279e-06, | |
| "loss": 0.3972, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 1.8382000957395883, | |
| "grad_norm": 0.2795656621456146, | |
| "learning_rate": 3.920072780229378e-06, | |
| "loss": 0.3976, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 1.8439444710387747, | |
| "grad_norm": 0.2547081410884857, | |
| "learning_rate": 3.887395330218429e-06, | |
| "loss": 0.4259, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 1.8496888463379606, | |
| "grad_norm": 0.22198687493801117, | |
| "learning_rate": 3.854767802361342e-06, | |
| "loss": 0.3728, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 1.855433221637147, | |
| "grad_norm": 0.2664637267589569, | |
| "learning_rate": 3.822191660643047e-06, | |
| "loss": 0.4256, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 1.8611775969363333, | |
| "grad_norm": 0.22866730391979218, | |
| "learning_rate": 3.789668366742792e-06, | |
| "loss": 0.4283, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 1.8669219722355193, | |
| "grad_norm": 0.24711646139621735, | |
| "learning_rate": 3.7571993799685675e-06, | |
| "loss": 0.4124, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 1.8726663475347056, | |
| "grad_norm": 0.2538555860519409, | |
| "learning_rate": 3.7247861571916183e-06, | |
| "loss": 0.3934, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 1.8784107228338918, | |
| "grad_norm": 0.2567046582698822, | |
| "learning_rate": 3.6924301527810856e-06, | |
| "loss": 0.411, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 1.884155098133078, | |
| "grad_norm": 0.2725870907306671, | |
| "learning_rate": 3.6601328185387364e-06, | |
| "loss": 0.4195, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 1.8898994734322643, | |
| "grad_norm": 0.24136009812355042, | |
| "learning_rate": 3.6278956036338397e-06, | |
| "loss": 0.3904, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 1.8956438487314504, | |
| "grad_norm": 0.2512601315975189, | |
| "learning_rate": 3.5957199545381216e-06, | |
| "loss": 0.3869, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 1.9013882240306366, | |
| "grad_norm": 0.24334044754505157, | |
| "learning_rate": 3.5636073149608824e-06, | |
| "loss": 0.4019, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 1.907132599329823, | |
| "grad_norm": 0.24103941023349762, | |
| "learning_rate": 3.5315591257842e-06, | |
| "loss": 0.3609, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 1.912876974629009, | |
| "grad_norm": 0.24972982704639435, | |
| "learning_rate": 3.4995768249982975e-06, | |
| "loss": 0.3843, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 1.9186213499281952, | |
| "grad_norm": 0.24723373353481293, | |
| "learning_rate": 3.467661847637001e-06, | |
| "loss": 0.4227, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 1.9243657252273816, | |
| "grad_norm": 0.24389074742794037, | |
| "learning_rate": 3.4358156257133644e-06, | |
| "loss": 0.4296, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 1.9301101005265677, | |
| "grad_norm": 0.2541641294956207, | |
| "learning_rate": 3.404039588155413e-06, | |
| "loss": 0.3727, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 1.9358544758257539, | |
| "grad_norm": 0.22656212747097015, | |
| "learning_rate": 3.372335160742022e-06, | |
| "loss": 0.3666, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 1.9415988511249402, | |
| "grad_norm": 0.24361655116081238, | |
| "learning_rate": 3.3407037660389474e-06, | |
| "loss": 0.388, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 1.9473432264241264, | |
| "grad_norm": 0.24967317283153534, | |
| "learning_rate": 3.3091468233349934e-06, | |
| "loss": 0.4433, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 1.9530876017233125, | |
| "grad_norm": 0.23722046613693237, | |
| "learning_rate": 3.2776657485783357e-06, | |
| "loss": 0.4124, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 1.958831977022499, | |
| "grad_norm": 0.24369174242019653, | |
| "learning_rate": 3.246261954312979e-06, | |
| "loss": 0.4216, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 1.964576352321685, | |
| "grad_norm": 0.23776546120643616, | |
| "learning_rate": 3.2149368496153856e-06, | |
| "loss": 0.4256, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 1.9703207276208712, | |
| "grad_norm": 0.25640493631362915, | |
| "learning_rate": 3.1836918400312387e-06, | |
| "loss": 0.3812, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 1.9760651029200575, | |
| "grad_norm": 0.24129056930541992, | |
| "learning_rate": 3.152528327512395e-06, | |
| "loss": 0.4002, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 1.9818094782192437, | |
| "grad_norm": 0.24882204830646515, | |
| "learning_rate": 3.1214477103539585e-06, | |
| "loss": 0.414, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 1.9875538535184298, | |
| "grad_norm": 0.24418820440769196, | |
| "learning_rate": 3.0904513831315563e-06, | |
| "loss": 0.4261, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 1.9932982288176162, | |
| "grad_norm": 0.2543267011642456, | |
| "learning_rate": 3.059540736638751e-06, | |
| "loss": 0.4272, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 1.9990426041168023, | |
| "grad_norm": 0.23077604174613953, | |
| "learning_rate": 3.028717157824652e-06, | |
| "loss": 0.3639, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 2.0047869794159885, | |
| "grad_norm": 0.9153398871421814, | |
| "learning_rate": 2.9979820297316652e-06, | |
| "loss": 0.6275, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 2.010531354715175, | |
| "grad_norm": 0.31731805205345154, | |
| "learning_rate": 2.9673367314334533e-06, | |
| "loss": 0.4267, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 2.016275730014361, | |
| "grad_norm": 0.2978307008743286, | |
| "learning_rate": 2.936782637973044e-06, | |
| "loss": 0.3788, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 2.022020105313547, | |
| "grad_norm": 0.28210750222206116, | |
| "learning_rate": 2.9063211203011443e-06, | |
| "loss": 0.383, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 2.0277644806127335, | |
| "grad_norm": 0.28862592577934265, | |
| "learning_rate": 2.8759535452146128e-06, | |
| "loss": 0.3827, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 2.0335088559119194, | |
| "grad_norm": 0.3073420524597168, | |
| "learning_rate": 2.8456812752951483e-06, | |
| "loss": 0.3858, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 2.039253231211106, | |
| "grad_norm": 0.27528196573257446, | |
| "learning_rate": 2.815505668848136e-06, | |
| "loss": 0.3536, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 2.044997606510292, | |
| "grad_norm": 0.25948259234428406, | |
| "learning_rate": 2.785428079841709e-06, | |
| "loss": 0.3844, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 2.050741981809478, | |
| "grad_norm": 0.3084884285926819, | |
| "learning_rate": 2.755449857845992e-06, | |
| "loss": 0.3772, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 2.0564863571086645, | |
| "grad_norm": 0.2928623855113983, | |
| "learning_rate": 2.725572347972558e-06, | |
| "loss": 0.3562, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 2.062230732407851, | |
| "grad_norm": 0.27207672595977783, | |
| "learning_rate": 2.6957968908140546e-06, | |
| "loss": 0.362, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 2.0679751077070367, | |
| "grad_norm": 0.2691848576068878, | |
| "learning_rate": 2.666124822384071e-06, | |
| "loss": 0.3807, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 2.073719483006223, | |
| "grad_norm": 0.22607873380184174, | |
| "learning_rate": 2.636557474057173e-06, | |
| "loss": 0.3436, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 2.0794638583054095, | |
| "grad_norm": 0.25091448426246643, | |
| "learning_rate": 2.607096172509187e-06, | |
| "loss": 0.374, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 2.0852082336045954, | |
| "grad_norm": 0.306155800819397, | |
| "learning_rate": 2.5777422396576503e-06, | |
| "loss": 0.4164, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 2.0909526089037818, | |
| "grad_norm": 0.25360941886901855, | |
| "learning_rate": 2.5484969926025114e-06, | |
| "loss": 0.3706, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 2.096696984202968, | |
| "grad_norm": 0.2415456920862198, | |
| "learning_rate": 2.5193617435670244e-06, | |
| "loss": 0.4131, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 2.102441359502154, | |
| "grad_norm": 0.23606958985328674, | |
| "learning_rate": 2.4903377998388783e-06, | |
| "loss": 0.3781, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 2.1081857348013404, | |
| "grad_norm": 0.2365504652261734, | |
| "learning_rate": 2.461426463711535e-06, | |
| "loss": 0.3769, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 2.113930110100527, | |
| "grad_norm": 0.23365849256515503, | |
| "learning_rate": 2.4326290324257896e-06, | |
| "loss": 0.3573, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 2.1196744853997127, | |
| "grad_norm": 0.24357256293296814, | |
| "learning_rate": 2.403946798111576e-06, | |
| "loss": 0.3505, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 2.125418860698899, | |
| "grad_norm": 0.28438547253608704, | |
| "learning_rate": 2.37538104772998e-06, | |
| "loss": 0.4116, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 2.131163235998085, | |
| "grad_norm": 0.21200957894325256, | |
| "learning_rate": 2.3469330630154974e-06, | |
| "loss": 0.326, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 2.1369076112972714, | |
| "grad_norm": 0.2492566853761673, | |
| "learning_rate": 2.318604120418521e-06, | |
| "loss": 0.3999, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 2.1426519865964577, | |
| "grad_norm": 0.22796769440174103, | |
| "learning_rate": 2.2903954910480746e-06, | |
| "loss": 0.391, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 2.1483963618956436, | |
| "grad_norm": 0.23675256967544556, | |
| "learning_rate": 2.2623084406147643e-06, | |
| "loss": 0.362, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 2.15414073719483, | |
| "grad_norm": 0.22599707543849945, | |
| "learning_rate": 2.234344229374003e-06, | |
| "loss": 0.3744, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 2.1598851124940164, | |
| "grad_norm": 0.23200853168964386, | |
| "learning_rate": 2.2065041120694487e-06, | |
| "loss": 0.3748, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 2.1656294877932023, | |
| "grad_norm": 0.22807936370372772, | |
| "learning_rate": 2.178789337876716e-06, | |
| "loss": 0.3762, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 2.1713738630923887, | |
| "grad_norm": 0.21814514696598053, | |
| "learning_rate": 2.151201150347318e-06, | |
| "loss": 0.3669, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 2.177118238391575, | |
| "grad_norm": 0.22707633674144745, | |
| "learning_rate": 2.123740787352872e-06, | |
| "loss": 0.341, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 2.182862613690761, | |
| "grad_norm": 0.2529568672180176, | |
| "learning_rate": 2.096409481029556e-06, | |
| "loss": 0.3723, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 2.1886069889899473, | |
| "grad_norm": 0.231676384806633, | |
| "learning_rate": 2.069208457722828e-06, | |
| "loss": 0.3581, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 2.1943513642891337, | |
| "grad_norm": 0.22723394632339478, | |
| "learning_rate": 2.042138937932388e-06, | |
| "loss": 0.3621, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 2.2000957395883196, | |
| "grad_norm": 0.24381379783153534, | |
| "learning_rate": 2.015202136257432e-06, | |
| "loss": 0.411, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 2.205840114887506, | |
| "grad_norm": 0.22075577080249786, | |
| "learning_rate": 1.988399261342135e-06, | |
| "loss": 0.3448, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 2.2115844901866923, | |
| "grad_norm": 0.2534730136394501, | |
| "learning_rate": 1.9617315158214363e-06, | |
| "loss": 0.3949, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 2.2173288654858783, | |
| "grad_norm": 0.2261105328798294, | |
| "learning_rate": 1.935200096267064e-06, | |
| "loss": 0.3752, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 2.2230732407850646, | |
| "grad_norm": 0.22049082815647125, | |
| "learning_rate": 1.908806193133855e-06, | |
| "loss": 0.3693, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 2.228817616084251, | |
| "grad_norm": 0.21172842383384705, | |
| "learning_rate": 1.8825509907063328e-06, | |
| "loss": 0.395, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 2.234561991383437, | |
| "grad_norm": 0.24431262910366058, | |
| "learning_rate": 1.856435667045577e-06, | |
| "loss": 0.3871, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 2.2403063666826233, | |
| "grad_norm": 0.24871356785297394, | |
| "learning_rate": 1.8304613939363531e-06, | |
| "loss": 0.3852, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 2.2460507419818097, | |
| "grad_norm": 0.21080052852630615, | |
| "learning_rate": 1.8046293368345485e-06, | |
| "loss": 0.3687, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 2.2517951172809956, | |
| "grad_norm": 0.1993173360824585, | |
| "learning_rate": 1.7789406548148647e-06, | |
| "loss": 0.3554, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 2.257539492580182, | |
| "grad_norm": 0.23289141058921814, | |
| "learning_rate": 1.7533965005188242e-06, | |
| "loss": 0.4427, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 2.2632838678793683, | |
| "grad_norm": 0.20976853370666504, | |
| "learning_rate": 1.7279980201030382e-06, | |
| "loss": 0.3516, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 2.2690282431785542, | |
| "grad_norm": 0.23207247257232666, | |
| "learning_rate": 1.7027463531877897e-06, | |
| "loss": 0.3759, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 2.2747726184777406, | |
| "grad_norm": 0.22939598560333252, | |
| "learning_rate": 1.677642632805892e-06, | |
| "loss": 0.3825, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 2.280516993776927, | |
| "grad_norm": 0.22694642841815948, | |
| "learning_rate": 1.6526879853518558e-06, | |
| "loss": 0.3804, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 2.286261369076113, | |
| "grad_norm": 0.2267027348279953, | |
| "learning_rate": 1.6278835305313462e-06, | |
| "loss": 0.3835, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 2.2920057443752992, | |
| "grad_norm": 0.23273469507694244, | |
| "learning_rate": 1.6032303813109368e-06, | |
| "loss": 0.3769, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 2.297750119674485, | |
| "grad_norm": 0.22379297018051147, | |
| "learning_rate": 1.578729643868181e-06, | |
| "loss": 0.3861, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 2.3034944949736715, | |
| "grad_norm": 0.2582657039165497, | |
| "learning_rate": 1.5543824175419691e-06, | |
| "loss": 0.3953, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 2.309238870272858, | |
| "grad_norm": 0.20334425568580627, | |
| "learning_rate": 1.5301897947832063e-06, | |
| "loss": 0.3766, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 2.3149832455720443, | |
| "grad_norm": 0.20961356163024902, | |
| "learning_rate": 1.5061528611057917e-06, | |
| "loss": 0.3515, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 2.32072762087123, | |
| "grad_norm": 0.226267471909523, | |
| "learning_rate": 1.4822726950379207e-06, | |
| "loss": 0.4064, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 2.3264719961704166, | |
| "grad_norm": 0.2132333666086197, | |
| "learning_rate": 1.4585503680736756e-06, | |
| "loss": 0.3639, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 2.3322163714696025, | |
| "grad_norm": 0.22161203622817993, | |
| "learning_rate": 1.4349869446249664e-06, | |
| "loss": 0.3643, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 2.337960746768789, | |
| "grad_norm": 0.21189740300178528, | |
| "learning_rate": 1.4115834819737534e-06, | |
| "loss": 0.3316, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 2.343705122067975, | |
| "grad_norm": 0.2345123291015625, | |
| "learning_rate": 1.3883410302246237e-06, | |
| "loss": 0.3701, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 2.349449497367161, | |
| "grad_norm": 0.23276999592781067, | |
| "learning_rate": 1.3652606322576606e-06, | |
| "loss": 0.3434, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 2.3551938726663475, | |
| "grad_norm": 0.21485473215579987, | |
| "learning_rate": 1.3423433236816563e-06, | |
| "loss": 0.3821, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 2.360938247965534, | |
| "grad_norm": 0.21681098639965057, | |
| "learning_rate": 1.3195901327876426e-06, | |
| "loss": 0.3717, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 2.36668262326472, | |
| "grad_norm": 0.2142193466424942, | |
| "learning_rate": 1.2970020805027555e-06, | |
| "loss": 0.3555, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 2.372426998563906, | |
| "grad_norm": 0.2359546720981598, | |
| "learning_rate": 1.2745801803444192e-06, | |
| "loss": 0.3989, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 2.3781713738630925, | |
| "grad_norm": 0.25387322902679443, | |
| "learning_rate": 1.25232543837488e-06, | |
| "loss": 0.3943, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 2.3839157491622784, | |
| "grad_norm": 0.21573728322982788, | |
| "learning_rate": 1.2302388531560515e-06, | |
| "loss": 0.3693, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 2.389660124461465, | |
| "grad_norm": 0.2112240195274353, | |
| "learning_rate": 1.2083214157047257e-06, | |
| "loss": 0.3896, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 2.395404499760651, | |
| "grad_norm": 0.21627095341682434, | |
| "learning_rate": 1.186574109448091e-06, | |
| "loss": 0.3827, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 2.401148875059837, | |
| "grad_norm": 0.20708608627319336, | |
| "learning_rate": 1.164997910179615e-06, | |
| "loss": 0.3978, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 2.4068932503590235, | |
| "grad_norm": 0.21296700835227966, | |
| "learning_rate": 1.1435937860152579e-06, | |
| "loss": 0.384, | |
| "step": 419 | |
| }, | |
| { | |
| "epoch": 2.41263762565821, | |
| "grad_norm": 0.21074937283992767, | |
| "learning_rate": 1.1223626973500395e-06, | |
| "loss": 0.3468, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 2.4183820009573957, | |
| "grad_norm": 0.22150783240795135, | |
| "learning_rate": 1.1013055968149343e-06, | |
| "loss": 0.3817, | |
| "step": 421 | |
| }, | |
| { | |
| "epoch": 2.424126376256582, | |
| "grad_norm": 0.21905986964702606, | |
| "learning_rate": 1.0804234292341426e-06, | |
| "loss": 0.3984, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 2.4298707515557685, | |
| "grad_norm": 0.20688596367835999, | |
| "learning_rate": 1.0597171315826805e-06, | |
| "loss": 0.3382, | |
| "step": 423 | |
| }, | |
| { | |
| "epoch": 2.4356151268549544, | |
| "grad_norm": 0.21128499507904053, | |
| "learning_rate": 1.0391876329443534e-06, | |
| "loss": 0.3582, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 2.4413595021541408, | |
| "grad_norm": 0.23447024822235107, | |
| "learning_rate": 1.0188358544700583e-06, | |
| "loss": 0.3776, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 2.447103877453327, | |
| "grad_norm": 0.21497882902622223, | |
| "learning_rate": 9.986627093364542e-07, | |
| "loss": 0.4066, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 2.452848252752513, | |
| "grad_norm": 0.20009803771972656, | |
| "learning_rate": 9.786691027049893e-07, | |
| "loss": 0.3365, | |
| "step": 427 | |
| }, | |
| { | |
| "epoch": 2.4585926280516994, | |
| "grad_norm": 0.24452891945838928, | |
| "learning_rate": 9.588559316812906e-07, | |
| "loss": 0.4195, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 2.4643370033508853, | |
| "grad_norm": 0.20632074773311615, | |
| "learning_rate": 9.392240852749007e-07, | |
| "loss": 0.3858, | |
| "step": 429 | |
| }, | |
| { | |
| "epoch": 2.4700813786500717, | |
| "grad_norm": 0.2089974731206894, | |
| "learning_rate": 9.197744443594003e-07, | |
| "loss": 0.3525, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 2.475825753949258, | |
| "grad_norm": 0.20910853147506714, | |
| "learning_rate": 9.005078816328772e-07, | |
| "loss": 0.4122, | |
| "step": 431 | |
| }, | |
| { | |
| "epoch": 2.4815701292484444, | |
| "grad_norm": 0.2021329402923584, | |
| "learning_rate": 8.814252615787661e-07, | |
| "loss": 0.3664, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 2.4873145045476304, | |
| "grad_norm": 0.22205480933189392, | |
| "learning_rate": 8.625274404270662e-07, | |
| "loss": 0.4036, | |
| "step": 433 | |
| }, | |
| { | |
| "epoch": 2.4930588798468167, | |
| "grad_norm": 0.21877549588680267, | |
| "learning_rate": 8.438152661159165e-07, | |
| "loss": 0.3557, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 2.4988032551460027, | |
| "grad_norm": 0.2282644808292389, | |
| "learning_rate": 8.252895782535569e-07, | |
| "loss": 0.3974, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 2.504547630445189, | |
| "grad_norm": 0.19086650013923645, | |
| "learning_rate": 8.069512080806441e-07, | |
| "loss": 0.3348, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 2.5102920057443754, | |
| "grad_norm": 0.22070568799972534, | |
| "learning_rate": 7.88800978432967e-07, | |
| "loss": 0.3948, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 2.5160363810435618, | |
| "grad_norm": 0.2190258651971817, | |
| "learning_rate": 7.708397037045129e-07, | |
| "loss": 0.4058, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 2.5217807563427477, | |
| "grad_norm": 0.19553068280220032, | |
| "learning_rate": 7.530681898109393e-07, | |
| "loss": 0.3539, | |
| "step": 439 | |
| }, | |
| { | |
| "epoch": 2.527525131641934, | |
| "grad_norm": 0.22662606835365295, | |
| "learning_rate": 7.35487234153402e-07, | |
| "loss": 0.4109, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 2.53326950694112, | |
| "grad_norm": 0.229027658700943, | |
| "learning_rate": 7.180976255827809e-07, | |
| "loss": 0.4039, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 2.5390138822403063, | |
| "grad_norm": 0.19389323890209198, | |
| "learning_rate": 7.009001443642843e-07, | |
| "loss": 0.3364, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 2.5447582575394927, | |
| "grad_norm": 0.18867188692092896, | |
| "learning_rate": 6.838955621424404e-07, | |
| "loss": 0.3302, | |
| "step": 443 | |
| }, | |
| { | |
| "epoch": 2.550502632838679, | |
| "grad_norm": 0.22040928900241852, | |
| "learning_rate": 6.67084641906468e-07, | |
| "loss": 0.4042, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 2.556247008137865, | |
| "grad_norm": 0.20983703434467316, | |
| "learning_rate": 6.50468137956049e-07, | |
| "loss": 0.3732, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 2.5619913834370514, | |
| "grad_norm": 0.1974366009235382, | |
| "learning_rate": 6.340467958674762e-07, | |
| "loss": 0.3608, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 2.5677357587362373, | |
| "grad_norm": 0.23682722449302673, | |
| "learning_rate": 6.178213524602061e-07, | |
| "loss": 0.396, | |
| "step": 447 | |
| }, | |
| { | |
| "epoch": 2.5734801340354236, | |
| "grad_norm": 0.21820193529129028, | |
| "learning_rate": 6.017925357637932e-07, | |
| "loss": 0.3567, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 2.57922450933461, | |
| "grad_norm": 0.22981062531471252, | |
| "learning_rate": 5.859610649852249e-07, | |
| "loss": 0.3642, | |
| "step": 449 | |
| }, | |
| { | |
| "epoch": 2.584968884633796, | |
| "grad_norm": 0.2145363688468933, | |
| "learning_rate": 5.703276504766514e-07, | |
| "loss": 0.3495, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 2.5907132599329823, | |
| "grad_norm": 0.19343282282352448, | |
| "learning_rate": 5.548929937035147e-07, | |
| "loss": 0.3339, | |
| "step": 451 | |
| }, | |
| { | |
| "epoch": 2.5964576352321687, | |
| "grad_norm": 0.20028981566429138, | |
| "learning_rate": 5.396577872130676e-07, | |
| "loss": 0.3781, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 2.6022020105313546, | |
| "grad_norm": 0.20916354656219482, | |
| "learning_rate": 5.246227146033089e-07, | |
| "loss": 0.398, | |
| "step": 453 | |
| }, | |
| { | |
| "epoch": 2.607946385830541, | |
| "grad_norm": 0.19891418516635895, | |
| "learning_rate": 5.097884504922996e-07, | |
| "loss": 0.3522, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 2.6136907611297273, | |
| "grad_norm": 0.22149665653705597, | |
| "learning_rate": 4.951556604879049e-07, | |
| "loss": 0.4384, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 2.6194351364289132, | |
| "grad_norm": 0.20566165447235107, | |
| "learning_rate": 4.807250011579168e-07, | |
| "loss": 0.3747, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 2.6251795117280996, | |
| "grad_norm": 0.20062044262886047, | |
| "learning_rate": 4.6649712000060297e-07, | |
| "loss": 0.3277, | |
| "step": 457 | |
| }, | |
| { | |
| "epoch": 2.6309238870272855, | |
| "grad_norm": 0.20748507976531982, | |
| "learning_rate": 4.5247265541564836e-07, | |
| "loss": 0.3733, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 2.636668262326472, | |
| "grad_norm": 0.20769307017326355, | |
| "learning_rate": 4.386522366755169e-07, | |
| "loss": 0.3715, | |
| "step": 459 | |
| }, | |
| { | |
| "epoch": 2.6424126376256583, | |
| "grad_norm": 0.21871528029441833, | |
| "learning_rate": 4.250364838972065e-07, | |
| "loss": 0.4001, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 2.6481570129248446, | |
| "grad_norm": 0.19270025193691254, | |
| "learning_rate": 4.116260080144352e-07, | |
| "loss": 0.3847, | |
| "step": 461 | |
| }, | |
| { | |
| "epoch": 2.6539013882240305, | |
| "grad_norm": 0.22346258163452148, | |
| "learning_rate": 3.98421410750221e-07, | |
| "loss": 0.3865, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 2.659645763523217, | |
| "grad_norm": 0.2189069241285324, | |
| "learning_rate": 3.854232845898859e-07, | |
| "loss": 0.3777, | |
| "step": 463 | |
| }, | |
| { | |
| "epoch": 2.665390138822403, | |
| "grad_norm": 0.19334660470485687, | |
| "learning_rate": 3.7263221275447125e-07, | |
| "loss": 0.3738, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 2.671134514121589, | |
| "grad_norm": 0.1997268944978714, | |
| "learning_rate": 3.60048769174568e-07, | |
| "loss": 0.3713, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 2.6768788894207756, | |
| "grad_norm": 0.21596620976924896, | |
| "learning_rate": 3.4767351846456744e-07, | |
| "loss": 0.3963, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 2.682623264719962, | |
| "grad_norm": 0.2167111337184906, | |
| "learning_rate": 3.355070158973212e-07, | |
| "loss": 0.3719, | |
| "step": 467 | |
| }, | |
| { | |
| "epoch": 2.688367640019148, | |
| "grad_norm": 0.22667430341243744, | |
| "learning_rate": 3.235498073792342e-07, | |
| "loss": 0.3777, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 2.694112015318334, | |
| "grad_norm": 0.22425274550914764, | |
| "learning_rate": 3.118024294257621e-07, | |
| "loss": 0.3934, | |
| "step": 469 | |
| }, | |
| { | |
| "epoch": 2.69985639061752, | |
| "grad_norm": 0.21314410865306854, | |
| "learning_rate": 3.002654091373453e-07, | |
| "loss": 0.3514, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 2.7056007659167065, | |
| "grad_norm": 0.20535998046398163, | |
| "learning_rate": 2.889392641757527e-07, | |
| "loss": 0.4027, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 2.711345141215893, | |
| "grad_norm": 0.19453689455986023, | |
| "learning_rate": 2.778245027408566e-07, | |
| "loss": 0.3619, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 2.7170895165150792, | |
| "grad_norm": 0.2257540374994278, | |
| "learning_rate": 2.669216235478295e-07, | |
| "loss": 0.4111, | |
| "step": 473 | |
| }, | |
| { | |
| "epoch": 2.722833891814265, | |
| "grad_norm": 0.22148896753787994, | |
| "learning_rate": 2.562311158047692e-07, | |
| "loss": 0.3853, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 2.7285782671134515, | |
| "grad_norm": 0.23967334628105164, | |
| "learning_rate": 2.45753459190744e-07, | |
| "loss": 0.3789, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 2.7343226424126374, | |
| "grad_norm": 0.21828240156173706, | |
| "learning_rate": 2.354891238342738e-07, | |
| "loss": 0.3635, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 2.740067017711824, | |
| "grad_norm": 0.22273699939250946, | |
| "learning_rate": 2.254385702922318e-07, | |
| "loss": 0.3866, | |
| "step": 477 | |
| }, | |
| { | |
| "epoch": 2.74581139301101, | |
| "grad_norm": 0.21624407172203064, | |
| "learning_rate": 2.1560224952918373e-07, | |
| "loss": 0.3334, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 2.751555768310196, | |
| "grad_norm": 0.20374171435832977, | |
| "learning_rate": 2.0598060289714893e-07, | |
| "loss": 0.333, | |
| "step": 479 | |
| }, | |
| { | |
| "epoch": 2.7573001436093825, | |
| "grad_norm": 0.21629025042057037, | |
| "learning_rate": 1.9657406211579966e-07, | |
| "loss": 0.4038, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 2.763044518908569, | |
| "grad_norm": 0.19769959151744843, | |
| "learning_rate": 1.8738304925308926e-07, | |
| "loss": 0.3912, | |
| "step": 481 | |
| }, | |
| { | |
| "epoch": 2.7687888942077548, | |
| "grad_norm": 0.20424911379814148, | |
| "learning_rate": 1.7840797670631572e-07, | |
| "loss": 0.3857, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 2.774533269506941, | |
| "grad_norm": 0.22209753096103668, | |
| "learning_rate": 1.6964924718361364e-07, | |
| "loss": 0.377, | |
| "step": 483 | |
| }, | |
| { | |
| "epoch": 2.7802776448061275, | |
| "grad_norm": 0.18916495144367218, | |
| "learning_rate": 1.6110725368589041e-07, | |
| "loss": 0.3567, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 2.7860220201053134, | |
| "grad_norm": 0.21668356657028198, | |
| "learning_rate": 1.5278237948918585e-07, | |
| "loss": 0.4076, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 2.7917663954045, | |
| "grad_norm": 0.20810151100158691, | |
| "learning_rate": 1.4467499812748143e-07, | |
| "loss": 0.3784, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 2.7975107707036857, | |
| "grad_norm": 0.2041124403476715, | |
| "learning_rate": 1.3678547337593494e-07, | |
| "loss": 0.3887, | |
| "step": 487 | |
| }, | |
| { | |
| "epoch": 2.803255146002872, | |
| "grad_norm": 0.24317069351673126, | |
| "learning_rate": 1.2911415923456017e-07, | |
| "loss": 0.3808, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 2.8089995213020584, | |
| "grad_norm": 0.19764302670955658, | |
| "learning_rate": 1.2166139991234227e-07, | |
| "loss": 0.3556, | |
| "step": 489 | |
| }, | |
| { | |
| "epoch": 2.814743896601245, | |
| "grad_norm": 0.22858253121376038, | |
| "learning_rate": 1.1442752981179527e-07, | |
| "loss": 0.3944, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 2.8204882719004307, | |
| "grad_norm": 0.20770110189914703, | |
| "learning_rate": 1.0741287351395402e-07, | |
| "loss": 0.4109, | |
| "step": 491 | |
| }, | |
| { | |
| "epoch": 2.826232647199617, | |
| "grad_norm": 0.20658861100673676, | |
| "learning_rate": 1.0061774576381411e-07, | |
| "loss": 0.375, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 2.831977022498803, | |
| "grad_norm": 0.21842309832572937, | |
| "learning_rate": 9.404245145620717e-08, | |
| "loss": 0.4091, | |
| "step": 493 | |
| }, | |
| { | |
| "epoch": 2.8377213977979894, | |
| "grad_norm": 0.20370720326900482, | |
| "learning_rate": 8.768728562211948e-08, | |
| "loss": 0.3252, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 2.8434657730971757, | |
| "grad_norm": 0.20084549486637115, | |
| "learning_rate": 8.155253341545655e-08, | |
| "loss": 0.4111, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 2.849210148396362, | |
| "grad_norm": 0.23278358578681946, | |
| "learning_rate": 7.563847010024716e-08, | |
| "loss": 0.3762, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 2.854954523695548, | |
| "grad_norm": 0.20328493416309357, | |
| "learning_rate": 6.994536103829164e-08, | |
| "loss": 0.3765, | |
| "step": 497 | |
| }, | |
| { | |
| "epoch": 2.8606988989947344, | |
| "grad_norm": 0.20402538776397705, | |
| "learning_rate": 6.447346167725688e-08, | |
| "loss": 0.3449, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 2.8664432742939203, | |
| "grad_norm": 0.22473885118961334, | |
| "learning_rate": 5.9223017539213335e-08, | |
| "loss": 0.4115, | |
| "step": 499 | |
| }, | |
| { | |
| "epoch": 2.8721876495931067, | |
| "grad_norm": 0.18734395503997803, | |
| "learning_rate": 5.4194264209617705e-08, | |
| "loss": 0.3474, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 2.877932024892293, | |
| "grad_norm": 0.2044266164302826, | |
| "learning_rate": 4.9387427326745287e-08, | |
| "loss": 0.3939, | |
| "step": 501 | |
| }, | |
| { | |
| "epoch": 2.8836764001914794, | |
| "grad_norm": 0.21485216915607452, | |
| "learning_rate": 4.4802722571561374e-08, | |
| "loss": 0.4016, | |
| "step": 502 | |
| }, | |
| { | |
| "epoch": 2.8894207754906653, | |
| "grad_norm": 0.22615912556648254, | |
| "learning_rate": 4.044035565804793e-08, | |
| "loss": 0.3599, | |
| "step": 503 | |
| }, | |
| { | |
| "epoch": 2.8951651507898517, | |
| "grad_norm": 0.2036847621202469, | |
| "learning_rate": 3.6300522323969855e-08, | |
| "loss": 0.3266, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 2.9009095260890376, | |
| "grad_norm": 0.2017732411623001, | |
| "learning_rate": 3.2383408322095856e-08, | |
| "loss": 0.4101, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 2.906653901388224, | |
| "grad_norm": 0.19980326294898987, | |
| "learning_rate": 2.8689189411859607e-08, | |
| "loss": 0.3482, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 2.9123982766874104, | |
| "grad_norm": 0.221945121884346, | |
| "learning_rate": 2.5218031351478268e-08, | |
| "loss": 0.4437, | |
| "step": 507 | |
| }, | |
| { | |
| "epoch": 2.9181426519865967, | |
| "grad_norm": 0.2403242588043213, | |
| "learning_rate": 2.1970089890509527e-08, | |
| "loss": 0.3528, | |
| "step": 508 | |
| }, | |
| { | |
| "epoch": 2.9238870272857826, | |
| "grad_norm": 0.22222495079040527, | |
| "learning_rate": 1.8945510762868325e-08, | |
| "loss": 0.3765, | |
| "step": 509 | |
| }, | |
| { | |
| "epoch": 2.929631402584969, | |
| "grad_norm": 0.20567429065704346, | |
| "learning_rate": 1.614442968028429e-08, | |
| "loss": 0.3844, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 2.935375777884155, | |
| "grad_norm": 0.1993645876646042, | |
| "learning_rate": 1.3566972326214956e-08, | |
| "loss": 0.3712, | |
| "step": 511 | |
| }, | |
| { | |
| "epoch": 2.9411201531833413, | |
| "grad_norm": 0.21140974760055542, | |
| "learning_rate": 1.1213254350202486e-08, | |
| "loss": 0.357, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 2.9468645284825277, | |
| "grad_norm": 0.20113182067871094, | |
| "learning_rate": 9.083381362690603e-09, | |
| "loss": 0.3692, | |
| "step": 513 | |
| }, | |
| { | |
| "epoch": 2.9526089037817136, | |
| "grad_norm": 0.2018483579158783, | |
| "learning_rate": 7.177448930279496e-09, | |
| "loss": 0.3817, | |
| "step": 514 | |
| }, | |
| { | |
| "epoch": 2.9583532790809, | |
| "grad_norm": 0.21870052814483643, | |
| "learning_rate": 5.495542571443135e-09, | |
| "loss": 0.3505, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 2.9640976543800863, | |
| "grad_norm": 0.2133331298828125, | |
| "learning_rate": 4.037737752686788e-09, | |
| "loss": 0.3767, | |
| "step": 516 | |
| }, | |
| { | |
| "epoch": 2.9698420296792722, | |
| "grad_norm": 0.19916494190692902, | |
| "learning_rate": 2.8040998851674996e-09, | |
| "loss": 0.374, | |
| "step": 517 | |
| }, | |
| { | |
| "epoch": 2.9755864049784586, | |
| "grad_norm": 0.2075948715209961, | |
| "learning_rate": 1.7946843217514498e-09, | |
| "loss": 0.3572, | |
| "step": 518 | |
| }, | |
| { | |
| "epoch": 2.981330780277645, | |
| "grad_norm": 0.20564815402030945, | |
| "learning_rate": 1.009536354537044e-09, | |
| "loss": 0.3296, | |
| "step": 519 | |
| }, | |
| { | |
| "epoch": 2.987075155576831, | |
| "grad_norm": 0.22070202231407166, | |
| "learning_rate": 4.486912128182086e-10, | |
| "loss": 0.4276, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 2.9928195308760173, | |
| "grad_norm": 0.19630590081214905, | |
| "learning_rate": 1.1217406150676457e-10, | |
| "loss": 0.3737, | |
| "step": 521 | |
| }, | |
| { | |
| "epoch": 2.998563906175203, | |
| "grad_norm": 0.20596085488796234, | |
| "learning_rate": 0.0, | |
| "loss": 0.3589, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 2.998563906175203, | |
| "step": 522, | |
| "total_flos": 1806937112969216.0, | |
| "train_loss": 0.43391252791516166, | |
| "train_runtime": 96882.8659, | |
| "train_samples_per_second": 0.517, | |
| "train_steps_per_second": 0.005 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 522, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 100, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1806937112969216.0, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |