| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.0, | |
| "eval_steps": 500, | |
| "global_step": 1052, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0019011406844106464, | |
| "grad_norm": 3.598050512245097, | |
| "learning_rate": 1.886792452830189e-07, | |
| "loss": 0.7008, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0038022813688212928, | |
| "grad_norm": 3.222735789322097, | |
| "learning_rate": 3.773584905660378e-07, | |
| "loss": 0.7132, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.005703422053231939, | |
| "grad_norm": 3.4043061777667836, | |
| "learning_rate": 5.660377358490567e-07, | |
| "loss": 0.7039, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.0076045627376425855, | |
| "grad_norm": 3.115345054078698, | |
| "learning_rate": 7.547169811320755e-07, | |
| "loss": 0.708, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.009505703422053232, | |
| "grad_norm": 3.427953314405625, | |
| "learning_rate": 9.433962264150944e-07, | |
| "loss": 0.7407, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.011406844106463879, | |
| "grad_norm": 3.0048680622712904, | |
| "learning_rate": 1.1320754716981133e-06, | |
| "loss": 0.6937, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.013307984790874524, | |
| "grad_norm": 3.149975323720877, | |
| "learning_rate": 1.3207547169811322e-06, | |
| "loss": 0.6933, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.015209125475285171, | |
| "grad_norm": 3.020574777751727, | |
| "learning_rate": 1.509433962264151e-06, | |
| "loss": 0.7245, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.017110266159695818, | |
| "grad_norm": 2.544753250834972, | |
| "learning_rate": 1.6981132075471698e-06, | |
| "loss": 0.6823, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.019011406844106463, | |
| "grad_norm": 2.386769229408845, | |
| "learning_rate": 1.8867924528301889e-06, | |
| "loss": 0.6793, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.02091254752851711, | |
| "grad_norm": 1.650172024780255, | |
| "learning_rate": 2.075471698113208e-06, | |
| "loss": 0.6568, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.022813688212927757, | |
| "grad_norm": 1.6418423155158648, | |
| "learning_rate": 2.2641509433962266e-06, | |
| "loss": 0.6026, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.024714828897338403, | |
| "grad_norm": 1.636093411749038, | |
| "learning_rate": 2.4528301886792453e-06, | |
| "loss": 0.6187, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.026615969581749048, | |
| "grad_norm": 1.4222794451135146, | |
| "learning_rate": 2.6415094339622644e-06, | |
| "loss": 0.5827, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.028517110266159697, | |
| "grad_norm": 1.8250450428140685, | |
| "learning_rate": 2.830188679245283e-06, | |
| "loss": 0.5992, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.030418250950570342, | |
| "grad_norm": 2.2561507778040966, | |
| "learning_rate": 3.018867924528302e-06, | |
| "loss": 0.5952, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.03231939163498099, | |
| "grad_norm": 2.3599660909284257, | |
| "learning_rate": 3.207547169811321e-06, | |
| "loss": 0.5797, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.034220532319391636, | |
| "grad_norm": 1.816283393729647, | |
| "learning_rate": 3.3962264150943395e-06, | |
| "loss": 0.5837, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.03612167300380228, | |
| "grad_norm": 1.558929649084608, | |
| "learning_rate": 3.5849056603773586e-06, | |
| "loss": 0.5578, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.03802281368821293, | |
| "grad_norm": 1.2328549776007172, | |
| "learning_rate": 3.7735849056603777e-06, | |
| "loss": 0.5611, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.039923954372623575, | |
| "grad_norm": 1.3698571744798604, | |
| "learning_rate": 3.962264150943396e-06, | |
| "loss": 0.5498, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.04182509505703422, | |
| "grad_norm": 1.4073837305383299, | |
| "learning_rate": 4.150943396226416e-06, | |
| "loss": 0.5934, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.043726235741444866, | |
| "grad_norm": 1.366486048929682, | |
| "learning_rate": 4.339622641509435e-06, | |
| "loss": 0.552, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.045627376425855515, | |
| "grad_norm": 1.1693078877710685, | |
| "learning_rate": 4.528301886792453e-06, | |
| "loss": 0.5546, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.04752851711026616, | |
| "grad_norm": 1.0306525847187065, | |
| "learning_rate": 4.716981132075472e-06, | |
| "loss": 0.5495, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.049429657794676805, | |
| "grad_norm": 1.0652699379717747, | |
| "learning_rate": 4.905660377358491e-06, | |
| "loss": 0.5567, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.051330798479087454, | |
| "grad_norm": 1.1288318750095117, | |
| "learning_rate": 5.09433962264151e-06, | |
| "loss": 0.5456, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.053231939163498096, | |
| "grad_norm": 1.0305076618323628, | |
| "learning_rate": 5.283018867924529e-06, | |
| "loss": 0.5381, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.055133079847908745, | |
| "grad_norm": 1.0513217010366536, | |
| "learning_rate": 5.4716981132075475e-06, | |
| "loss": 0.5197, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.057034220532319393, | |
| "grad_norm": 0.8873862190154846, | |
| "learning_rate": 5.660377358490566e-06, | |
| "loss": 0.5082, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.058935361216730035, | |
| "grad_norm": 0.9197339521107429, | |
| "learning_rate": 5.849056603773585e-06, | |
| "loss": 0.5345, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.060836501901140684, | |
| "grad_norm": 0.90825774894367, | |
| "learning_rate": 6.037735849056604e-06, | |
| "loss": 0.4996, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.06273764258555133, | |
| "grad_norm": 0.8984940790409877, | |
| "learning_rate": 6.226415094339623e-06, | |
| "loss": 0.51, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.06463878326996197, | |
| "grad_norm": 0.9415933669252435, | |
| "learning_rate": 6.415094339622642e-06, | |
| "loss": 0.5199, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.06653992395437262, | |
| "grad_norm": 1.0302526669350698, | |
| "learning_rate": 6.60377358490566e-06, | |
| "loss": 0.5366, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.06844106463878327, | |
| "grad_norm": 0.9084013093908937, | |
| "learning_rate": 6.792452830188679e-06, | |
| "loss": 0.5033, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.07034220532319392, | |
| "grad_norm": 0.8650785916212774, | |
| "learning_rate": 6.981132075471699e-06, | |
| "loss": 0.4695, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.07224334600760456, | |
| "grad_norm": 0.9116002715575098, | |
| "learning_rate": 7.169811320754717e-06, | |
| "loss": 0.5263, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.0741444866920152, | |
| "grad_norm": 0.9029623814193115, | |
| "learning_rate": 7.358490566037736e-06, | |
| "loss": 0.5375, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.07604562737642585, | |
| "grad_norm": 0.8587793106124799, | |
| "learning_rate": 7.5471698113207555e-06, | |
| "loss": 0.473, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.0779467680608365, | |
| "grad_norm": 0.8798818833396722, | |
| "learning_rate": 7.735849056603775e-06, | |
| "loss": 0.5157, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.07984790874524715, | |
| "grad_norm": 0.7801648412434248, | |
| "learning_rate": 7.924528301886793e-06, | |
| "loss": 0.4773, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.0817490494296578, | |
| "grad_norm": 0.8905633818275244, | |
| "learning_rate": 8.113207547169812e-06, | |
| "loss": 0.476, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.08365019011406843, | |
| "grad_norm": 0.8123040933518004, | |
| "learning_rate": 8.301886792452832e-06, | |
| "loss": 0.464, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.08555133079847908, | |
| "grad_norm": 0.833827324168966, | |
| "learning_rate": 8.49056603773585e-06, | |
| "loss": 0.4882, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.08745247148288973, | |
| "grad_norm": 0.8564193334198396, | |
| "learning_rate": 8.67924528301887e-06, | |
| "loss": 0.4824, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.08935361216730038, | |
| "grad_norm": 0.8131532198096886, | |
| "learning_rate": 8.867924528301887e-06, | |
| "loss": 0.4981, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.09125475285171103, | |
| "grad_norm": 0.840301232922172, | |
| "learning_rate": 9.056603773584907e-06, | |
| "loss": 0.4906, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.09315589353612168, | |
| "grad_norm": 0.8265081752458253, | |
| "learning_rate": 9.245283018867926e-06, | |
| "loss": 0.4975, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.09505703422053231, | |
| "grad_norm": 0.7966849143697176, | |
| "learning_rate": 9.433962264150944e-06, | |
| "loss": 0.5035, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.09695817490494296, | |
| "grad_norm": 0.8318050996554929, | |
| "learning_rate": 9.622641509433963e-06, | |
| "loss": 0.4825, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.09885931558935361, | |
| "grad_norm": 0.8403828914754736, | |
| "learning_rate": 9.811320754716981e-06, | |
| "loss": 0.5005, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.10076045627376426, | |
| "grad_norm": 0.9627733443688251, | |
| "learning_rate": 1e-05, | |
| "loss": 0.4862, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.10266159695817491, | |
| "grad_norm": 0.9247047149546374, | |
| "learning_rate": 1.018867924528302e-05, | |
| "loss": 0.4792, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.10456273764258556, | |
| "grad_norm": 0.8899276303618266, | |
| "learning_rate": 1.0377358490566038e-05, | |
| "loss": 0.486, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.10646387832699619, | |
| "grad_norm": 0.8547648145033194, | |
| "learning_rate": 1.0566037735849058e-05, | |
| "loss": 0.4785, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.10836501901140684, | |
| "grad_norm": 1.031691560957074, | |
| "learning_rate": 1.0754716981132076e-05, | |
| "loss": 0.4629, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.11026615969581749, | |
| "grad_norm": 0.9089724306418459, | |
| "learning_rate": 1.0943396226415095e-05, | |
| "loss": 0.476, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.11216730038022814, | |
| "grad_norm": 1.0707381450328022, | |
| "learning_rate": 1.1132075471698115e-05, | |
| "loss": 0.4834, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.11406844106463879, | |
| "grad_norm": 0.9949120156467437, | |
| "learning_rate": 1.1320754716981132e-05, | |
| "loss": 0.4955, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.11596958174904944, | |
| "grad_norm": 0.8113558860123722, | |
| "learning_rate": 1.1509433962264152e-05, | |
| "loss": 0.4575, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.11787072243346007, | |
| "grad_norm": 1.1718599720447724, | |
| "learning_rate": 1.169811320754717e-05, | |
| "loss": 0.4789, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.11977186311787072, | |
| "grad_norm": 0.8719308223497783, | |
| "learning_rate": 1.188679245283019e-05, | |
| "loss": 0.4587, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.12167300380228137, | |
| "grad_norm": 0.9582260008521769, | |
| "learning_rate": 1.2075471698113209e-05, | |
| "loss": 0.5159, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.12357414448669202, | |
| "grad_norm": 0.9219641777374403, | |
| "learning_rate": 1.2264150943396227e-05, | |
| "loss": 0.48, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.12547528517110265, | |
| "grad_norm": 0.8741327580988341, | |
| "learning_rate": 1.2452830188679246e-05, | |
| "loss": 0.469, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.12737642585551331, | |
| "grad_norm": 0.8879883821498404, | |
| "learning_rate": 1.2641509433962264e-05, | |
| "loss": 0.4626, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.12927756653992395, | |
| "grad_norm": 0.8983624330010218, | |
| "learning_rate": 1.2830188679245283e-05, | |
| "loss": 0.4748, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.1311787072243346, | |
| "grad_norm": 1.023876542390637, | |
| "learning_rate": 1.3018867924528303e-05, | |
| "loss": 0.5029, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.13307984790874525, | |
| "grad_norm": 0.8580232869489458, | |
| "learning_rate": 1.320754716981132e-05, | |
| "loss": 0.4741, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.13498098859315588, | |
| "grad_norm": 0.8932503673893808, | |
| "learning_rate": 1.339622641509434e-05, | |
| "loss": 0.4789, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.13688212927756654, | |
| "grad_norm": 1.211759664237497, | |
| "learning_rate": 1.3584905660377358e-05, | |
| "loss": 0.4943, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.13878326996197718, | |
| "grad_norm": 0.8804748478498138, | |
| "learning_rate": 1.3773584905660378e-05, | |
| "loss": 0.4705, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.14068441064638784, | |
| "grad_norm": 1.011622605170241, | |
| "learning_rate": 1.3962264150943397e-05, | |
| "loss": 0.4494, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.14258555133079848, | |
| "grad_norm": 0.9004993750763927, | |
| "learning_rate": 1.4150943396226415e-05, | |
| "loss": 0.4624, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.1444866920152091, | |
| "grad_norm": 0.7956792694064247, | |
| "learning_rate": 1.4339622641509435e-05, | |
| "loss": 0.4808, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.14638783269961977, | |
| "grad_norm": 0.9842331250602806, | |
| "learning_rate": 1.4528301886792452e-05, | |
| "loss": 0.4882, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.1482889733840304, | |
| "grad_norm": 0.85179215030062, | |
| "learning_rate": 1.4716981132075472e-05, | |
| "loss": 0.4583, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.15019011406844107, | |
| "grad_norm": 0.9269990431217079, | |
| "learning_rate": 1.4905660377358491e-05, | |
| "loss": 0.4942, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.1520912547528517, | |
| "grad_norm": 0.9073116710795202, | |
| "learning_rate": 1.5094339622641511e-05, | |
| "loss": 0.484, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.15399239543726237, | |
| "grad_norm": 0.8476204081356586, | |
| "learning_rate": 1.5283018867924532e-05, | |
| "loss": 0.4785, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.155893536121673, | |
| "grad_norm": 0.8599202675175366, | |
| "learning_rate": 1.547169811320755e-05, | |
| "loss": 0.4591, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.15779467680608364, | |
| "grad_norm": 0.8641475590202745, | |
| "learning_rate": 1.5660377358490568e-05, | |
| "loss": 0.4691, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.1596958174904943, | |
| "grad_norm": 0.842333986456053, | |
| "learning_rate": 1.5849056603773586e-05, | |
| "loss": 0.4807, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.16159695817490494, | |
| "grad_norm": 0.8466176923787454, | |
| "learning_rate": 1.6037735849056607e-05, | |
| "loss": 0.4699, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.1634980988593156, | |
| "grad_norm": 0.9621768416052041, | |
| "learning_rate": 1.6226415094339625e-05, | |
| "loss": 0.4571, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.16539923954372623, | |
| "grad_norm": 0.8778035741384473, | |
| "learning_rate": 1.6415094339622643e-05, | |
| "loss": 0.4728, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.16730038022813687, | |
| "grad_norm": 0.9530578688086553, | |
| "learning_rate": 1.6603773584905664e-05, | |
| "loss": 0.4852, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.16920152091254753, | |
| "grad_norm": 0.878964203485296, | |
| "learning_rate": 1.679245283018868e-05, | |
| "loss": 0.4794, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.17110266159695817, | |
| "grad_norm": 0.9494945608604918, | |
| "learning_rate": 1.69811320754717e-05, | |
| "loss": 0.476, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.17300380228136883, | |
| "grad_norm": 1.0011134094245464, | |
| "learning_rate": 1.716981132075472e-05, | |
| "loss": 0.459, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.17490494296577946, | |
| "grad_norm": 0.9295830798016033, | |
| "learning_rate": 1.735849056603774e-05, | |
| "loss": 0.4856, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.17680608365019013, | |
| "grad_norm": 0.9102596357028369, | |
| "learning_rate": 1.7547169811320756e-05, | |
| "loss": 0.474, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.17870722433460076, | |
| "grad_norm": 0.9473784221597169, | |
| "learning_rate": 1.7735849056603774e-05, | |
| "loss": 0.4827, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.1806083650190114, | |
| "grad_norm": 0.9886128023581761, | |
| "learning_rate": 1.7924528301886795e-05, | |
| "loss": 0.4824, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.18250950570342206, | |
| "grad_norm": 0.8726410005744043, | |
| "learning_rate": 1.8113207547169813e-05, | |
| "loss": 0.4636, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.1844106463878327, | |
| "grad_norm": 0.9672983519070151, | |
| "learning_rate": 1.830188679245283e-05, | |
| "loss": 0.472, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.18631178707224336, | |
| "grad_norm": 0.9283823986259061, | |
| "learning_rate": 1.8490566037735852e-05, | |
| "loss": 0.4307, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.188212927756654, | |
| "grad_norm": 0.8774609335752139, | |
| "learning_rate": 1.867924528301887e-05, | |
| "loss": 0.471, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.19011406844106463, | |
| "grad_norm": 0.8932072760913167, | |
| "learning_rate": 1.8867924528301888e-05, | |
| "loss": 0.4744, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.1920152091254753, | |
| "grad_norm": 0.9836811644730346, | |
| "learning_rate": 1.905660377358491e-05, | |
| "loss": 0.5015, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.19391634980988592, | |
| "grad_norm": 0.8600609508730563, | |
| "learning_rate": 1.9245283018867927e-05, | |
| "loss": 0.4586, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.1958174904942966, | |
| "grad_norm": 0.9906751714047475, | |
| "learning_rate": 1.9433962264150945e-05, | |
| "loss": 0.4891, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.19771863117870722, | |
| "grad_norm": 0.8624735722376163, | |
| "learning_rate": 1.9622641509433963e-05, | |
| "loss": 0.4499, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.19961977186311788, | |
| "grad_norm": 0.9865747819597135, | |
| "learning_rate": 1.9811320754716984e-05, | |
| "loss": 0.4703, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.20152091254752852, | |
| "grad_norm": 1.0461627483887608, | |
| "learning_rate": 2e-05, | |
| "loss": 0.4707, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.20342205323193915, | |
| "grad_norm": 0.9829746725364998, | |
| "learning_rate": 1.9999944857420527e-05, | |
| "loss": 0.4549, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.20532319391634982, | |
| "grad_norm": 0.888493065537834, | |
| "learning_rate": 1.9999779430290247e-05, | |
| "loss": 0.4783, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.20722433460076045, | |
| "grad_norm": 1.1068482956474874, | |
| "learning_rate": 1.9999503720433575e-05, | |
| "loss": 0.4877, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.20912547528517111, | |
| "grad_norm": 0.9569473619666661, | |
| "learning_rate": 1.999911773089118e-05, | |
| "loss": 0.4806, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.21102661596958175, | |
| "grad_norm": 0.979736028181881, | |
| "learning_rate": 1.999862146591996e-05, | |
| "loss": 0.4509, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.21292775665399238, | |
| "grad_norm": 0.8769396900268799, | |
| "learning_rate": 1.9998014930992976e-05, | |
| "loss": 0.4585, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.21482889733840305, | |
| "grad_norm": 0.992114936596131, | |
| "learning_rate": 1.9997298132799408e-05, | |
| "loss": 0.4844, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.21673003802281368, | |
| "grad_norm": 0.8951932916770349, | |
| "learning_rate": 1.9996471079244477e-05, | |
| "loss": 0.4866, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.21863117870722434, | |
| "grad_norm": 0.8954599005047487, | |
| "learning_rate": 1.999553377944936e-05, | |
| "loss": 0.4517, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.22053231939163498, | |
| "grad_norm": 0.8957633475639653, | |
| "learning_rate": 1.9994486243751076e-05, | |
| "loss": 0.4731, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.2224334600760456, | |
| "grad_norm": 0.8305802234979098, | |
| "learning_rate": 1.9993328483702393e-05, | |
| "loss": 0.4728, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.22433460076045628, | |
| "grad_norm": 1.0005758201469903, | |
| "learning_rate": 1.999206051207169e-05, | |
| "loss": 0.4691, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.2262357414448669, | |
| "grad_norm": 0.9360355297134823, | |
| "learning_rate": 1.9990682342842805e-05, | |
| "loss": 0.4868, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.22813688212927757, | |
| "grad_norm": 0.9599743182141269, | |
| "learning_rate": 1.99891939912149e-05, | |
| "loss": 0.4426, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.2300380228136882, | |
| "grad_norm": 0.9339692499906854, | |
| "learning_rate": 1.9987595473602292e-05, | |
| "loss": 0.476, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.23193916349809887, | |
| "grad_norm": 1.0388460956106484, | |
| "learning_rate": 1.9985886807634246e-05, | |
| "loss": 0.5048, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.2338403041825095, | |
| "grad_norm": 0.8392082779200816, | |
| "learning_rate": 1.9984068012154824e-05, | |
| "loss": 0.455, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.23574144486692014, | |
| "grad_norm": 0.8936571870811483, | |
| "learning_rate": 1.9982139107222634e-05, | |
| "loss": 0.4634, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.2376425855513308, | |
| "grad_norm": 0.9578146683967285, | |
| "learning_rate": 1.9980100114110637e-05, | |
| "loss": 0.4619, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.23954372623574144, | |
| "grad_norm": 0.8936648916445861, | |
| "learning_rate": 1.99779510553059e-05, | |
| "loss": 0.4593, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.2414448669201521, | |
| "grad_norm": 1.1160196451914657, | |
| "learning_rate": 1.9975691954509347e-05, | |
| "loss": 0.4883, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.24334600760456274, | |
| "grad_norm": 0.8333646075556067, | |
| "learning_rate": 1.9973322836635517e-05, | |
| "loss": 0.4773, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.24524714828897337, | |
| "grad_norm": 0.9327645169099568, | |
| "learning_rate": 1.997084372781226e-05, | |
| "loss": 0.5083, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.24714828897338403, | |
| "grad_norm": 0.9377344493674896, | |
| "learning_rate": 1.9968254655380465e-05, | |
| "loss": 0.4674, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.24904942965779467, | |
| "grad_norm": 0.9943797725214091, | |
| "learning_rate": 1.996555564789376e-05, | |
| "loss": 0.4653, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.2509505703422053, | |
| "grad_norm": 0.8565817113547802, | |
| "learning_rate": 1.996274673511819e-05, | |
| "loss": 0.4383, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.25285171102661597, | |
| "grad_norm": 0.7945448471326897, | |
| "learning_rate": 1.99598279480319e-05, | |
| "loss": 0.4555, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.25475285171102663, | |
| "grad_norm": 0.9070676499449211, | |
| "learning_rate": 1.9956799318824776e-05, | |
| "loss": 0.4435, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.25665399239543724, | |
| "grad_norm": 0.8344360110769314, | |
| "learning_rate": 1.99536608808981e-05, | |
| "loss": 0.4604, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.2585551330798479, | |
| "grad_norm": 0.8598192641853587, | |
| "learning_rate": 1.995041266886419e-05, | |
| "loss": 0.4776, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.26045627376425856, | |
| "grad_norm": 0.811119813066209, | |
| "learning_rate": 1.9947054718545996e-05, | |
| "loss": 0.4754, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.2623574144486692, | |
| "grad_norm": 0.8197155147312594, | |
| "learning_rate": 1.994358706697674e-05, | |
| "loss": 0.4666, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.26425855513307983, | |
| "grad_norm": 0.8018884574989606, | |
| "learning_rate": 1.9940009752399462e-05, | |
| "loss": 0.459, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.2661596958174905, | |
| "grad_norm": 0.8517022979530432, | |
| "learning_rate": 1.9936322814266634e-05, | |
| "loss": 0.4746, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.26806083650190116, | |
| "grad_norm": 0.8121330617395145, | |
| "learning_rate": 1.9932526293239713e-05, | |
| "loss": 0.4769, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.26996197718631176, | |
| "grad_norm": 0.7920947873626231, | |
| "learning_rate": 1.9928620231188694e-05, | |
| "loss": 0.4479, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.2718631178707224, | |
| "grad_norm": 0.7964597777270268, | |
| "learning_rate": 1.992460467119164e-05, | |
| "loss": 0.4738, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.2737642585551331, | |
| "grad_norm": 0.776214376258977, | |
| "learning_rate": 1.992047965753422e-05, | |
| "loss": 0.4879, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.27566539923954375, | |
| "grad_norm": 0.7895946064380274, | |
| "learning_rate": 1.991624523570922e-05, | |
| "loss": 0.4925, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.27756653992395436, | |
| "grad_norm": 0.7852600559045596, | |
| "learning_rate": 1.9911901452416012e-05, | |
| "loss": 0.4578, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.279467680608365, | |
| "grad_norm": 0.7771047616533184, | |
| "learning_rate": 1.9907448355560094e-05, | |
| "loss": 0.4515, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.2813688212927757, | |
| "grad_norm": 0.8035671970897588, | |
| "learning_rate": 1.9902885994252506e-05, | |
| "loss": 0.456, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.2832699619771863, | |
| "grad_norm": 0.8100035586208262, | |
| "learning_rate": 1.989821441880933e-05, | |
| "loss": 0.4728, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.28517110266159695, | |
| "grad_norm": 0.8169806110583785, | |
| "learning_rate": 1.9893433680751105e-05, | |
| "loss": 0.4589, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.2870722433460076, | |
| "grad_norm": 0.7188719907313003, | |
| "learning_rate": 1.9888543832802277e-05, | |
| "loss": 0.4355, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.2889733840304182, | |
| "grad_norm": 0.804774903120944, | |
| "learning_rate": 1.9883544928890612e-05, | |
| "loss": 0.4738, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.2908745247148289, | |
| "grad_norm": 0.8865858954793044, | |
| "learning_rate": 1.9878437024146603e-05, | |
| "loss": 0.4987, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.29277566539923955, | |
| "grad_norm": 0.8687579384020276, | |
| "learning_rate": 1.9873220174902857e-05, | |
| "loss": 0.4593, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.2946768060836502, | |
| "grad_norm": 0.8618810489244619, | |
| "learning_rate": 1.986789443869348e-05, | |
| "loss": 0.4765, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.2965779467680608, | |
| "grad_norm": 0.7946092111002868, | |
| "learning_rate": 1.9862459874253438e-05, | |
| "loss": 0.4416, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.2984790874524715, | |
| "grad_norm": 0.8546189063067687, | |
| "learning_rate": 1.985691654151791e-05, | |
| "loss": 0.4474, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.30038022813688214, | |
| "grad_norm": 0.847773543073196, | |
| "learning_rate": 1.9851264501621635e-05, | |
| "loss": 0.4638, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.30228136882129275, | |
| "grad_norm": 0.8409675125740901, | |
| "learning_rate": 1.984550381689822e-05, | |
| "loss": 0.457, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.3041825095057034, | |
| "grad_norm": 0.8023469360674511, | |
| "learning_rate": 1.983963455087946e-05, | |
| "loss": 0.4579, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.3060836501901141, | |
| "grad_norm": 0.8154473976201643, | |
| "learning_rate": 1.983365676829466e-05, | |
| "loss": 0.4595, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.30798479087452474, | |
| "grad_norm": 0.8106057446420789, | |
| "learning_rate": 1.982757053506989e-05, | |
| "loss": 0.452, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.30988593155893535, | |
| "grad_norm": 0.7873998704531001, | |
| "learning_rate": 1.9821375918327268e-05, | |
| "loss": 0.4834, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.311787072243346, | |
| "grad_norm": 0.7714296647926087, | |
| "learning_rate": 1.981507298638422e-05, | |
| "loss": 0.4441, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.31368821292775667, | |
| "grad_norm": 0.8671179015870749, | |
| "learning_rate": 1.9808661808752735e-05, | |
| "loss": 0.4865, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.3155893536121673, | |
| "grad_norm": 0.8232650135913239, | |
| "learning_rate": 1.980214245613858e-05, | |
| "loss": 0.4536, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.31749049429657794, | |
| "grad_norm": 0.8634430589338169, | |
| "learning_rate": 1.979551500044055e-05, | |
| "loss": 0.4645, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.3193916349809886, | |
| "grad_norm": 0.7906356784769415, | |
| "learning_rate": 1.9788779514749635e-05, | |
| "loss": 0.4631, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.32129277566539927, | |
| "grad_norm": 0.7639558306583075, | |
| "learning_rate": 1.978193607334826e-05, | |
| "loss": 0.4532, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.3231939163498099, | |
| "grad_norm": 0.7045630459237094, | |
| "learning_rate": 1.977498475170941e-05, | |
| "loss": 0.4692, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.32509505703422054, | |
| "grad_norm": 0.7749376875822258, | |
| "learning_rate": 1.9767925626495857e-05, | |
| "loss": 0.4739, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.3269961977186312, | |
| "grad_norm": 0.7809190833820437, | |
| "learning_rate": 1.9760758775559275e-05, | |
| "loss": 0.4702, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.3288973384030418, | |
| "grad_norm": 0.8493336406396681, | |
| "learning_rate": 1.975348427793939e-05, | |
| "loss": 0.476, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.33079847908745247, | |
| "grad_norm": 1.4043169138862521, | |
| "learning_rate": 1.9746102213863113e-05, | |
| "loss": 0.4594, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.33269961977186313, | |
| "grad_norm": 1.119352907197034, | |
| "learning_rate": 1.973861266474366e-05, | |
| "loss": 0.4533, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.33460076045627374, | |
| "grad_norm": 1.1001157489832107, | |
| "learning_rate": 1.9731015713179643e-05, | |
| "loss": 0.4692, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.3365019011406844, | |
| "grad_norm": 0.9683350709825659, | |
| "learning_rate": 1.9723311442954163e-05, | |
| "loss": 0.4716, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.33840304182509506, | |
| "grad_norm": 1.1798306220724755, | |
| "learning_rate": 1.9715499939033883e-05, | |
| "loss": 0.4493, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.3403041825095057, | |
| "grad_norm": 0.7438294133881727, | |
| "learning_rate": 1.9707581287568094e-05, | |
| "loss": 0.4345, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.34220532319391633, | |
| "grad_norm": 1.1263804540140947, | |
| "learning_rate": 1.969955557588778e-05, | |
| "loss": 0.4412, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.344106463878327, | |
| "grad_norm": 0.8281149064993888, | |
| "learning_rate": 1.9691422892504626e-05, | |
| "loss": 0.4373, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.34600760456273766, | |
| "grad_norm": 0.9169416205299551, | |
| "learning_rate": 1.968318332711006e-05, | |
| "loss": 0.4826, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.34790874524714827, | |
| "grad_norm": 0.9541633121419573, | |
| "learning_rate": 1.9674836970574253e-05, | |
| "loss": 0.4699, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.34980988593155893, | |
| "grad_norm": 0.8129052918132557, | |
| "learning_rate": 1.966638391494514e-05, | |
| "loss": 0.4672, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.3517110266159696, | |
| "grad_norm": 0.9304786890457217, | |
| "learning_rate": 1.9657824253447378e-05, | |
| "loss": 0.4396, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.35361216730038025, | |
| "grad_norm": 0.8245223893709127, | |
| "learning_rate": 1.9649158080481327e-05, | |
| "loss": 0.4632, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.35551330798479086, | |
| "grad_norm": 1.056255037406037, | |
| "learning_rate": 1.964038549162201e-05, | |
| "loss": 0.4413, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.3574144486692015, | |
| "grad_norm": 0.8515738728771085, | |
| "learning_rate": 1.963150658361807e-05, | |
| "loss": 0.466, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.3593155893536122, | |
| "grad_norm": 0.8958945543039234, | |
| "learning_rate": 1.962252145439068e-05, | |
| "loss": 0.4314, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.3612167300380228, | |
| "grad_norm": 0.9496159095843592, | |
| "learning_rate": 1.9613430203032486e-05, | |
| "loss": 0.4438, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.36311787072243346, | |
| "grad_norm": 0.7795836391021491, | |
| "learning_rate": 1.9604232929806493e-05, | |
| "loss": 0.4613, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.3650190114068441, | |
| "grad_norm": 0.9615077067174417, | |
| "learning_rate": 1.9594929736144978e-05, | |
| "loss": 0.4596, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.3669201520912547, | |
| "grad_norm": 0.7729567190375889, | |
| "learning_rate": 1.9585520724648354e-05, | |
| "loss": 0.4473, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.3688212927756654, | |
| "grad_norm": 0.7398723081217433, | |
| "learning_rate": 1.957600599908406e-05, | |
| "loss": 0.4515, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.37072243346007605, | |
| "grad_norm": 0.9535746180449084, | |
| "learning_rate": 1.95663856643854e-05, | |
| "loss": 0.4519, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.3726235741444867, | |
| "grad_norm": 0.7738276583322528, | |
| "learning_rate": 1.955665982665038e-05, | |
| "loss": 0.4511, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.3745247148288973, | |
| "grad_norm": 0.7359192010735204, | |
| "learning_rate": 1.9546828593140565e-05, | |
| "loss": 0.4428, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.376425855513308, | |
| "grad_norm": 0.9056444677383114, | |
| "learning_rate": 1.9536892072279863e-05, | |
| "loss": 0.4551, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.37832699619771865, | |
| "grad_norm": 0.7984813908381183, | |
| "learning_rate": 1.9526850373653356e-05, | |
| "loss": 0.4761, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.38022813688212925, | |
| "grad_norm": 0.718990722490752, | |
| "learning_rate": 1.9516703608006074e-05, | |
| "loss": 0.4101, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.3821292775665399, | |
| "grad_norm": 0.8290974389891955, | |
| "learning_rate": 1.9506451887241787e-05, | |
| "loss": 0.4471, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.3840304182509506, | |
| "grad_norm": 0.7975717712410405, | |
| "learning_rate": 1.949609532442176e-05, | |
| "loss": 0.4663, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.38593155893536124, | |
| "grad_norm": 0.828599830844592, | |
| "learning_rate": 1.9485634033763507e-05, | |
| "loss": 0.4366, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.38783269961977185, | |
| "grad_norm": 0.840634994618327, | |
| "learning_rate": 1.9475068130639543e-05, | |
| "loss": 0.4324, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.3897338403041825, | |
| "grad_norm": 0.8009886603430634, | |
| "learning_rate": 1.9464397731576093e-05, | |
| "loss": 0.4468, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.3916349809885932, | |
| "grad_norm": 0.9080602068434928, | |
| "learning_rate": 1.945362295425183e-05, | |
| "loss": 0.4451, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.3935361216730038, | |
| "grad_norm": 0.7440527020444839, | |
| "learning_rate": 1.944274391749655e-05, | |
| "loss": 0.4362, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.39543726235741444, | |
| "grad_norm": 0.8324183032174326, | |
| "learning_rate": 1.9431760741289886e-05, | |
| "loss": 0.4449, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.3973384030418251, | |
| "grad_norm": 0.7845606727276162, | |
| "learning_rate": 1.942067354675997e-05, | |
| "loss": 0.4704, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.39923954372623577, | |
| "grad_norm": 0.7466045030247327, | |
| "learning_rate": 1.9409482456182105e-05, | |
| "loss": 0.4521, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.4011406844106464, | |
| "grad_norm": 0.7945062011002727, | |
| "learning_rate": 1.939818759297741e-05, | |
| "loss": 0.4695, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.40304182509505704, | |
| "grad_norm": 0.7780959310819049, | |
| "learning_rate": 1.9386789081711465e-05, | |
| "loss": 0.4565, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.4049429657794677, | |
| "grad_norm": 0.7521928077113523, | |
| "learning_rate": 1.9375287048092927e-05, | |
| "loss": 0.4466, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.4068441064638783, | |
| "grad_norm": 0.7742588425754313, | |
| "learning_rate": 1.9363681618972166e-05, | |
| "loss": 0.4472, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.40874524714828897, | |
| "grad_norm": 0.8399358737342701, | |
| "learning_rate": 1.9351972922339835e-05, | |
| "loss": 0.4738, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.41064638783269963, | |
| "grad_norm": 0.772602248327352, | |
| "learning_rate": 1.9340161087325483e-05, | |
| "loss": 0.4577, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.41254752851711024, | |
| "grad_norm": 0.8051426753790454, | |
| "learning_rate": 1.9328246244196117e-05, | |
| "loss": 0.4377, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.4144486692015209, | |
| "grad_norm": 0.825540114072043, | |
| "learning_rate": 1.931622852435478e-05, | |
| "loss": 0.45, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.41634980988593157, | |
| "grad_norm": 0.8689156455841902, | |
| "learning_rate": 1.930410806033908e-05, | |
| "loss": 0.4407, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.41825095057034223, | |
| "grad_norm": 0.7605796586705885, | |
| "learning_rate": 1.929188498581975e-05, | |
| "loss": 0.4347, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.42015209125475284, | |
| "grad_norm": 0.7746227751267498, | |
| "learning_rate": 1.9279559435599164e-05, | |
| "loss": 0.4608, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.4220532319391635, | |
| "grad_norm": 0.8093343597747852, | |
| "learning_rate": 1.926713154560984e-05, | |
| "loss": 0.4645, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.42395437262357416, | |
| "grad_norm": 0.7943642509313467, | |
| "learning_rate": 1.9254601452912972e-05, | |
| "loss": 0.474, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.42585551330798477, | |
| "grad_norm": 0.7489339716856304, | |
| "learning_rate": 1.924196929569688e-05, | |
| "loss": 0.4364, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.42775665399239543, | |
| "grad_norm": 0.7289211877219848, | |
| "learning_rate": 1.922923521327551e-05, | |
| "loss": 0.4966, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.4296577946768061, | |
| "grad_norm": 0.8083873629142393, | |
| "learning_rate": 1.9216399346086893e-05, | |
| "loss": 0.4665, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.43155893536121676, | |
| "grad_norm": 0.7785475877926267, | |
| "learning_rate": 1.9203461835691596e-05, | |
| "loss": 0.4555, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.43346007604562736, | |
| "grad_norm": 0.7652279927614211, | |
| "learning_rate": 1.9190422824771158e-05, | |
| "loss": 0.4423, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.435361216730038, | |
| "grad_norm": 0.7533410112419021, | |
| "learning_rate": 1.9177282457126515e-05, | |
| "loss": 0.4428, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.4372623574144487, | |
| "grad_norm": 0.9035323704611021, | |
| "learning_rate": 1.9164040877676425e-05, | |
| "loss": 0.4486, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.4391634980988593, | |
| "grad_norm": 0.7514514116540132, | |
| "learning_rate": 1.9150698232455853e-05, | |
| "loss": 0.4065, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.44106463878326996, | |
| "grad_norm": 0.7834862504557109, | |
| "learning_rate": 1.913725466861438e-05, | |
| "loss": 0.4609, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.4429657794676806, | |
| "grad_norm": 0.9488369950834241, | |
| "learning_rate": 1.9123710334414552e-05, | |
| "loss": 0.4617, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.4448669201520912, | |
| "grad_norm": 0.8229549620865543, | |
| "learning_rate": 1.911006537923029e-05, | |
| "loss": 0.4388, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.4467680608365019, | |
| "grad_norm": 0.8678227434871482, | |
| "learning_rate": 1.9096319953545186e-05, | |
| "loss": 0.4355, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.44866920152091255, | |
| "grad_norm": 0.8299700711642316, | |
| "learning_rate": 1.908247420895089e-05, | |
| "loss": 0.4525, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.4505703422053232, | |
| "grad_norm": 0.8783167721865093, | |
| "learning_rate": 1.9068528298145418e-05, | |
| "loss": 0.4535, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.4524714828897338, | |
| "grad_norm": 0.8028815413858434, | |
| "learning_rate": 1.905448237493147e-05, | |
| "loss": 0.4554, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.4543726235741445, | |
| "grad_norm": 0.7485166341772101, | |
| "learning_rate": 1.9040336594214727e-05, | |
| "loss": 0.4349, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.45627376425855515, | |
| "grad_norm": 0.7975213172290808, | |
| "learning_rate": 1.9026091112002163e-05, | |
| "loss": 0.4216, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.45817490494296575, | |
| "grad_norm": 0.6793025144403352, | |
| "learning_rate": 1.90117460854003e-05, | |
| "loss": 0.4323, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.4600760456273764, | |
| "grad_norm": 0.806728156561214, | |
| "learning_rate": 1.8997301672613496e-05, | |
| "loss": 0.4777, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.4619771863117871, | |
| "grad_norm": 0.7513800330978121, | |
| "learning_rate": 1.8982758032942184e-05, | |
| "loss": 0.4686, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.46387832699619774, | |
| "grad_norm": 0.7472709784074971, | |
| "learning_rate": 1.896811532678113e-05, | |
| "loss": 0.4359, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.46577946768060835, | |
| "grad_norm": 0.7178309504766717, | |
| "learning_rate": 1.8953373715617646e-05, | |
| "loss": 0.4631, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.467680608365019, | |
| "grad_norm": 0.7222581622532535, | |
| "learning_rate": 1.893853336202983e-05, | |
| "loss": 0.4338, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.4695817490494297, | |
| "grad_norm": 0.7953544220547667, | |
| "learning_rate": 1.892359442968475e-05, | |
| "loss": 0.4602, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 0.4714828897338403, | |
| "grad_norm": 0.7670752990176973, | |
| "learning_rate": 1.8908557083336668e-05, | |
| "loss": 0.4373, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.47338403041825095, | |
| "grad_norm": 0.7963978968168441, | |
| "learning_rate": 1.889342148882519e-05, | |
| "loss": 0.4475, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 0.4752851711026616, | |
| "grad_norm": 0.7485934032500751, | |
| "learning_rate": 1.8878187813073465e-05, | |
| "loss": 0.4689, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.47718631178707227, | |
| "grad_norm": 0.752477034304505, | |
| "learning_rate": 1.886285622408633e-05, | |
| "loss": 0.4227, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 0.4790874524714829, | |
| "grad_norm": 0.7656883403463771, | |
| "learning_rate": 1.8847426890948447e-05, | |
| "loss": 0.4566, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.48098859315589354, | |
| "grad_norm": 0.8287248358891999, | |
| "learning_rate": 1.8831899983822475e-05, | |
| "loss": 0.4567, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 0.4828897338403042, | |
| "grad_norm": 0.709592560308173, | |
| "learning_rate": 1.8816275673947148e-05, | |
| "loss": 0.4276, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.4847908745247148, | |
| "grad_norm": 0.7647258766146171, | |
| "learning_rate": 1.8800554133635417e-05, | |
| "loss": 0.429, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.4866920152091255, | |
| "grad_norm": 0.7736986610690838, | |
| "learning_rate": 1.8784735536272543e-05, | |
| "loss": 0.4411, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.48859315589353614, | |
| "grad_norm": 0.7603112348009117, | |
| "learning_rate": 1.8768820056314173e-05, | |
| "loss": 0.4264, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 0.49049429657794674, | |
| "grad_norm": 0.7654292049195688, | |
| "learning_rate": 1.875280786928444e-05, | |
| "loss": 0.4478, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.4923954372623574, | |
| "grad_norm": 0.8158171408640096, | |
| "learning_rate": 1.873669915177399e-05, | |
| "loss": 0.4537, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 0.49429657794676807, | |
| "grad_norm": 0.7872680482383151, | |
| "learning_rate": 1.872049408143808e-05, | |
| "loss": 0.4327, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.49619771863117873, | |
| "grad_norm": 0.7147941850165362, | |
| "learning_rate": 1.8704192836994578e-05, | |
| "loss": 0.4242, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 0.49809885931558934, | |
| "grad_norm": 0.8170134227015007, | |
| "learning_rate": 1.8687795598222024e-05, | |
| "loss": 0.446, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "grad_norm": 0.7812141373841187, | |
| "learning_rate": 1.8671302545957628e-05, | |
| "loss": 0.4373, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 0.5019011406844106, | |
| "grad_norm": 0.7482512485302047, | |
| "learning_rate": 1.8654713862095272e-05, | |
| "loss": 0.4412, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.5038022813688213, | |
| "grad_norm": 0.7083970649149763, | |
| "learning_rate": 1.8638029729583524e-05, | |
| "loss": 0.4572, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.5057034220532319, | |
| "grad_norm": 0.7999224754557839, | |
| "learning_rate": 1.8621250332423603e-05, | |
| "loss": 0.4386, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.5076045627376425, | |
| "grad_norm": 0.8196419564051675, | |
| "learning_rate": 1.860437585566736e-05, | |
| "loss": 0.4315, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 0.5095057034220533, | |
| "grad_norm": 0.7554009836821762, | |
| "learning_rate": 1.8587406485415226e-05, | |
| "loss": 0.4572, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.5114068441064639, | |
| "grad_norm": 0.8235034166864555, | |
| "learning_rate": 1.8570342408814173e-05, | |
| "loss": 0.4457, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 0.5133079847908745, | |
| "grad_norm": 0.7609746796349174, | |
| "learning_rate": 1.855318381405564e-05, | |
| "loss": 0.4621, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.5152091254752852, | |
| "grad_norm": 0.8084724698206379, | |
| "learning_rate": 1.8535930890373467e-05, | |
| "loss": 0.4431, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 0.5171102661596958, | |
| "grad_norm": 0.765025495545704, | |
| "learning_rate": 1.8518583828041787e-05, | |
| "loss": 0.4537, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.5190114068441065, | |
| "grad_norm": 0.8157046318114198, | |
| "learning_rate": 1.8501142818372964e-05, | |
| "loss": 0.4404, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 0.5209125475285171, | |
| "grad_norm": 0.7341734917298394, | |
| "learning_rate": 1.848360805371544e-05, | |
| "loss": 0.4586, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.5228136882129277, | |
| "grad_norm": 0.7802589801858523, | |
| "learning_rate": 1.8465979727451653e-05, | |
| "loss": 0.4365, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.5247148288973384, | |
| "grad_norm": 0.7427878659815825, | |
| "learning_rate": 1.8448258033995877e-05, | |
| "loss": 0.4385, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.526615969581749, | |
| "grad_norm": 0.7621942231413598, | |
| "learning_rate": 1.8430443168792087e-05, | |
| "loss": 0.4173, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 0.5285171102661597, | |
| "grad_norm": 0.7216327068077534, | |
| "learning_rate": 1.8412535328311813e-05, | |
| "loss": 0.4467, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 0.5304182509505704, | |
| "grad_norm": 0.7353414839073464, | |
| "learning_rate": 1.8394534710051956e-05, | |
| "loss": 0.4524, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 0.532319391634981, | |
| "grad_norm": 0.7251640075230803, | |
| "learning_rate": 1.8376441512532617e-05, | |
| "loss": 0.4462, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.5342205323193916, | |
| "grad_norm": 0.6893097817389318, | |
| "learning_rate": 1.835825593529492e-05, | |
| "loss": 0.4284, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 0.5361216730038023, | |
| "grad_norm": 0.6838924104200217, | |
| "learning_rate": 1.833997817889878e-05, | |
| "loss": 0.4298, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 0.5380228136882129, | |
| "grad_norm": 0.7091549458682664, | |
| "learning_rate": 1.8321608444920738e-05, | |
| "loss": 0.4193, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 0.5399239543726235, | |
| "grad_norm": 0.7179860942288903, | |
| "learning_rate": 1.830314693595169e-05, | |
| "loss": 0.4354, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.5418250950570342, | |
| "grad_norm": 0.7509839555279627, | |
| "learning_rate": 1.828459385559468e-05, | |
| "loss": 0.4245, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.5437262357414449, | |
| "grad_norm": 0.7086227492353241, | |
| "learning_rate": 1.8265949408462657e-05, | |
| "loss": 0.4325, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 0.5456273764258555, | |
| "grad_norm": 0.806516436800297, | |
| "learning_rate": 1.8247213800176192e-05, | |
| "loss": 0.4523, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 0.5475285171102662, | |
| "grad_norm": 0.8308682529867073, | |
| "learning_rate": 1.8228387237361245e-05, | |
| "loss": 0.4272, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.5494296577946768, | |
| "grad_norm": 0.7828482638716789, | |
| "learning_rate": 1.8209469927646863e-05, | |
| "loss": 0.4297, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 0.5513307984790875, | |
| "grad_norm": 0.8127665202479064, | |
| "learning_rate": 1.8190462079662897e-05, | |
| "loss": 0.449, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.5532319391634981, | |
| "grad_norm": 0.7713123918315768, | |
| "learning_rate": 1.81713639030377e-05, | |
| "loss": 0.44, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.5551330798479087, | |
| "grad_norm": 0.857355338897827, | |
| "learning_rate": 1.8152175608395814e-05, | |
| "loss": 0.4225, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 0.5570342205323194, | |
| "grad_norm": 0.842314632308363, | |
| "learning_rate": 1.8132897407355657e-05, | |
| "loss": 0.4379, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 0.55893536121673, | |
| "grad_norm": 0.7785154109213545, | |
| "learning_rate": 1.811352951252717e-05, | |
| "loss": 0.4016, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 0.5608365019011406, | |
| "grad_norm": 0.7137872696323513, | |
| "learning_rate": 1.809407213750949e-05, | |
| "loss": 0.4274, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.5627376425855514, | |
| "grad_norm": 0.7913459160736689, | |
| "learning_rate": 1.807452549688859e-05, | |
| "loss": 0.4595, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 0.564638783269962, | |
| "grad_norm": 0.73962562599915, | |
| "learning_rate": 1.8054889806234906e-05, | |
| "loss": 0.47, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 0.5665399239543726, | |
| "grad_norm": 0.8688373235468253, | |
| "learning_rate": 1.8035165282100963e-05, | |
| "loss": 0.4473, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 0.5684410646387833, | |
| "grad_norm": 0.7656361283591083, | |
| "learning_rate": 1.8015352142018984e-05, | |
| "loss": 0.4366, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 0.5703422053231939, | |
| "grad_norm": 0.6852748546626566, | |
| "learning_rate": 1.799545060449851e-05, | |
| "loss": 0.4154, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.5722433460076045, | |
| "grad_norm": 0.9409842339787114, | |
| "learning_rate": 1.797546088902396e-05, | |
| "loss": 0.4386, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 0.5741444866920152, | |
| "grad_norm": 0.7233829290836847, | |
| "learning_rate": 1.7955383216052224e-05, | |
| "loss": 0.4521, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 0.5760456273764258, | |
| "grad_norm": 0.7386515969178572, | |
| "learning_rate": 1.7935217807010238e-05, | |
| "loss": 0.4306, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 0.5779467680608364, | |
| "grad_norm": 0.8490256490662417, | |
| "learning_rate": 1.7914964884292543e-05, | |
| "loss": 0.4275, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 0.5798479087452472, | |
| "grad_norm": 0.7204866607387883, | |
| "learning_rate": 1.7894624671258813e-05, | |
| "loss": 0.422, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.5817490494296578, | |
| "grad_norm": 0.8723449339230218, | |
| "learning_rate": 1.7874197392231414e-05, | |
| "loss": 0.4341, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 0.5836501901140685, | |
| "grad_norm": 0.7256868786917025, | |
| "learning_rate": 1.7853683272492913e-05, | |
| "loss": 0.4254, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 0.5855513307984791, | |
| "grad_norm": 0.7155265513213723, | |
| "learning_rate": 1.7833082538283615e-05, | |
| "loss": 0.4238, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 0.5874524714828897, | |
| "grad_norm": 0.6889360386336122, | |
| "learning_rate": 1.7812395416799034e-05, | |
| "loss": 0.4305, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 0.5893536121673004, | |
| "grad_norm": 0.7415893528252747, | |
| "learning_rate": 1.7791622136187422e-05, | |
| "loss": 0.4551, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.591254752851711, | |
| "grad_norm": 0.7466600182812532, | |
| "learning_rate": 1.7770762925547235e-05, | |
| "loss": 0.456, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 0.5931558935361216, | |
| "grad_norm": 0.663674802960924, | |
| "learning_rate": 1.7749818014924612e-05, | |
| "loss": 0.4251, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 0.5950570342205324, | |
| "grad_norm": 0.6517189143740555, | |
| "learning_rate": 1.7728787635310828e-05, | |
| "loss": 0.4197, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 0.596958174904943, | |
| "grad_norm": 0.6700424379527684, | |
| "learning_rate": 1.770767201863976e-05, | |
| "loss": 0.4279, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.5988593155893536, | |
| "grad_norm": 0.6580610103209431, | |
| "learning_rate": 1.7686471397785322e-05, | |
| "loss": 0.4207, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.6007604562737643, | |
| "grad_norm": 0.7063861733419377, | |
| "learning_rate": 1.76651860065589e-05, | |
| "loss": 0.4135, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 0.6026615969581749, | |
| "grad_norm": 0.7361275577613828, | |
| "learning_rate": 1.764381607970677e-05, | |
| "loss": 0.4506, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 0.6045627376425855, | |
| "grad_norm": 0.694876466358444, | |
| "learning_rate": 1.7622361852907506e-05, | |
| "loss": 0.4321, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 0.6064638783269962, | |
| "grad_norm": 0.7402451133352099, | |
| "learning_rate": 1.760082356276939e-05, | |
| "loss": 0.437, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 0.6083650190114068, | |
| "grad_norm": 0.683231783487908, | |
| "learning_rate": 1.75792014468278e-05, | |
| "loss": 0.4268, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.6102661596958175, | |
| "grad_norm": 0.7027432535840766, | |
| "learning_rate": 1.7557495743542586e-05, | |
| "loss": 0.4394, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 0.6121673003802282, | |
| "grad_norm": 0.7274061100398459, | |
| "learning_rate": 1.7535706692295436e-05, | |
| "loss": 0.4476, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 0.6140684410646388, | |
| "grad_norm": 0.7392791763710189, | |
| "learning_rate": 1.7513834533387256e-05, | |
| "loss": 0.4239, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 0.6159695817490495, | |
| "grad_norm": 0.7028877326905897, | |
| "learning_rate": 1.7491879508035488e-05, | |
| "loss": 0.4299, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 0.6178707224334601, | |
| "grad_norm": 0.6945237566271814, | |
| "learning_rate": 1.746984185837149e-05, | |
| "loss": 0.423, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.6197718631178707, | |
| "grad_norm": 0.6793444484269197, | |
| "learning_rate": 1.744772182743782e-05, | |
| "loss": 0.4336, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 0.6216730038022814, | |
| "grad_norm": 0.6743325676517699, | |
| "learning_rate": 1.7425519659185596e-05, | |
| "loss": 0.4371, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 0.623574144486692, | |
| "grad_norm": 0.8025660005405725, | |
| "learning_rate": 1.740323559847179e-05, | |
| "loss": 0.4288, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 0.6254752851711026, | |
| "grad_norm": 0.7063101237110045, | |
| "learning_rate": 1.738086989105651e-05, | |
| "loss": 0.4147, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 0.6273764258555133, | |
| "grad_norm": 0.7030359168637568, | |
| "learning_rate": 1.735842278360032e-05, | |
| "loss": 0.4533, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.629277566539924, | |
| "grad_norm": 0.7828178041360292, | |
| "learning_rate": 1.73358945236615e-05, | |
| "loss": 0.4253, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 0.6311787072243346, | |
| "grad_norm": 0.7105048580452298, | |
| "learning_rate": 1.7313285359693322e-05, | |
| "loss": 0.4374, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 0.6330798479087453, | |
| "grad_norm": 0.6458786718258576, | |
| "learning_rate": 1.7290595541041312e-05, | |
| "loss": 0.4143, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 0.6349809885931559, | |
| "grad_norm": 0.7086682254770856, | |
| "learning_rate": 1.7267825317940494e-05, | |
| "loss": 0.4479, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 0.6368821292775665, | |
| "grad_norm": 0.7085159753230028, | |
| "learning_rate": 1.724497494151264e-05, | |
| "loss": 0.4171, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.6387832699619772, | |
| "grad_norm": 0.7247315192105774, | |
| "learning_rate": 1.7222044663763484e-05, | |
| "loss": 0.4289, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 0.6406844106463878, | |
| "grad_norm": 0.7477867821989687, | |
| "learning_rate": 1.7199034737579962e-05, | |
| "loss": 0.4247, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 0.6425855513307985, | |
| "grad_norm": 0.7062498080690548, | |
| "learning_rate": 1.7175945416727405e-05, | |
| "loss": 0.4368, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 0.6444866920152091, | |
| "grad_norm": 0.7389439527765631, | |
| "learning_rate": 1.7152776955846768e-05, | |
| "loss": 0.4219, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 0.6463878326996197, | |
| "grad_norm": 0.6930098857476008, | |
| "learning_rate": 1.7129529610451775e-05, | |
| "loss": 0.397, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.6482889733840305, | |
| "grad_norm": 0.6902076953593139, | |
| "learning_rate": 1.7106203636926154e-05, | |
| "loss": 0.4306, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 0.6501901140684411, | |
| "grad_norm": 0.7891735288780114, | |
| "learning_rate": 1.7082799292520767e-05, | |
| "loss": 0.424, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 0.6520912547528517, | |
| "grad_norm": 0.6884572355199468, | |
| "learning_rate": 1.7059316835350806e-05, | |
| "loss": 0.4416, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 0.6539923954372624, | |
| "grad_norm": 0.7235608035967601, | |
| "learning_rate": 1.7035756524392924e-05, | |
| "loss": 0.4333, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 0.655893536121673, | |
| "grad_norm": 0.7326410359193561, | |
| "learning_rate": 1.7012118619482376e-05, | |
| "loss": 0.4383, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.6577946768060836, | |
| "grad_norm": 0.6976061620228513, | |
| "learning_rate": 1.6988403381310177e-05, | |
| "loss": 0.455, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 0.6596958174904943, | |
| "grad_norm": 0.7088495814837974, | |
| "learning_rate": 1.696461107142021e-05, | |
| "loss": 0.4363, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 0.6615969581749049, | |
| "grad_norm": 0.6967864059290054, | |
| "learning_rate": 1.6940741952206342e-05, | |
| "loss": 0.4204, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 0.6634980988593155, | |
| "grad_norm": 0.729074169761856, | |
| "learning_rate": 1.691679628690953e-05, | |
| "loss": 0.4324, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 0.6653992395437263, | |
| "grad_norm": 0.7196004832510822, | |
| "learning_rate": 1.6892774339614927e-05, | |
| "loss": 0.417, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.6673003802281369, | |
| "grad_norm": 0.7163535471808641, | |
| "learning_rate": 1.686867637524896e-05, | |
| "loss": 0.4396, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 0.6692015209125475, | |
| "grad_norm": 0.6898671355445452, | |
| "learning_rate": 1.6844502659576414e-05, | |
| "loss": 0.4286, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 0.6711026615969582, | |
| "grad_norm": 0.7912939800800087, | |
| "learning_rate": 1.6820253459197493e-05, | |
| "loss": 0.4522, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 0.6730038022813688, | |
| "grad_norm": 0.781608366020826, | |
| "learning_rate": 1.679592904154489e-05, | |
| "loss": 0.4483, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 0.6749049429657795, | |
| "grad_norm": 0.7181614890135484, | |
| "learning_rate": 1.677152967488084e-05, | |
| "loss": 0.4325, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.6768060836501901, | |
| "grad_norm": 0.7171424421706101, | |
| "learning_rate": 1.6747055628294134e-05, | |
| "loss": 0.4235, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 0.6787072243346007, | |
| "grad_norm": 0.8184585866740236, | |
| "learning_rate": 1.6722507171697184e-05, | |
| "loss": 0.4405, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 0.6806083650190115, | |
| "grad_norm": 0.6770453983230084, | |
| "learning_rate": 1.669788457582304e-05, | |
| "loss": 0.4506, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.6825095057034221, | |
| "grad_norm": 0.7451899787415116, | |
| "learning_rate": 1.6673188112222394e-05, | |
| "loss": 0.4217, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 0.6844106463878327, | |
| "grad_norm": 0.7372141212655945, | |
| "learning_rate": 1.6648418053260585e-05, | |
| "loss": 0.4322, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.6863117870722434, | |
| "grad_norm": 0.6483777881857093, | |
| "learning_rate": 1.6623574672114596e-05, | |
| "loss": 0.4358, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 0.688212927756654, | |
| "grad_norm": 0.6720140582954877, | |
| "learning_rate": 1.6598658242770054e-05, | |
| "loss": 0.4365, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 0.6901140684410646, | |
| "grad_norm": 0.6620537262139077, | |
| "learning_rate": 1.6573669040018202e-05, | |
| "loss": 0.407, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 0.6920152091254753, | |
| "grad_norm": 0.6520035630226115, | |
| "learning_rate": 1.6548607339452853e-05, | |
| "loss": 0.4065, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 0.6939163498098859, | |
| "grad_norm": 0.6424752525593794, | |
| "learning_rate": 1.652347341746737e-05, | |
| "loss": 0.4365, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.6958174904942965, | |
| "grad_norm": 0.6761986424108621, | |
| "learning_rate": 1.6498267551251618e-05, | |
| "loss": 0.4236, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 0.6977186311787072, | |
| "grad_norm": 0.6313797598784099, | |
| "learning_rate": 1.6472990018788884e-05, | |
| "loss": 0.4263, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 0.6996197718631179, | |
| "grad_norm": 0.6691804141094988, | |
| "learning_rate": 1.644764109885284e-05, | |
| "loss": 0.4374, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 0.7015209125475285, | |
| "grad_norm": 0.646926280134475, | |
| "learning_rate": 1.642222107100446e-05, | |
| "loss": 0.4345, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 0.7034220532319392, | |
| "grad_norm": 0.736308712440305, | |
| "learning_rate": 1.6396730215588913e-05, | |
| "loss": 0.4255, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.7053231939163498, | |
| "grad_norm": 0.6795105637559926, | |
| "learning_rate": 1.6371168813732514e-05, | |
| "loss": 0.413, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 0.7072243346007605, | |
| "grad_norm": 0.7177204878528012, | |
| "learning_rate": 1.6345537147339578e-05, | |
| "loss": 0.4228, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 0.7091254752851711, | |
| "grad_norm": 0.699580210415486, | |
| "learning_rate": 1.6319835499089358e-05, | |
| "loss": 0.4106, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 0.7110266159695817, | |
| "grad_norm": 0.6485833165506162, | |
| "learning_rate": 1.6294064152432878e-05, | |
| "loss": 0.4153, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 0.7129277566539924, | |
| "grad_norm": 0.6748926577699425, | |
| "learning_rate": 1.626822339158985e-05, | |
| "loss": 0.4137, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.714828897338403, | |
| "grad_norm": 0.6422789519155716, | |
| "learning_rate": 1.6242313501545522e-05, | |
| "loss": 0.4254, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 0.7167300380228137, | |
| "grad_norm": 0.7045660563799033, | |
| "learning_rate": 1.621633476804752e-05, | |
| "loss": 0.4458, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 0.7186311787072244, | |
| "grad_norm": 0.7064805151191417, | |
| "learning_rate": 1.6190287477602716e-05, | |
| "loss": 0.4273, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 0.720532319391635, | |
| "grad_norm": 0.7485928725544002, | |
| "learning_rate": 1.6164171917474078e-05, | |
| "loss": 0.4221, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 0.7224334600760456, | |
| "grad_norm": 0.7870926378422516, | |
| "learning_rate": 1.6137988375677466e-05, | |
| "loss": 0.4703, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.7243346007604563, | |
| "grad_norm": 0.6695750340051236, | |
| "learning_rate": 1.6111737140978495e-05, | |
| "loss": 0.4252, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 0.7262357414448669, | |
| "grad_norm": 0.6694008109102478, | |
| "learning_rate": 1.6085418502889315e-05, | |
| "loss": 0.4204, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 0.7281368821292775, | |
| "grad_norm": 0.7078291674703513, | |
| "learning_rate": 1.6059032751665454e-05, | |
| "loss": 0.4239, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 0.7300380228136882, | |
| "grad_norm": 0.7233812060220794, | |
| "learning_rate": 1.6032580178302585e-05, | |
| "loss": 0.4434, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 0.7319391634980988, | |
| "grad_norm": 0.6852596612012354, | |
| "learning_rate": 1.600606107453333e-05, | |
| "loss": 0.4326, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.7338403041825095, | |
| "grad_norm": 0.7152705349830047, | |
| "learning_rate": 1.597947573282405e-05, | |
| "loss": 0.4261, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 0.7357414448669202, | |
| "grad_norm": 0.7314014584662948, | |
| "learning_rate": 1.5952824446371608e-05, | |
| "loss": 0.463, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 0.7376425855513308, | |
| "grad_norm": 0.7219181501291495, | |
| "learning_rate": 1.592610750910014e-05, | |
| "loss": 0.4348, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 0.7395437262357415, | |
| "grad_norm": 0.6530017696086983, | |
| "learning_rate": 1.589932521565781e-05, | |
| "loss": 0.406, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 0.7414448669201521, | |
| "grad_norm": 0.6898391463555436, | |
| "learning_rate": 1.587247786141358e-05, | |
| "loss": 0.4259, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.7433460076045627, | |
| "grad_norm": 0.6508079630701724, | |
| "learning_rate": 1.5845565742453906e-05, | |
| "loss": 0.4343, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 0.7452471482889734, | |
| "grad_norm": 0.701984224294415, | |
| "learning_rate": 1.581858915557953e-05, | |
| "loss": 0.417, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 0.747148288973384, | |
| "grad_norm": 0.649465549135197, | |
| "learning_rate": 1.5791548398302167e-05, | |
| "loss": 0.4232, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 0.7490494296577946, | |
| "grad_norm": 0.7111287126122185, | |
| "learning_rate": 1.5764443768841234e-05, | |
| "loss": 0.3993, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 0.7509505703422054, | |
| "grad_norm": 0.7483369961975148, | |
| "learning_rate": 1.5737275566120577e-05, | |
| "loss": 0.4378, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.752851711026616, | |
| "grad_norm": 0.7103429507864539, | |
| "learning_rate": 1.5710044089765144e-05, | |
| "loss": 0.4339, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 0.7547528517110266, | |
| "grad_norm": 0.7426731098474025, | |
| "learning_rate": 1.5682749640097708e-05, | |
| "loss": 0.4241, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 0.7566539923954373, | |
| "grad_norm": 0.6195135205880833, | |
| "learning_rate": 1.565539251813554e-05, | |
| "loss": 0.3932, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 0.7585551330798479, | |
| "grad_norm": 0.77497987544636, | |
| "learning_rate": 1.5627973025587093e-05, | |
| "loss": 0.4303, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 0.7604562737642585, | |
| "grad_norm": 0.7728057300596839, | |
| "learning_rate": 1.560049146484868e-05, | |
| "loss": 0.4293, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.7623574144486692, | |
| "grad_norm": 0.7639787800632488, | |
| "learning_rate": 1.5572948139001128e-05, | |
| "loss": 0.43, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 0.7642585551330798, | |
| "grad_norm": 0.7069452847068256, | |
| "learning_rate": 1.5545343351806443e-05, | |
| "loss": 0.4214, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 0.7661596958174905, | |
| "grad_norm": 0.71579783171243, | |
| "learning_rate": 1.551767740770446e-05, | |
| "loss": 0.4081, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 0.7680608365019012, | |
| "grad_norm": 0.7300557884621219, | |
| "learning_rate": 1.5489950611809484e-05, | |
| "loss": 0.4204, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 0.7699619771863118, | |
| "grad_norm": 0.7326667676314269, | |
| "learning_rate": 1.5462163269906928e-05, | |
| "loss": 0.422, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.7718631178707225, | |
| "grad_norm": 0.6729518795102971, | |
| "learning_rate": 1.5434315688449924e-05, | |
| "loss": 0.4402, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 0.7737642585551331, | |
| "grad_norm": 0.8180062516601114, | |
| "learning_rate": 1.5406408174555978e-05, | |
| "loss": 0.4118, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 0.7756653992395437, | |
| "grad_norm": 0.7711034949785363, | |
| "learning_rate": 1.5378441036003543e-05, | |
| "loss": 0.438, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 0.7775665399239544, | |
| "grad_norm": 0.6754089476056547, | |
| "learning_rate": 1.535041458122865e-05, | |
| "loss": 0.4111, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 0.779467680608365, | |
| "grad_norm": 0.8030264732154252, | |
| "learning_rate": 1.5322329119321508e-05, | |
| "loss": 0.4259, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.7813688212927756, | |
| "grad_norm": 0.847871016885185, | |
| "learning_rate": 1.529418496002308e-05, | |
| "loss": 0.4244, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 0.7832699619771863, | |
| "grad_norm": 0.7184378133667302, | |
| "learning_rate": 1.5265982413721662e-05, | |
| "loss": 0.4031, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 0.785171102661597, | |
| "grad_norm": 0.9023298345299018, | |
| "learning_rate": 1.5237721791449497e-05, | |
| "loss": 0.4233, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 0.7870722433460076, | |
| "grad_norm": 0.6709862565190281, | |
| "learning_rate": 1.5209403404879305e-05, | |
| "loss": 0.425, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 0.7889733840304183, | |
| "grad_norm": 0.6642161203432194, | |
| "learning_rate": 1.5181027566320858e-05, | |
| "loss": 0.4124, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.7908745247148289, | |
| "grad_norm": 0.8842384997073621, | |
| "learning_rate": 1.5152594588717544e-05, | |
| "loss": 0.435, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 0.7927756653992395, | |
| "grad_norm": 0.6768800462682988, | |
| "learning_rate": 1.5124104785642909e-05, | |
| "loss": 0.4333, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 0.7946768060836502, | |
| "grad_norm": 0.9218188555328675, | |
| "learning_rate": 1.5095558471297196e-05, | |
| "loss": 0.4263, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 0.7965779467680608, | |
| "grad_norm": 0.6980077336892946, | |
| "learning_rate": 1.5066955960503893e-05, | |
| "loss": 0.4424, | |
| "step": 419 | |
| }, | |
| { | |
| "epoch": 0.7984790874524715, | |
| "grad_norm": 0.7327685022091915, | |
| "learning_rate": 1.5038297568706244e-05, | |
| "loss": 0.4265, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.8003802281368821, | |
| "grad_norm": 0.8009085383102565, | |
| "learning_rate": 1.5009583611963772e-05, | |
| "loss": 0.433, | |
| "step": 421 | |
| }, | |
| { | |
| "epoch": 0.8022813688212928, | |
| "grad_norm": 0.6391551593198743, | |
| "learning_rate": 1.4980814406948806e-05, | |
| "loss": 0.433, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 0.8041825095057035, | |
| "grad_norm": 0.7094621590510084, | |
| "learning_rate": 1.4951990270942991e-05, | |
| "loss": 0.418, | |
| "step": 423 | |
| }, | |
| { | |
| "epoch": 0.8060836501901141, | |
| "grad_norm": 0.7396853097710299, | |
| "learning_rate": 1.492311152183376e-05, | |
| "loss": 0.3893, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 0.8079847908745247, | |
| "grad_norm": 0.7038078501104372, | |
| "learning_rate": 1.4894178478110856e-05, | |
| "loss": 0.406, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.8098859315589354, | |
| "grad_norm": 0.6443174626165765, | |
| "learning_rate": 1.4865191458862816e-05, | |
| "loss": 0.4057, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 0.811787072243346, | |
| "grad_norm": 0.8136266581435002, | |
| "learning_rate": 1.4836150783773442e-05, | |
| "loss": 0.4079, | |
| "step": 427 | |
| }, | |
| { | |
| "epoch": 0.8136882129277566, | |
| "grad_norm": 0.7732014105345911, | |
| "learning_rate": 1.4807056773118276e-05, | |
| "loss": 0.4372, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 0.8155893536121673, | |
| "grad_norm": 0.6742068185369734, | |
| "learning_rate": 1.4777909747761085e-05, | |
| "loss": 0.4003, | |
| "step": 429 | |
| }, | |
| { | |
| "epoch": 0.8174904942965779, | |
| "grad_norm": 0.813737383644273, | |
| "learning_rate": 1.4748710029150296e-05, | |
| "loss": 0.4447, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.8193916349809885, | |
| "grad_norm": 0.6947186195723536, | |
| "learning_rate": 1.4719457939315468e-05, | |
| "loss": 0.4262, | |
| "step": 431 | |
| }, | |
| { | |
| "epoch": 0.8212927756653993, | |
| "grad_norm": 0.6839551277893491, | |
| "learning_rate": 1.4690153800863743e-05, | |
| "loss": 0.4068, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 0.8231939163498099, | |
| "grad_norm": 0.6996801267842218, | |
| "learning_rate": 1.4660797936976278e-05, | |
| "loss": 0.4079, | |
| "step": 433 | |
| }, | |
| { | |
| "epoch": 0.8250950570342205, | |
| "grad_norm": 0.6825162579877742, | |
| "learning_rate": 1.4631390671404682e-05, | |
| "loss": 0.415, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 0.8269961977186312, | |
| "grad_norm": 0.6837018519935446, | |
| "learning_rate": 1.460193232846745e-05, | |
| "loss": 0.4363, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.8288973384030418, | |
| "grad_norm": 0.7837297737441469, | |
| "learning_rate": 1.4572423233046386e-05, | |
| "loss": 0.4389, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 0.8307984790874525, | |
| "grad_norm": 0.678754843806813, | |
| "learning_rate": 1.4542863710583022e-05, | |
| "loss": 0.44, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 0.8326996197718631, | |
| "grad_norm": 0.6403838911944043, | |
| "learning_rate": 1.4513254087075015e-05, | |
| "loss": 0.4352, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 0.8346007604562737, | |
| "grad_norm": 0.6726459525079256, | |
| "learning_rate": 1.4483594689072571e-05, | |
| "loss": 0.4151, | |
| "step": 439 | |
| }, | |
| { | |
| "epoch": 0.8365019011406845, | |
| "grad_norm": 0.7335435052384716, | |
| "learning_rate": 1.4453885843674837e-05, | |
| "loss": 0.4077, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.8384030418250951, | |
| "grad_norm": 0.6793133084295226, | |
| "learning_rate": 1.4424127878526278e-05, | |
| "loss": 0.4257, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 0.8403041825095057, | |
| "grad_norm": 0.7935664736067994, | |
| "learning_rate": 1.4394321121813093e-05, | |
| "loss": 0.4173, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 0.8422053231939164, | |
| "grad_norm": 0.7227476595374016, | |
| "learning_rate": 1.436446590225957e-05, | |
| "loss": 0.4208, | |
| "step": 443 | |
| }, | |
| { | |
| "epoch": 0.844106463878327, | |
| "grad_norm": 0.6678629475323585, | |
| "learning_rate": 1.433456254912447e-05, | |
| "loss": 0.4012, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 0.8460076045627376, | |
| "grad_norm": 0.6448963430672735, | |
| "learning_rate": 1.4304611392197399e-05, | |
| "loss": 0.4193, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.8479087452471483, | |
| "grad_norm": 0.6543661102411391, | |
| "learning_rate": 1.427461276179517e-05, | |
| "loss": 0.4263, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 0.8498098859315589, | |
| "grad_norm": 0.6547563998208504, | |
| "learning_rate": 1.4244566988758152e-05, | |
| "loss": 0.4131, | |
| "step": 447 | |
| }, | |
| { | |
| "epoch": 0.8517110266159695, | |
| "grad_norm": 0.6650732446677436, | |
| "learning_rate": 1.4214474404446633e-05, | |
| "loss": 0.4202, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 0.8536121673003803, | |
| "grad_norm": 0.6761353702881292, | |
| "learning_rate": 1.4184335340737158e-05, | |
| "loss": 0.4369, | |
| "step": 449 | |
| }, | |
| { | |
| "epoch": 0.8555133079847909, | |
| "grad_norm": 0.688855649983525, | |
| "learning_rate": 1.4154150130018867e-05, | |
| "loss": 0.4258, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.8574144486692015, | |
| "grad_norm": 0.6321326951477404, | |
| "learning_rate": 1.4123919105189836e-05, | |
| "loss": 0.4248, | |
| "step": 451 | |
| }, | |
| { | |
| "epoch": 0.8593155893536122, | |
| "grad_norm": 0.6638448710072258, | |
| "learning_rate": 1.4093642599653406e-05, | |
| "loss": 0.4262, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 0.8612167300380228, | |
| "grad_norm": 0.6549201870840491, | |
| "learning_rate": 1.40633209473145e-05, | |
| "loss": 0.4169, | |
| "step": 453 | |
| }, | |
| { | |
| "epoch": 0.8631178707224335, | |
| "grad_norm": 0.6313771370577579, | |
| "learning_rate": 1.4032954482575938e-05, | |
| "loss": 0.4292, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 0.8650190114068441, | |
| "grad_norm": 0.6484955312950095, | |
| "learning_rate": 1.4002543540334766e-05, | |
| "loss": 0.4236, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.8669201520912547, | |
| "grad_norm": 0.656956449031631, | |
| "learning_rate": 1.3972088455978537e-05, | |
| "loss": 0.4253, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 0.8688212927756654, | |
| "grad_norm": 0.662583220503101, | |
| "learning_rate": 1.3941589565381635e-05, | |
| "loss": 0.4178, | |
| "step": 457 | |
| }, | |
| { | |
| "epoch": 0.870722433460076, | |
| "grad_norm": 0.7033456875213991, | |
| "learning_rate": 1.391104720490156e-05, | |
| "loss": 0.4454, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 0.8726235741444867, | |
| "grad_norm": 0.7374160060892516, | |
| "learning_rate": 1.3880461711375224e-05, | |
| "loss": 0.408, | |
| "step": 459 | |
| }, | |
| { | |
| "epoch": 0.8745247148288974, | |
| "grad_norm": 0.6686102131433704, | |
| "learning_rate": 1.3849833422115221e-05, | |
| "loss": 0.4344, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.876425855513308, | |
| "grad_norm": 0.7629027518761643, | |
| "learning_rate": 1.3819162674906134e-05, | |
| "loss": 0.4112, | |
| "step": 461 | |
| }, | |
| { | |
| "epoch": 0.8783269961977186, | |
| "grad_norm": 0.6388093480009844, | |
| "learning_rate": 1.378844980800078e-05, | |
| "loss": 0.4078, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 0.8802281368821293, | |
| "grad_norm": 0.6913803710646702, | |
| "learning_rate": 1.3757695160116502e-05, | |
| "loss": 0.445, | |
| "step": 463 | |
| }, | |
| { | |
| "epoch": 0.8821292775665399, | |
| "grad_norm": 0.6615809094332875, | |
| "learning_rate": 1.3726899070431423e-05, | |
| "loss": 0.4132, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 0.8840304182509505, | |
| "grad_norm": 0.6999713843413127, | |
| "learning_rate": 1.3696061878580707e-05, | |
| "loss": 0.4333, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.8859315589353612, | |
| "grad_norm": 0.7165602120598136, | |
| "learning_rate": 1.3665183924652817e-05, | |
| "loss": 0.4285, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 0.8878326996197718, | |
| "grad_norm": 0.6461666245422538, | |
| "learning_rate": 1.3634265549185755e-05, | |
| "loss": 0.4061, | |
| "step": 467 | |
| }, | |
| { | |
| "epoch": 0.8897338403041825, | |
| "grad_norm": 0.7215709337291267, | |
| "learning_rate": 1.3603307093163319e-05, | |
| "loss": 0.4459, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 0.8916349809885932, | |
| "grad_norm": 0.6770151518487906, | |
| "learning_rate": 1.3572308898011328e-05, | |
| "loss": 0.4146, | |
| "step": 469 | |
| }, | |
| { | |
| "epoch": 0.8935361216730038, | |
| "grad_norm": 0.6453842527053798, | |
| "learning_rate": 1.3541271305593878e-05, | |
| "loss": 0.4241, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.8954372623574145, | |
| "grad_norm": 0.7616681457699916, | |
| "learning_rate": 1.3510194658209547e-05, | |
| "loss": 0.4236, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 0.8973384030418251, | |
| "grad_norm": 0.6717336238606672, | |
| "learning_rate": 1.3479079298587634e-05, | |
| "loss": 0.4217, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 0.8992395437262357, | |
| "grad_norm": 0.6539668909382668, | |
| "learning_rate": 1.3447925569884374e-05, | |
| "loss": 0.4216, | |
| "step": 473 | |
| }, | |
| { | |
| "epoch": 0.9011406844106464, | |
| "grad_norm": 0.6744068537060459, | |
| "learning_rate": 1.3416733815679166e-05, | |
| "loss": 0.4264, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 0.903041825095057, | |
| "grad_norm": 0.6717567879048116, | |
| "learning_rate": 1.3385504379970764e-05, | |
| "loss": 0.4389, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.9049429657794676, | |
| "grad_norm": 0.6804088979571221, | |
| "learning_rate": 1.3354237607173494e-05, | |
| "loss": 0.4163, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 0.9068441064638784, | |
| "grad_norm": 0.6732741100091546, | |
| "learning_rate": 1.3322933842113457e-05, | |
| "loss": 0.4246, | |
| "step": 477 | |
| }, | |
| { | |
| "epoch": 0.908745247148289, | |
| "grad_norm": 0.6431774533831092, | |
| "learning_rate": 1.3291593430024727e-05, | |
| "loss": 0.4272, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 0.9106463878326996, | |
| "grad_norm": 0.7586987452523507, | |
| "learning_rate": 1.3260216716545534e-05, | |
| "loss": 0.4275, | |
| "step": 479 | |
| }, | |
| { | |
| "epoch": 0.9125475285171103, | |
| "grad_norm": 0.6456000679532741, | |
| "learning_rate": 1.3228804047714462e-05, | |
| "loss": 0.4096, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.9144486692015209, | |
| "grad_norm": 0.7059982640469601, | |
| "learning_rate": 1.319735576996663e-05, | |
| "loss": 0.4421, | |
| "step": 481 | |
| }, | |
| { | |
| "epoch": 0.9163498098859315, | |
| "grad_norm": 0.6526420230151465, | |
| "learning_rate": 1.3165872230129869e-05, | |
| "loss": 0.4219, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 0.9182509505703422, | |
| "grad_norm": 0.6786042105569517, | |
| "learning_rate": 1.3134353775420895e-05, | |
| "loss": 0.4125, | |
| "step": 483 | |
| }, | |
| { | |
| "epoch": 0.9201520912547528, | |
| "grad_norm": 0.6693710839527324, | |
| "learning_rate": 1.3102800753441488e-05, | |
| "loss": 0.4198, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 0.9220532319391636, | |
| "grad_norm": 0.5918993502199404, | |
| "learning_rate": 1.3071213512174655e-05, | |
| "loss": 0.386, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.9239543726235742, | |
| "grad_norm": 0.6746150514169127, | |
| "learning_rate": 1.3039592399980785e-05, | |
| "loss": 0.399, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 0.9258555133079848, | |
| "grad_norm": 0.6527752188311693, | |
| "learning_rate": 1.3007937765593818e-05, | |
| "loss": 0.4356, | |
| "step": 487 | |
| }, | |
| { | |
| "epoch": 0.9277566539923955, | |
| "grad_norm": 0.6770488025653373, | |
| "learning_rate": 1.2976249958117395e-05, | |
| "loss": 0.4293, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 0.9296577946768061, | |
| "grad_norm": 0.659256555706261, | |
| "learning_rate": 1.2944529327021002e-05, | |
| "loss": 0.4132, | |
| "step": 489 | |
| }, | |
| { | |
| "epoch": 0.9315589353612167, | |
| "grad_norm": 0.6391553757253262, | |
| "learning_rate": 1.291277622213612e-05, | |
| "loss": 0.4052, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.9334600760456274, | |
| "grad_norm": 0.6485286268222028, | |
| "learning_rate": 1.2880990993652379e-05, | |
| "loss": 0.4289, | |
| "step": 491 | |
| }, | |
| { | |
| "epoch": 0.935361216730038, | |
| "grad_norm": 0.665755527198907, | |
| "learning_rate": 1.2849173992113669e-05, | |
| "loss": 0.4248, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 0.9372623574144486, | |
| "grad_norm": 0.6642338936130883, | |
| "learning_rate": 1.2817325568414299e-05, | |
| "loss": 0.4156, | |
| "step": 493 | |
| }, | |
| { | |
| "epoch": 0.9391634980988594, | |
| "grad_norm": 0.644585517729116, | |
| "learning_rate": 1.2785446073795118e-05, | |
| "loss": 0.4082, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 0.94106463878327, | |
| "grad_norm": 0.745715005249675, | |
| "learning_rate": 1.2753535859839638e-05, | |
| "loss": 0.4179, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.9429657794676806, | |
| "grad_norm": 0.6217374039298305, | |
| "learning_rate": 1.272159527847016e-05, | |
| "loss": 0.4042, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 0.9448669201520913, | |
| "grad_norm": 0.670510313330144, | |
| "learning_rate": 1.2689624681943897e-05, | |
| "loss": 0.4455, | |
| "step": 497 | |
| }, | |
| { | |
| "epoch": 0.9467680608365019, | |
| "grad_norm": 0.6872909174712374, | |
| "learning_rate": 1.2657624422849077e-05, | |
| "loss": 0.4273, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 0.9486692015209125, | |
| "grad_norm": 0.6780355836634759, | |
| "learning_rate": 1.2625594854101066e-05, | |
| "loss": 0.4261, | |
| "step": 499 | |
| }, | |
| { | |
| "epoch": 0.9505703422053232, | |
| "grad_norm": 0.6642064785471654, | |
| "learning_rate": 1.2593536328938471e-05, | |
| "loss": 0.4236, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.9524714828897338, | |
| "grad_norm": 0.6145685165927349, | |
| "learning_rate": 1.2561449200919253e-05, | |
| "loss": 0.3974, | |
| "step": 501 | |
| }, | |
| { | |
| "epoch": 0.9543726235741445, | |
| "grad_norm": 0.718311203068776, | |
| "learning_rate": 1.2529333823916807e-05, | |
| "loss": 0.3998, | |
| "step": 502 | |
| }, | |
| { | |
| "epoch": 0.9562737642585551, | |
| "grad_norm": 0.6626553757532538, | |
| "learning_rate": 1.2497190552116082e-05, | |
| "loss": 0.4295, | |
| "step": 503 | |
| }, | |
| { | |
| "epoch": 0.9581749049429658, | |
| "grad_norm": 0.654130247793398, | |
| "learning_rate": 1.2465019740009662e-05, | |
| "loss": 0.4009, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 0.9600760456273765, | |
| "grad_norm": 0.7352584819357333, | |
| "learning_rate": 1.2432821742393854e-05, | |
| "loss": 0.419, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 0.9619771863117871, | |
| "grad_norm": 0.6632601475224529, | |
| "learning_rate": 1.2400596914364792e-05, | |
| "loss": 0.4167, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 0.9638783269961977, | |
| "grad_norm": 0.7329668969062761, | |
| "learning_rate": 1.2368345611314508e-05, | |
| "loss": 0.4205, | |
| "step": 507 | |
| }, | |
| { | |
| "epoch": 0.9657794676806084, | |
| "grad_norm": 0.6106373023588458, | |
| "learning_rate": 1.2336068188927002e-05, | |
| "loss": 0.3997, | |
| "step": 508 | |
| }, | |
| { | |
| "epoch": 0.967680608365019, | |
| "grad_norm": 0.6459626459087109, | |
| "learning_rate": 1.2303765003174342e-05, | |
| "loss": 0.4214, | |
| "step": 509 | |
| }, | |
| { | |
| "epoch": 0.9695817490494296, | |
| "grad_norm": 0.65402687722963, | |
| "learning_rate": 1.2271436410312727e-05, | |
| "loss": 0.4542, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.9714828897338403, | |
| "grad_norm": 0.6694991446957526, | |
| "learning_rate": 1.2239082766878557e-05, | |
| "loss": 0.4222, | |
| "step": 511 | |
| }, | |
| { | |
| "epoch": 0.973384030418251, | |
| "grad_norm": 0.6928486520358509, | |
| "learning_rate": 1.2206704429684504e-05, | |
| "loss": 0.4191, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 0.9752851711026616, | |
| "grad_norm": 0.6329104242278863, | |
| "learning_rate": 1.2174301755815572e-05, | |
| "loss": 0.4232, | |
| "step": 513 | |
| }, | |
| { | |
| "epoch": 0.9771863117870723, | |
| "grad_norm": 0.6332053708166463, | |
| "learning_rate": 1.2141875102625166e-05, | |
| "loss": 0.412, | |
| "step": 514 | |
| }, | |
| { | |
| "epoch": 0.9790874524714829, | |
| "grad_norm": 0.6764515794970526, | |
| "learning_rate": 1.2109424827731144e-05, | |
| "loss": 0.435, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 0.9809885931558935, | |
| "grad_norm": 0.6424890801593467, | |
| "learning_rate": 1.2076951289011884e-05, | |
| "loss": 0.4328, | |
| "step": 516 | |
| }, | |
| { | |
| "epoch": 0.9828897338403042, | |
| "grad_norm": 0.6692580779196105, | |
| "learning_rate": 1.204445484460232e-05, | |
| "loss": 0.4317, | |
| "step": 517 | |
| }, | |
| { | |
| "epoch": 0.9847908745247148, | |
| "grad_norm": 0.6593325074461684, | |
| "learning_rate": 1.2011935852890004e-05, | |
| "loss": 0.4194, | |
| "step": 518 | |
| }, | |
| { | |
| "epoch": 0.9866920152091255, | |
| "grad_norm": 0.6288279629501167, | |
| "learning_rate": 1.1979394672511156e-05, | |
| "loss": 0.392, | |
| "step": 519 | |
| }, | |
| { | |
| "epoch": 0.9885931558935361, | |
| "grad_norm": 0.6849376387633753, | |
| "learning_rate": 1.19468316623467e-05, | |
| "loss": 0.4233, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.9904942965779467, | |
| "grad_norm": 0.661605094801409, | |
| "learning_rate": 1.1914247181518312e-05, | |
| "loss": 0.4035, | |
| "step": 521 | |
| }, | |
| { | |
| "epoch": 0.9923954372623575, | |
| "grad_norm": 0.7016232148582198, | |
| "learning_rate": 1.1881641589384456e-05, | |
| "loss": 0.4236, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 0.9942965779467681, | |
| "grad_norm": 0.6511937343652139, | |
| "learning_rate": 1.1849015245536424e-05, | |
| "loss": 0.4336, | |
| "step": 523 | |
| }, | |
| { | |
| "epoch": 0.9961977186311787, | |
| "grad_norm": 0.6666873500706701, | |
| "learning_rate": 1.1816368509794365e-05, | |
| "loss": 0.4148, | |
| "step": 524 | |
| }, | |
| { | |
| "epoch": 0.9980988593155894, | |
| "grad_norm": 0.613840476872732, | |
| "learning_rate": 1.1783701742203326e-05, | |
| "loss": 0.3902, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 0.7220188313535454, | |
| "learning_rate": 1.1751015303029272e-05, | |
| "loss": 0.4128, | |
| "step": 526 | |
| }, | |
| { | |
| "epoch": 1.0019011406844107, | |
| "grad_norm": 0.7938902982800389, | |
| "learning_rate": 1.1718309552755118e-05, | |
| "loss": 0.3281, | |
| "step": 527 | |
| }, | |
| { | |
| "epoch": 1.0038022813688212, | |
| "grad_norm": 0.7710418537178484, | |
| "learning_rate": 1.1685584852076746e-05, | |
| "loss": 0.3344, | |
| "step": 528 | |
| }, | |
| { | |
| "epoch": 1.005703422053232, | |
| "grad_norm": 0.6588876219838334, | |
| "learning_rate": 1.1652841561899042e-05, | |
| "loss": 0.33, | |
| "step": 529 | |
| }, | |
| { | |
| "epoch": 1.0076045627376427, | |
| "grad_norm": 0.6670065353733651, | |
| "learning_rate": 1.1620080043331901e-05, | |
| "loss": 0.3119, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 1.0095057034220531, | |
| "grad_norm": 0.7320990528950818, | |
| "learning_rate": 1.1587300657686254e-05, | |
| "loss": 0.3221, | |
| "step": 531 | |
| }, | |
| { | |
| "epoch": 1.0114068441064639, | |
| "grad_norm": 0.8361019502675807, | |
| "learning_rate": 1.1554503766470069e-05, | |
| "loss": 0.3257, | |
| "step": 532 | |
| }, | |
| { | |
| "epoch": 1.0133079847908746, | |
| "grad_norm": 0.8087176940985482, | |
| "learning_rate": 1.1521689731384391e-05, | |
| "loss": 0.3298, | |
| "step": 533 | |
| }, | |
| { | |
| "epoch": 1.015209125475285, | |
| "grad_norm": 0.8017009458803728, | |
| "learning_rate": 1.1488858914319321e-05, | |
| "loss": 0.32, | |
| "step": 534 | |
| }, | |
| { | |
| "epoch": 1.0171102661596958, | |
| "grad_norm": 0.6768484899929694, | |
| "learning_rate": 1.1456011677350052e-05, | |
| "loss": 0.3086, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 1.0190114068441065, | |
| "grad_norm": 0.6967484680507494, | |
| "learning_rate": 1.1423148382732854e-05, | |
| "loss": 0.322, | |
| "step": 536 | |
| }, | |
| { | |
| "epoch": 1.020912547528517, | |
| "grad_norm": 0.7156227684922336, | |
| "learning_rate": 1.1390269392901096e-05, | |
| "loss": 0.307, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 1.0228136882129277, | |
| "grad_norm": 0.6659704012560265, | |
| "learning_rate": 1.1357375070461241e-05, | |
| "loss": 0.3102, | |
| "step": 538 | |
| }, | |
| { | |
| "epoch": 1.0247148288973384, | |
| "grad_norm": 0.7046590488149735, | |
| "learning_rate": 1.1324465778188846e-05, | |
| "loss": 0.3272, | |
| "step": 539 | |
| }, | |
| { | |
| "epoch": 1.026615969581749, | |
| "grad_norm": 0.6721601117930766, | |
| "learning_rate": 1.1291541879024568e-05, | |
| "loss": 0.3318, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 1.0285171102661597, | |
| "grad_norm": 0.6692242575472586, | |
| "learning_rate": 1.1258603736070145e-05, | |
| "loss": 0.2994, | |
| "step": 541 | |
| }, | |
| { | |
| "epoch": 1.0304182509505704, | |
| "grad_norm": 0.7032457414639703, | |
| "learning_rate": 1.1225651712584413e-05, | |
| "loss": 0.3168, | |
| "step": 542 | |
| }, | |
| { | |
| "epoch": 1.032319391634981, | |
| "grad_norm": 0.6828807917013936, | |
| "learning_rate": 1.1192686171979288e-05, | |
| "loss": 0.3203, | |
| "step": 543 | |
| }, | |
| { | |
| "epoch": 1.0342205323193916, | |
| "grad_norm": 0.7272600523459498, | |
| "learning_rate": 1.1159707477815756e-05, | |
| "loss": 0.3032, | |
| "step": 544 | |
| }, | |
| { | |
| "epoch": 1.0361216730038023, | |
| "grad_norm": 0.7190024950121244, | |
| "learning_rate": 1.1126715993799875e-05, | |
| "loss": 0.3102, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 1.038022813688213, | |
| "grad_norm": 0.7054833345167262, | |
| "learning_rate": 1.1093712083778748e-05, | |
| "loss": 0.3086, | |
| "step": 546 | |
| }, | |
| { | |
| "epoch": 1.0399239543726235, | |
| "grad_norm": 0.6827971285645381, | |
| "learning_rate": 1.1060696111736515e-05, | |
| "loss": 0.3132, | |
| "step": 547 | |
| }, | |
| { | |
| "epoch": 1.0418250950570342, | |
| "grad_norm": 0.694040452693297, | |
| "learning_rate": 1.1027668441790358e-05, | |
| "loss": 0.3019, | |
| "step": 548 | |
| }, | |
| { | |
| "epoch": 1.043726235741445, | |
| "grad_norm": 0.702207814637265, | |
| "learning_rate": 1.099462943818646e-05, | |
| "loss": 0.3199, | |
| "step": 549 | |
| }, | |
| { | |
| "epoch": 1.0456273764258555, | |
| "grad_norm": 0.7317299133154515, | |
| "learning_rate": 1.0961579465295987e-05, | |
| "loss": 0.297, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.0475285171102662, | |
| "grad_norm": 0.7496286358979238, | |
| "learning_rate": 1.0928518887611099e-05, | |
| "loss": 0.3208, | |
| "step": 551 | |
| }, | |
| { | |
| "epoch": 1.049429657794677, | |
| "grad_norm": 0.7051496988239422, | |
| "learning_rate": 1.0895448069740902e-05, | |
| "loss": 0.3175, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 1.0513307984790874, | |
| "grad_norm": 0.8011345894922628, | |
| "learning_rate": 1.0862367376407433e-05, | |
| "loss": 0.3018, | |
| "step": 553 | |
| }, | |
| { | |
| "epoch": 1.053231939163498, | |
| "grad_norm": 0.7191736733193583, | |
| "learning_rate": 1.0829277172441648e-05, | |
| "loss": 0.3164, | |
| "step": 554 | |
| }, | |
| { | |
| "epoch": 1.0551330798479088, | |
| "grad_norm": 0.6443370056628915, | |
| "learning_rate": 1.0796177822779384e-05, | |
| "loss": 0.3027, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 1.0570342205323193, | |
| "grad_norm": 0.7688161850207594, | |
| "learning_rate": 1.0763069692457346e-05, | |
| "loss": 0.3108, | |
| "step": 556 | |
| }, | |
| { | |
| "epoch": 1.05893536121673, | |
| "grad_norm": 0.6975214565791557, | |
| "learning_rate": 1.0729953146609076e-05, | |
| "loss": 0.3181, | |
| "step": 557 | |
| }, | |
| { | |
| "epoch": 1.0608365019011408, | |
| "grad_norm": 0.7002492398021611, | |
| "learning_rate": 1.0696828550460928e-05, | |
| "loss": 0.3089, | |
| "step": 558 | |
| }, | |
| { | |
| "epoch": 1.0627376425855513, | |
| "grad_norm": 0.753789638960337, | |
| "learning_rate": 1.0663696269328034e-05, | |
| "loss": 0.3236, | |
| "step": 559 | |
| }, | |
| { | |
| "epoch": 1.064638783269962, | |
| "grad_norm": 0.801912153585207, | |
| "learning_rate": 1.0630556668610286e-05, | |
| "loss": 0.29, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 1.0665399239543727, | |
| "grad_norm": 0.7306418697466098, | |
| "learning_rate": 1.059741011378829e-05, | |
| "loss": 0.2979, | |
| "step": 561 | |
| }, | |
| { | |
| "epoch": 1.0684410646387832, | |
| "grad_norm": 0.6812672181140943, | |
| "learning_rate": 1.0564256970419367e-05, | |
| "loss": 0.3086, | |
| "step": 562 | |
| }, | |
| { | |
| "epoch": 1.070342205323194, | |
| "grad_norm": 0.8258781587521762, | |
| "learning_rate": 1.0531097604133473e-05, | |
| "loss": 0.3062, | |
| "step": 563 | |
| }, | |
| { | |
| "epoch": 1.0722433460076046, | |
| "grad_norm": 0.8594408865792439, | |
| "learning_rate": 1.0497932380629207e-05, | |
| "loss": 0.3096, | |
| "step": 564 | |
| }, | |
| { | |
| "epoch": 1.0741444866920151, | |
| "grad_norm": 0.6682275197078438, | |
| "learning_rate": 1.0464761665669771e-05, | |
| "loss": 0.322, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 1.0760456273764258, | |
| "grad_norm": 0.9387510593538437, | |
| "learning_rate": 1.0431585825078916e-05, | |
| "loss": 0.3181, | |
| "step": 566 | |
| }, | |
| { | |
| "epoch": 1.0779467680608366, | |
| "grad_norm": 0.8781806060809328, | |
| "learning_rate": 1.0398405224736927e-05, | |
| "loss": 0.3068, | |
| "step": 567 | |
| }, | |
| { | |
| "epoch": 1.079847908745247, | |
| "grad_norm": 0.7082452337737895, | |
| "learning_rate": 1.0365220230576592e-05, | |
| "loss": 0.3055, | |
| "step": 568 | |
| }, | |
| { | |
| "epoch": 1.0817490494296578, | |
| "grad_norm": 0.8337658875558616, | |
| "learning_rate": 1.0332031208579133e-05, | |
| "loss": 0.3184, | |
| "step": 569 | |
| }, | |
| { | |
| "epoch": 1.0836501901140685, | |
| "grad_norm": 0.7599118793898617, | |
| "learning_rate": 1.0298838524770212e-05, | |
| "loss": 0.2952, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 1.085551330798479, | |
| "grad_norm": 0.6534766509118429, | |
| "learning_rate": 1.0265642545215872e-05, | |
| "loss": 0.312, | |
| "step": 571 | |
| }, | |
| { | |
| "epoch": 1.0874524714828897, | |
| "grad_norm": 0.6975672983616702, | |
| "learning_rate": 1.0232443636018502e-05, | |
| "loss": 0.3287, | |
| "step": 572 | |
| }, | |
| { | |
| "epoch": 1.0893536121673004, | |
| "grad_norm": 0.9082610382277593, | |
| "learning_rate": 1.0199242163312794e-05, | |
| "loss": 0.2998, | |
| "step": 573 | |
| }, | |
| { | |
| "epoch": 1.091254752851711, | |
| "grad_norm": 0.634021008852441, | |
| "learning_rate": 1.0166038493261723e-05, | |
| "loss": 0.299, | |
| "step": 574 | |
| }, | |
| { | |
| "epoch": 1.0931558935361216, | |
| "grad_norm": 0.723084580740578, | |
| "learning_rate": 1.013283299205249e-05, | |
| "loss": 0.309, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 1.0950570342205324, | |
| "grad_norm": 0.6915487377322532, | |
| "learning_rate": 1.0099626025892491e-05, | |
| "loss": 0.2969, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 1.0969581749049429, | |
| "grad_norm": 0.7062556639156321, | |
| "learning_rate": 1.0066417961005283e-05, | |
| "loss": 0.309, | |
| "step": 577 | |
| }, | |
| { | |
| "epoch": 1.0988593155893536, | |
| "grad_norm": 0.6800057765118989, | |
| "learning_rate": 1.0033209163626539e-05, | |
| "loss": 0.3174, | |
| "step": 578 | |
| }, | |
| { | |
| "epoch": 1.1007604562737643, | |
| "grad_norm": 0.7139064812806064, | |
| "learning_rate": 1e-05, | |
| "loss": 0.3084, | |
| "step": 579 | |
| }, | |
| { | |
| "epoch": 1.102661596958175, | |
| "grad_norm": 0.7822597316470364, | |
| "learning_rate": 9.966790836373465e-06, | |
| "loss": 0.2984, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 1.1045627376425855, | |
| "grad_norm": 0.6919829401110732, | |
| "learning_rate": 9.933582038994719e-06, | |
| "loss": 0.3044, | |
| "step": 581 | |
| }, | |
| { | |
| "epoch": 1.1064638783269962, | |
| "grad_norm": 0.7545979295915479, | |
| "learning_rate": 9.90037397410751e-06, | |
| "loss": 0.3036, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 1.108365019011407, | |
| "grad_norm": 0.7021792330437671, | |
| "learning_rate": 9.867167007947511e-06, | |
| "loss": 0.3142, | |
| "step": 583 | |
| }, | |
| { | |
| "epoch": 1.1102661596958174, | |
| "grad_norm": 0.6913586393114494, | |
| "learning_rate": 9.833961506738282e-06, | |
| "loss": 0.3216, | |
| "step": 584 | |
| }, | |
| { | |
| "epoch": 1.1121673003802282, | |
| "grad_norm": 0.6867501973190906, | |
| "learning_rate": 9.80075783668721e-06, | |
| "loss": 0.299, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 1.1140684410646389, | |
| "grad_norm": 0.7478263914018145, | |
| "learning_rate": 9.767556363981503e-06, | |
| "loss": 0.3285, | |
| "step": 586 | |
| }, | |
| { | |
| "epoch": 1.1159695817490494, | |
| "grad_norm": 0.6630697975763085, | |
| "learning_rate": 9.734357454784131e-06, | |
| "loss": 0.3069, | |
| "step": 587 | |
| }, | |
| { | |
| "epoch": 1.11787072243346, | |
| "grad_norm": 0.6707275675557878, | |
| "learning_rate": 9.701161475229791e-06, | |
| "loss": 0.309, | |
| "step": 588 | |
| }, | |
| { | |
| "epoch": 1.1197718631178708, | |
| "grad_norm": 0.6706410134490758, | |
| "learning_rate": 9.66796879142087e-06, | |
| "loss": 0.2992, | |
| "step": 589 | |
| }, | |
| { | |
| "epoch": 1.1216730038022813, | |
| "grad_norm": 0.6649865652528664, | |
| "learning_rate": 9.634779769423412e-06, | |
| "loss": 0.2951, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 1.123574144486692, | |
| "grad_norm": 0.7208037488589544, | |
| "learning_rate": 9.601594775263073e-06, | |
| "loss": 0.3093, | |
| "step": 591 | |
| }, | |
| { | |
| "epoch": 1.1254752851711027, | |
| "grad_norm": 0.7489663081585506, | |
| "learning_rate": 9.568414174921085e-06, | |
| "loss": 0.3125, | |
| "step": 592 | |
| }, | |
| { | |
| "epoch": 1.1273764258555132, | |
| "grad_norm": 0.7124994075715532, | |
| "learning_rate": 9.535238334330234e-06, | |
| "loss": 0.2912, | |
| "step": 593 | |
| }, | |
| { | |
| "epoch": 1.129277566539924, | |
| "grad_norm": 0.7125344213306348, | |
| "learning_rate": 9.502067619370794e-06, | |
| "loss": 0.3048, | |
| "step": 594 | |
| }, | |
| { | |
| "epoch": 1.1311787072243347, | |
| "grad_norm": 0.7445753403912206, | |
| "learning_rate": 9.468902395866532e-06, | |
| "loss": 0.3044, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 1.1330798479087452, | |
| "grad_norm": 0.7597430750499627, | |
| "learning_rate": 9.435743029580638e-06, | |
| "loss": 0.2982, | |
| "step": 596 | |
| }, | |
| { | |
| "epoch": 1.1349809885931559, | |
| "grad_norm": 0.6694566298828142, | |
| "learning_rate": 9.402589886211711e-06, | |
| "loss": 0.3031, | |
| "step": 597 | |
| }, | |
| { | |
| "epoch": 1.1368821292775666, | |
| "grad_norm": 0.702742968343329, | |
| "learning_rate": 9.369443331389718e-06, | |
| "loss": 0.3268, | |
| "step": 598 | |
| }, | |
| { | |
| "epoch": 1.138783269961977, | |
| "grad_norm": 0.7996938849473036, | |
| "learning_rate": 9.336303730671968e-06, | |
| "loss": 0.3078, | |
| "step": 599 | |
| }, | |
| { | |
| "epoch": 1.1406844106463878, | |
| "grad_norm": 0.7194860814635625, | |
| "learning_rate": 9.303171449539074e-06, | |
| "loss": 0.3114, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.1425855513307985, | |
| "grad_norm": 0.6962744537190348, | |
| "learning_rate": 9.270046853390924e-06, | |
| "loss": 0.291, | |
| "step": 601 | |
| }, | |
| { | |
| "epoch": 1.144486692015209, | |
| "grad_norm": 0.7383607387197505, | |
| "learning_rate": 9.236930307542654e-06, | |
| "loss": 0.3082, | |
| "step": 602 | |
| }, | |
| { | |
| "epoch": 1.1463878326996197, | |
| "grad_norm": 0.6693795195758491, | |
| "learning_rate": 9.203822177220621e-06, | |
| "loss": 0.2923, | |
| "step": 603 | |
| }, | |
| { | |
| "epoch": 1.1482889733840305, | |
| "grad_norm": 0.7654729190391708, | |
| "learning_rate": 9.170722827558357e-06, | |
| "loss": 0.3245, | |
| "step": 604 | |
| }, | |
| { | |
| "epoch": 1.1501901140684412, | |
| "grad_norm": 0.742544399270309, | |
| "learning_rate": 9.13763262359257e-06, | |
| "loss": 0.3216, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 1.1520912547528517, | |
| "grad_norm": 0.7009665092281706, | |
| "learning_rate": 9.104551930259101e-06, | |
| "loss": 0.3041, | |
| "step": 606 | |
| }, | |
| { | |
| "epoch": 1.1539923954372624, | |
| "grad_norm": 0.7126410097325343, | |
| "learning_rate": 9.071481112388905e-06, | |
| "loss": 0.3217, | |
| "step": 607 | |
| }, | |
| { | |
| "epoch": 1.1558935361216731, | |
| "grad_norm": 0.6652375515738913, | |
| "learning_rate": 9.038420534704015e-06, | |
| "loss": 0.3096, | |
| "step": 608 | |
| }, | |
| { | |
| "epoch": 1.1577946768060836, | |
| "grad_norm": 0.7633170942763475, | |
| "learning_rate": 9.005370561813545e-06, | |
| "loss": 0.3206, | |
| "step": 609 | |
| }, | |
| { | |
| "epoch": 1.1596958174904943, | |
| "grad_norm": 0.7552903350439375, | |
| "learning_rate": 8.972331558209644e-06, | |
| "loss": 0.3254, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 1.161596958174905, | |
| "grad_norm": 0.6406068379021298, | |
| "learning_rate": 8.939303888263485e-06, | |
| "loss": 0.3072, | |
| "step": 611 | |
| }, | |
| { | |
| "epoch": 1.1634980988593155, | |
| "grad_norm": 0.6662184576443853, | |
| "learning_rate": 8.906287916221259e-06, | |
| "loss": 0.3079, | |
| "step": 612 | |
| }, | |
| { | |
| "epoch": 1.1653992395437263, | |
| "grad_norm": 0.8267285203331385, | |
| "learning_rate": 8.873284006200129e-06, | |
| "loss": 0.3027, | |
| "step": 613 | |
| }, | |
| { | |
| "epoch": 1.167300380228137, | |
| "grad_norm": 0.6812130104045184, | |
| "learning_rate": 8.840292522184247e-06, | |
| "loss": 0.2921, | |
| "step": 614 | |
| }, | |
| { | |
| "epoch": 1.1692015209125475, | |
| "grad_norm": 0.6936642858306603, | |
| "learning_rate": 8.807313828020715e-06, | |
| "loss": 0.3285, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 1.1711026615969582, | |
| "grad_norm": 0.795822470872818, | |
| "learning_rate": 8.774348287415589e-06, | |
| "loss": 0.3143, | |
| "step": 616 | |
| }, | |
| { | |
| "epoch": 1.173003802281369, | |
| "grad_norm": 0.704238172102989, | |
| "learning_rate": 8.74139626392986e-06, | |
| "loss": 0.3223, | |
| "step": 617 | |
| }, | |
| { | |
| "epoch": 1.1749049429657794, | |
| "grad_norm": 0.7193899031212281, | |
| "learning_rate": 8.708458120975436e-06, | |
| "loss": 0.3063, | |
| "step": 618 | |
| }, | |
| { | |
| "epoch": 1.1768060836501901, | |
| "grad_norm": 0.7429543284494067, | |
| "learning_rate": 8.675534221811156e-06, | |
| "loss": 0.305, | |
| "step": 619 | |
| }, | |
| { | |
| "epoch": 1.1787072243346008, | |
| "grad_norm": 0.6747514013477584, | |
| "learning_rate": 8.64262492953876e-06, | |
| "loss": 0.3071, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 1.1806083650190113, | |
| "grad_norm": 0.7540621765837212, | |
| "learning_rate": 8.60973060709891e-06, | |
| "loss": 0.3347, | |
| "step": 621 | |
| }, | |
| { | |
| "epoch": 1.182509505703422, | |
| "grad_norm": 0.6686865609476355, | |
| "learning_rate": 8.576851617267151e-06, | |
| "loss": 0.3116, | |
| "step": 622 | |
| }, | |
| { | |
| "epoch": 1.1844106463878328, | |
| "grad_norm": 0.7360122163711854, | |
| "learning_rate": 8.543988322649954e-06, | |
| "loss": 0.3218, | |
| "step": 623 | |
| }, | |
| { | |
| "epoch": 1.1863117870722433, | |
| "grad_norm": 0.7132387070914369, | |
| "learning_rate": 8.511141085680684e-06, | |
| "loss": 0.3234, | |
| "step": 624 | |
| }, | |
| { | |
| "epoch": 1.188212927756654, | |
| "grad_norm": 0.686612573171493, | |
| "learning_rate": 8.478310268615612e-06, | |
| "loss": 0.3159, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 1.1901140684410647, | |
| "grad_norm": 0.7142051237784038, | |
| "learning_rate": 8.445496233529934e-06, | |
| "loss": 0.3096, | |
| "step": 626 | |
| }, | |
| { | |
| "epoch": 1.1920152091254752, | |
| "grad_norm": 0.7035998589690597, | |
| "learning_rate": 8.41269934231375e-06, | |
| "loss": 0.3371, | |
| "step": 627 | |
| }, | |
| { | |
| "epoch": 1.193916349809886, | |
| "grad_norm": 0.6820808233068352, | |
| "learning_rate": 8.3799199566681e-06, | |
| "loss": 0.2963, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 1.1958174904942966, | |
| "grad_norm": 0.6341171405976705, | |
| "learning_rate": 8.34715843810096e-06, | |
| "loss": 0.2908, | |
| "step": 629 | |
| }, | |
| { | |
| "epoch": 1.1977186311787071, | |
| "grad_norm": 0.7096354479902686, | |
| "learning_rate": 8.314415147923254e-06, | |
| "loss": 0.3192, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 1.1996197718631179, | |
| "grad_norm": 0.6773611800368086, | |
| "learning_rate": 8.281690447244887e-06, | |
| "loss": 0.2975, | |
| "step": 631 | |
| }, | |
| { | |
| "epoch": 1.2015209125475286, | |
| "grad_norm": 0.7137046028229448, | |
| "learning_rate": 8.248984696970732e-06, | |
| "loss": 0.3049, | |
| "step": 632 | |
| }, | |
| { | |
| "epoch": 1.203422053231939, | |
| "grad_norm": 0.6973273531506592, | |
| "learning_rate": 8.216298257796677e-06, | |
| "loss": 0.3535, | |
| "step": 633 | |
| }, | |
| { | |
| "epoch": 1.2053231939163498, | |
| "grad_norm": 0.6780366775714444, | |
| "learning_rate": 8.183631490205636e-06, | |
| "loss": 0.3004, | |
| "step": 634 | |
| }, | |
| { | |
| "epoch": 1.2072243346007605, | |
| "grad_norm": 0.7126495254064852, | |
| "learning_rate": 8.150984754463578e-06, | |
| "loss": 0.3136, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 1.209125475285171, | |
| "grad_norm": 0.7076475605667071, | |
| "learning_rate": 8.118358410615545e-06, | |
| "loss": 0.3048, | |
| "step": 636 | |
| }, | |
| { | |
| "epoch": 1.2110266159695817, | |
| "grad_norm": 0.7408441793817443, | |
| "learning_rate": 8.08575281848169e-06, | |
| "loss": 0.2945, | |
| "step": 637 | |
| }, | |
| { | |
| "epoch": 1.2129277566539924, | |
| "grad_norm": 0.7219749936544865, | |
| "learning_rate": 8.0531683376533e-06, | |
| "loss": 0.2941, | |
| "step": 638 | |
| }, | |
| { | |
| "epoch": 1.214828897338403, | |
| "grad_norm": 0.667494244968256, | |
| "learning_rate": 8.020605327488846e-06, | |
| "loss": 0.3131, | |
| "step": 639 | |
| }, | |
| { | |
| "epoch": 1.2167300380228137, | |
| "grad_norm": 0.7686295999791597, | |
| "learning_rate": 7.988064147110001e-06, | |
| "loss": 0.3154, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 1.2186311787072244, | |
| "grad_norm": 0.7407037835116259, | |
| "learning_rate": 7.955545155397684e-06, | |
| "loss": 0.3057, | |
| "step": 641 | |
| }, | |
| { | |
| "epoch": 1.2205323193916349, | |
| "grad_norm": 0.6981577214679701, | |
| "learning_rate": 7.923048710988119e-06, | |
| "loss": 0.3174, | |
| "step": 642 | |
| }, | |
| { | |
| "epoch": 1.2224334600760456, | |
| "grad_norm": 0.7393698291756108, | |
| "learning_rate": 7.890575172268858e-06, | |
| "loss": 0.3248, | |
| "step": 643 | |
| }, | |
| { | |
| "epoch": 1.2243346007604563, | |
| "grad_norm": 0.7274077693988548, | |
| "learning_rate": 7.858124897374837e-06, | |
| "loss": 0.2984, | |
| "step": 644 | |
| }, | |
| { | |
| "epoch": 1.2262357414448668, | |
| "grad_norm": 0.7001334809953396, | |
| "learning_rate": 7.825698244184432e-06, | |
| "loss": 0.2963, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 1.2281368821292775, | |
| "grad_norm": 0.6912628746706484, | |
| "learning_rate": 7.7932955703155e-06, | |
| "loss": 0.2949, | |
| "step": 646 | |
| }, | |
| { | |
| "epoch": 1.2300380228136882, | |
| "grad_norm": 0.7525462900276383, | |
| "learning_rate": 7.760917233121443e-06, | |
| "loss": 0.3215, | |
| "step": 647 | |
| }, | |
| { | |
| "epoch": 1.231939163498099, | |
| "grad_norm": 0.7616626967163781, | |
| "learning_rate": 7.728563589687275e-06, | |
| "loss": 0.2989, | |
| "step": 648 | |
| }, | |
| { | |
| "epoch": 1.2338403041825095, | |
| "grad_norm": 0.6973443613016403, | |
| "learning_rate": 7.696234996825663e-06, | |
| "loss": 0.3107, | |
| "step": 649 | |
| }, | |
| { | |
| "epoch": 1.2357414448669202, | |
| "grad_norm": 0.6586019595160517, | |
| "learning_rate": 7.663931811073003e-06, | |
| "loss": 0.312, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.2376425855513309, | |
| "grad_norm": 0.7544869284916628, | |
| "learning_rate": 7.631654388685496e-06, | |
| "loss": 0.3153, | |
| "step": 651 | |
| }, | |
| { | |
| "epoch": 1.2395437262357414, | |
| "grad_norm": 0.75562124043933, | |
| "learning_rate": 7.599403085635208e-06, | |
| "loss": 0.3086, | |
| "step": 652 | |
| }, | |
| { | |
| "epoch": 1.241444866920152, | |
| "grad_norm": 0.7283788553688064, | |
| "learning_rate": 7.567178257606147e-06, | |
| "loss": 0.3117, | |
| "step": 653 | |
| }, | |
| { | |
| "epoch": 1.2433460076045628, | |
| "grad_norm": 0.6911635540159975, | |
| "learning_rate": 7.534980259990341e-06, | |
| "loss": 0.3027, | |
| "step": 654 | |
| }, | |
| { | |
| "epoch": 1.2452471482889733, | |
| "grad_norm": 0.7194684435446679, | |
| "learning_rate": 7.50280944788392e-06, | |
| "loss": 0.3086, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 1.247148288973384, | |
| "grad_norm": 0.7595594224939554, | |
| "learning_rate": 7.470666176083193e-06, | |
| "loss": 0.3209, | |
| "step": 656 | |
| }, | |
| { | |
| "epoch": 1.2490494296577948, | |
| "grad_norm": 0.7247414717009433, | |
| "learning_rate": 7.438550799080746e-06, | |
| "loss": 0.3105, | |
| "step": 657 | |
| }, | |
| { | |
| "epoch": 1.2509505703422052, | |
| "grad_norm": 0.6588781359568728, | |
| "learning_rate": 7.40646367106153e-06, | |
| "loss": 0.3052, | |
| "step": 658 | |
| }, | |
| { | |
| "epoch": 1.252851711026616, | |
| "grad_norm": 0.72590626388867, | |
| "learning_rate": 7.3744051458989395e-06, | |
| "loss": 0.3062, | |
| "step": 659 | |
| }, | |
| { | |
| "epoch": 1.2547528517110267, | |
| "grad_norm": 0.7679108764797695, | |
| "learning_rate": 7.342375577150928e-06, | |
| "loss": 0.325, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 1.2566539923954372, | |
| "grad_norm": 0.7298043496511923, | |
| "learning_rate": 7.310375318056107e-06, | |
| "loss": 0.3162, | |
| "step": 661 | |
| }, | |
| { | |
| "epoch": 1.258555133079848, | |
| "grad_norm": 0.7033753148471557, | |
| "learning_rate": 7.278404721529843e-06, | |
| "loss": 0.3246, | |
| "step": 662 | |
| }, | |
| { | |
| "epoch": 1.2604562737642586, | |
| "grad_norm": 0.7028455691802423, | |
| "learning_rate": 7.246464140160365e-06, | |
| "loss": 0.3046, | |
| "step": 663 | |
| }, | |
| { | |
| "epoch": 1.2623574144486693, | |
| "grad_norm": 0.7217660122185159, | |
| "learning_rate": 7.214553926204884e-06, | |
| "loss": 0.3065, | |
| "step": 664 | |
| }, | |
| { | |
| "epoch": 1.2642585551330798, | |
| "grad_norm": 0.7020472858032674, | |
| "learning_rate": 7.182674431585703e-06, | |
| "loss": 0.3114, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 1.2661596958174905, | |
| "grad_norm": 0.7319868845907703, | |
| "learning_rate": 7.150826007886334e-06, | |
| "loss": 0.3134, | |
| "step": 666 | |
| }, | |
| { | |
| "epoch": 1.2680608365019013, | |
| "grad_norm": 0.7600082276976634, | |
| "learning_rate": 7.119009006347625e-06, | |
| "loss": 0.292, | |
| "step": 667 | |
| }, | |
| { | |
| "epoch": 1.2699619771863118, | |
| "grad_norm": 0.7672017400731106, | |
| "learning_rate": 7.087223777863883e-06, | |
| "loss": 0.312, | |
| "step": 668 | |
| }, | |
| { | |
| "epoch": 1.2718631178707225, | |
| "grad_norm": 0.7494689315967313, | |
| "learning_rate": 7.055470672979003e-06, | |
| "loss": 0.3301, | |
| "step": 669 | |
| }, | |
| { | |
| "epoch": 1.2737642585551332, | |
| "grad_norm": 0.6634920770565246, | |
| "learning_rate": 7.023750041882609e-06, | |
| "loss": 0.3096, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 1.2756653992395437, | |
| "grad_norm": 0.740699377467611, | |
| "learning_rate": 6.992062234406185e-06, | |
| "loss": 0.3068, | |
| "step": 671 | |
| }, | |
| { | |
| "epoch": 1.2775665399239544, | |
| "grad_norm": 0.6775779975277, | |
| "learning_rate": 6.960407600019217e-06, | |
| "loss": 0.2964, | |
| "step": 672 | |
| }, | |
| { | |
| "epoch": 1.2794676806083651, | |
| "grad_norm": 0.687194787281401, | |
| "learning_rate": 6.9287864878253475e-06, | |
| "loss": 0.3169, | |
| "step": 673 | |
| }, | |
| { | |
| "epoch": 1.2813688212927756, | |
| "grad_norm": 0.7247430890363922, | |
| "learning_rate": 6.897199246558515e-06, | |
| "loss": 0.3064, | |
| "step": 674 | |
| }, | |
| { | |
| "epoch": 1.2832699619771863, | |
| "grad_norm": 0.684875208553655, | |
| "learning_rate": 6.865646224579108e-06, | |
| "loss": 0.3026, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 1.285171102661597, | |
| "grad_norm": 0.7196713309024741, | |
| "learning_rate": 6.834127769870134e-06, | |
| "loss": 0.313, | |
| "step": 676 | |
| }, | |
| { | |
| "epoch": 1.2870722433460076, | |
| "grad_norm": 0.6853306913900206, | |
| "learning_rate": 6.802644230033373e-06, | |
| "loss": 0.3093, | |
| "step": 677 | |
| }, | |
| { | |
| "epoch": 1.2889733840304183, | |
| "grad_norm": 0.7064322834584811, | |
| "learning_rate": 6.771195952285541e-06, | |
| "loss": 0.2928, | |
| "step": 678 | |
| }, | |
| { | |
| "epoch": 1.290874524714829, | |
| "grad_norm": 0.7450418023802602, | |
| "learning_rate": 6.739783283454469e-06, | |
| "loss": 0.3101, | |
| "step": 679 | |
| }, | |
| { | |
| "epoch": 1.2927756653992395, | |
| "grad_norm": 0.6698132476398562, | |
| "learning_rate": 6.708406569975274e-06, | |
| "loss": 0.2922, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 1.2946768060836502, | |
| "grad_norm": 0.6774129090226048, | |
| "learning_rate": 6.6770661578865444e-06, | |
| "loss": 0.2941, | |
| "step": 681 | |
| }, | |
| { | |
| "epoch": 1.296577946768061, | |
| "grad_norm": 0.728747713038511, | |
| "learning_rate": 6.645762392826509e-06, | |
| "loss": 0.3031, | |
| "step": 682 | |
| }, | |
| { | |
| "epoch": 1.2984790874524714, | |
| "grad_norm": 0.7158530380750532, | |
| "learning_rate": 6.614495620029238e-06, | |
| "loss": 0.3121, | |
| "step": 683 | |
| }, | |
| { | |
| "epoch": 1.3003802281368821, | |
| "grad_norm": 0.6876892485372104, | |
| "learning_rate": 6.583266184320836e-06, | |
| "loss": 0.2973, | |
| "step": 684 | |
| }, | |
| { | |
| "epoch": 1.3022813688212929, | |
| "grad_norm": 0.7526740736775194, | |
| "learning_rate": 6.552074430115624e-06, | |
| "loss": 0.3302, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 1.3041825095057034, | |
| "grad_norm": 0.6621942680447704, | |
| "learning_rate": 6.520920701412371e-06, | |
| "loss": 0.3001, | |
| "step": 686 | |
| }, | |
| { | |
| "epoch": 1.306083650190114, | |
| "grad_norm": 0.6940318609946915, | |
| "learning_rate": 6.489805341790456e-06, | |
| "loss": 0.2946, | |
| "step": 687 | |
| }, | |
| { | |
| "epoch": 1.3079847908745248, | |
| "grad_norm": 0.6935505117903904, | |
| "learning_rate": 6.458728694406124e-06, | |
| "loss": 0.3121, | |
| "step": 688 | |
| }, | |
| { | |
| "epoch": 1.3098859315589353, | |
| "grad_norm": 0.7362624171138598, | |
| "learning_rate": 6.427691101988673e-06, | |
| "loss": 0.3129, | |
| "step": 689 | |
| }, | |
| { | |
| "epoch": 1.311787072243346, | |
| "grad_norm": 0.7176633571916394, | |
| "learning_rate": 6.396692906836686e-06, | |
| "loss": 0.3354, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 1.3136882129277567, | |
| "grad_norm": 0.6993371324282728, | |
| "learning_rate": 6.3657344508142495e-06, | |
| "loss": 0.3109, | |
| "step": 691 | |
| }, | |
| { | |
| "epoch": 1.3155893536121672, | |
| "grad_norm": 0.6418582649261754, | |
| "learning_rate": 6.334816075347185e-06, | |
| "loss": 0.2965, | |
| "step": 692 | |
| }, | |
| { | |
| "epoch": 1.317490494296578, | |
| "grad_norm": 0.6894355355476369, | |
| "learning_rate": 6.303938121419295e-06, | |
| "loss": 0.2998, | |
| "step": 693 | |
| }, | |
| { | |
| "epoch": 1.3193916349809887, | |
| "grad_norm": 0.7092080687512982, | |
| "learning_rate": 6.273100929568579e-06, | |
| "loss": 0.3059, | |
| "step": 694 | |
| }, | |
| { | |
| "epoch": 1.3212927756653992, | |
| "grad_norm": 0.7071394520696178, | |
| "learning_rate": 6.242304839883502e-06, | |
| "loss": 0.302, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 1.3231939163498099, | |
| "grad_norm": 0.7357174419049026, | |
| "learning_rate": 6.211550191999223e-06, | |
| "loss": 0.3217, | |
| "step": 696 | |
| }, | |
| { | |
| "epoch": 1.3250950570342206, | |
| "grad_norm": 0.708168698330685, | |
| "learning_rate": 6.18083732509387e-06, | |
| "loss": 0.3338, | |
| "step": 697 | |
| }, | |
| { | |
| "epoch": 1.326996197718631, | |
| "grad_norm": 0.6837935452913564, | |
| "learning_rate": 6.150166577884781e-06, | |
| "loss": 0.3053, | |
| "step": 698 | |
| }, | |
| { | |
| "epoch": 1.3288973384030418, | |
| "grad_norm": 0.7557429731232368, | |
| "learning_rate": 6.119538288624778e-06, | |
| "loss": 0.2978, | |
| "step": 699 | |
| }, | |
| { | |
| "epoch": 1.3307984790874525, | |
| "grad_norm": 0.712034838272552, | |
| "learning_rate": 6.088952795098442e-06, | |
| "loss": 0.3071, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.332699619771863, | |
| "grad_norm": 0.686095792711308, | |
| "learning_rate": 6.058410434618367e-06, | |
| "loss": 0.3051, | |
| "step": 701 | |
| }, | |
| { | |
| "epoch": 1.3346007604562737, | |
| "grad_norm": 0.6716735881892217, | |
| "learning_rate": 6.027911544021465e-06, | |
| "loss": 0.3053, | |
| "step": 702 | |
| }, | |
| { | |
| "epoch": 1.3365019011406845, | |
| "grad_norm": 0.6618116933259053, | |
| "learning_rate": 5.997456459665237e-06, | |
| "loss": 0.3106, | |
| "step": 703 | |
| }, | |
| { | |
| "epoch": 1.338403041825095, | |
| "grad_norm": 0.6836697612756923, | |
| "learning_rate": 5.967045517424062e-06, | |
| "loss": 0.3116, | |
| "step": 704 | |
| }, | |
| { | |
| "epoch": 1.3403041825095057, | |
| "grad_norm": 0.7344335764835925, | |
| "learning_rate": 5.936679052685505e-06, | |
| "loss": 0.3223, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 1.3422053231939164, | |
| "grad_norm": 0.771553492341178, | |
| "learning_rate": 5.906357400346596e-06, | |
| "loss": 0.3247, | |
| "step": 706 | |
| }, | |
| { | |
| "epoch": 1.3441064638783269, | |
| "grad_norm": 0.695301483237602, | |
| "learning_rate": 5.876080894810167e-06, | |
| "loss": 0.3043, | |
| "step": 707 | |
| }, | |
| { | |
| "epoch": 1.3460076045627376, | |
| "grad_norm": 0.6846690888356044, | |
| "learning_rate": 5.845849869981137e-06, | |
| "loss": 0.296, | |
| "step": 708 | |
| }, | |
| { | |
| "epoch": 1.3479087452471483, | |
| "grad_norm": 0.7588005584448377, | |
| "learning_rate": 5.815664659262845e-06, | |
| "loss": 0.3208, | |
| "step": 709 | |
| }, | |
| { | |
| "epoch": 1.3498098859315588, | |
| "grad_norm": 0.7329430579242778, | |
| "learning_rate": 5.78552559555337e-06, | |
| "loss": 0.3105, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 1.3517110266159695, | |
| "grad_norm": 0.6449731705572702, | |
| "learning_rate": 5.755433011241851e-06, | |
| "loss": 0.2976, | |
| "step": 711 | |
| }, | |
| { | |
| "epoch": 1.3536121673003803, | |
| "grad_norm": 0.6773874588342192, | |
| "learning_rate": 5.725387238204831e-06, | |
| "loss": 0.3066, | |
| "step": 712 | |
| }, | |
| { | |
| "epoch": 1.3555133079847907, | |
| "grad_norm": 0.7906417113347532, | |
| "learning_rate": 5.695388607802603e-06, | |
| "loss": 0.3056, | |
| "step": 713 | |
| }, | |
| { | |
| "epoch": 1.3574144486692015, | |
| "grad_norm": 0.6721165121641532, | |
| "learning_rate": 5.665437450875534e-06, | |
| "loss": 0.3128, | |
| "step": 714 | |
| }, | |
| { | |
| "epoch": 1.3593155893536122, | |
| "grad_norm": 0.7052957807476773, | |
| "learning_rate": 5.635534097740435e-06, | |
| "loss": 0.3052, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 1.3612167300380227, | |
| "grad_norm": 0.6934352840931263, | |
| "learning_rate": 5.605678878186911e-06, | |
| "loss": 0.3018, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 1.3631178707224334, | |
| "grad_norm": 0.7081882345309076, | |
| "learning_rate": 5.575872121473722e-06, | |
| "loss": 0.2941, | |
| "step": 717 | |
| }, | |
| { | |
| "epoch": 1.3650190114068441, | |
| "grad_norm": 0.7596719291801347, | |
| "learning_rate": 5.546114156325166e-06, | |
| "loss": 0.3059, | |
| "step": 718 | |
| }, | |
| { | |
| "epoch": 1.3669201520912546, | |
| "grad_norm": 0.7074931260489689, | |
| "learning_rate": 5.516405310927431e-06, | |
| "loss": 0.3071, | |
| "step": 719 | |
| }, | |
| { | |
| "epoch": 1.3688212927756653, | |
| "grad_norm": 0.688182332663401, | |
| "learning_rate": 5.4867459129249846e-06, | |
| "loss": 0.3034, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 1.370722433460076, | |
| "grad_norm": 0.7354037403064927, | |
| "learning_rate": 5.4571362894169795e-06, | |
| "loss": 0.3181, | |
| "step": 721 | |
| }, | |
| { | |
| "epoch": 1.3726235741444868, | |
| "grad_norm": 0.7885001601333944, | |
| "learning_rate": 5.427576766953615e-06, | |
| "loss": 0.3011, | |
| "step": 722 | |
| }, | |
| { | |
| "epoch": 1.3745247148288973, | |
| "grad_norm": 0.6947399978371279, | |
| "learning_rate": 5.398067671532554e-06, | |
| "loss": 0.3045, | |
| "step": 723 | |
| }, | |
| { | |
| "epoch": 1.376425855513308, | |
| "grad_norm": 0.651226361151284, | |
| "learning_rate": 5.368609328595323e-06, | |
| "loss": 0.2954, | |
| "step": 724 | |
| }, | |
| { | |
| "epoch": 1.3783269961977187, | |
| "grad_norm": 0.6589235503500636, | |
| "learning_rate": 5.339202063023727e-06, | |
| "loss": 0.2941, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 1.3802281368821292, | |
| "grad_norm": 0.6676407357838474, | |
| "learning_rate": 5.309846199136258e-06, | |
| "loss": 0.2948, | |
| "step": 726 | |
| }, | |
| { | |
| "epoch": 1.38212927756654, | |
| "grad_norm": 0.703508231084904, | |
| "learning_rate": 5.280542060684535e-06, | |
| "loss": 0.3093, | |
| "step": 727 | |
| }, | |
| { | |
| "epoch": 1.3840304182509506, | |
| "grad_norm": 0.7085809481799236, | |
| "learning_rate": 5.2512899708497086e-06, | |
| "loss": 0.298, | |
| "step": 728 | |
| }, | |
| { | |
| "epoch": 1.3859315589353614, | |
| "grad_norm": 0.6913247334573938, | |
| "learning_rate": 5.222090252238916e-06, | |
| "loss": 0.3061, | |
| "step": 729 | |
| }, | |
| { | |
| "epoch": 1.3878326996197718, | |
| "grad_norm": 0.6994985457609108, | |
| "learning_rate": 5.192943226881724e-06, | |
| "loss": 0.2865, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 1.3897338403041826, | |
| "grad_norm": 0.7551420287756099, | |
| "learning_rate": 5.163849216226562e-06, | |
| "loss": 0.3034, | |
| "step": 731 | |
| }, | |
| { | |
| "epoch": 1.3916349809885933, | |
| "grad_norm": 0.7093111145773565, | |
| "learning_rate": 5.134808541137183e-06, | |
| "loss": 0.3212, | |
| "step": 732 | |
| }, | |
| { | |
| "epoch": 1.3935361216730038, | |
| "grad_norm": 0.7017954163378031, | |
| "learning_rate": 5.105821521889147e-06, | |
| "loss": 0.3002, | |
| "step": 733 | |
| }, | |
| { | |
| "epoch": 1.3954372623574145, | |
| "grad_norm": 0.6844260915059592, | |
| "learning_rate": 5.076888478166247e-06, | |
| "loss": 0.3172, | |
| "step": 734 | |
| }, | |
| { | |
| "epoch": 1.3973384030418252, | |
| "grad_norm": 0.7445025681124675, | |
| "learning_rate": 5.048009729057012e-06, | |
| "loss": 0.2995, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 1.3992395437262357, | |
| "grad_norm": 0.7417707440857404, | |
| "learning_rate": 5.0191855930511946e-06, | |
| "loss": 0.3067, | |
| "step": 736 | |
| }, | |
| { | |
| "epoch": 1.4011406844106464, | |
| "grad_norm": 0.7193583350817124, | |
| "learning_rate": 4.990416388036233e-06, | |
| "loss": 0.2872, | |
| "step": 737 | |
| }, | |
| { | |
| "epoch": 1.4030418250950571, | |
| "grad_norm": 0.6927748525401025, | |
| "learning_rate": 4.961702431293759e-06, | |
| "loss": 0.3093, | |
| "step": 738 | |
| }, | |
| { | |
| "epoch": 1.4049429657794676, | |
| "grad_norm": 0.7023504069249282, | |
| "learning_rate": 4.933044039496107e-06, | |
| "loss": 0.2984, | |
| "step": 739 | |
| }, | |
| { | |
| "epoch": 1.4068441064638784, | |
| "grad_norm": 0.6848494189095844, | |
| "learning_rate": 4.904441528702806e-06, | |
| "loss": 0.2959, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 1.408745247148289, | |
| "grad_norm": 0.7198982235176652, | |
| "learning_rate": 4.875895214357093e-06, | |
| "loss": 0.3036, | |
| "step": 741 | |
| }, | |
| { | |
| "epoch": 1.4106463878326996, | |
| "grad_norm": 0.7350074319257783, | |
| "learning_rate": 4.847405411282462e-06, | |
| "loss": 0.3172, | |
| "step": 742 | |
| }, | |
| { | |
| "epoch": 1.4125475285171103, | |
| "grad_norm": 0.7005431827789544, | |
| "learning_rate": 4.818972433679145e-06, | |
| "loss": 0.3006, | |
| "step": 743 | |
| }, | |
| { | |
| "epoch": 1.414448669201521, | |
| "grad_norm": 0.770541260160109, | |
| "learning_rate": 4.790596595120699e-06, | |
| "loss": 0.2942, | |
| "step": 744 | |
| }, | |
| { | |
| "epoch": 1.4163498098859315, | |
| "grad_norm": 0.659834513137269, | |
| "learning_rate": 4.762278208550505e-06, | |
| "loss": 0.2891, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 1.4182509505703422, | |
| "grad_norm": 0.703917700314252, | |
| "learning_rate": 4.734017586278337e-06, | |
| "loss": 0.2925, | |
| "step": 746 | |
| }, | |
| { | |
| "epoch": 1.420152091254753, | |
| "grad_norm": 0.7194526917686687, | |
| "learning_rate": 4.7058150399769245e-06, | |
| "loss": 0.3161, | |
| "step": 747 | |
| }, | |
| { | |
| "epoch": 1.4220532319391634, | |
| "grad_norm": 0.7057889128392096, | |
| "learning_rate": 4.677670880678493e-06, | |
| "loss": 0.3097, | |
| "step": 748 | |
| }, | |
| { | |
| "epoch": 1.4239543726235742, | |
| "grad_norm": 0.6666131969888315, | |
| "learning_rate": 4.649585418771348e-06, | |
| "loss": 0.2945, | |
| "step": 749 | |
| }, | |
| { | |
| "epoch": 1.4258555133079849, | |
| "grad_norm": 0.7535754883252909, | |
| "learning_rate": 4.621558963996458e-06, | |
| "loss": 0.3081, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.4277566539923954, | |
| "grad_norm": 0.7186051807264096, | |
| "learning_rate": 4.593591825444028e-06, | |
| "loss": 0.2998, | |
| "step": 751 | |
| }, | |
| { | |
| "epoch": 1.429657794676806, | |
| "grad_norm": 0.6975678085811556, | |
| "learning_rate": 4.565684311550077e-06, | |
| "loss": 0.3049, | |
| "step": 752 | |
| }, | |
| { | |
| "epoch": 1.4315589353612168, | |
| "grad_norm": 0.7803727945801265, | |
| "learning_rate": 4.537836730093077e-06, | |
| "loss": 0.3077, | |
| "step": 753 | |
| }, | |
| { | |
| "epoch": 1.4334600760456273, | |
| "grad_norm": 0.6913590260336043, | |
| "learning_rate": 4.510049388190518e-06, | |
| "loss": 0.3067, | |
| "step": 754 | |
| }, | |
| { | |
| "epoch": 1.435361216730038, | |
| "grad_norm": 0.7166680142868859, | |
| "learning_rate": 4.482322592295541e-06, | |
| "loss": 0.2963, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 1.4372623574144487, | |
| "grad_norm": 0.6752490565850662, | |
| "learning_rate": 4.454656648193559e-06, | |
| "loss": 0.2852, | |
| "step": 756 | |
| }, | |
| { | |
| "epoch": 1.4391634980988592, | |
| "grad_norm": 0.683229026032993, | |
| "learning_rate": 4.427051860998877e-06, | |
| "loss": 0.2941, | |
| "step": 757 | |
| }, | |
| { | |
| "epoch": 1.44106463878327, | |
| "grad_norm": 0.6679065265383225, | |
| "learning_rate": 4.399508535151321e-06, | |
| "loss": 0.2872, | |
| "step": 758 | |
| }, | |
| { | |
| "epoch": 1.4429657794676807, | |
| "grad_norm": 0.706472371237432, | |
| "learning_rate": 4.372026974412907e-06, | |
| "loss": 0.294, | |
| "step": 759 | |
| }, | |
| { | |
| "epoch": 1.4448669201520912, | |
| "grad_norm": 0.7786955137050446, | |
| "learning_rate": 4.344607481864466e-06, | |
| "loss": 0.302, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 1.446768060836502, | |
| "grad_norm": 0.7512754489808825, | |
| "learning_rate": 4.317250359902295e-06, | |
| "loss": 0.2992, | |
| "step": 761 | |
| }, | |
| { | |
| "epoch": 1.4486692015209126, | |
| "grad_norm": 0.6790065380764603, | |
| "learning_rate": 4.2899559102348585e-06, | |
| "loss": 0.3102, | |
| "step": 762 | |
| }, | |
| { | |
| "epoch": 1.450570342205323, | |
| "grad_norm": 0.7177361065097103, | |
| "learning_rate": 4.262724433879427e-06, | |
| "loss": 0.3124, | |
| "step": 763 | |
| }, | |
| { | |
| "epoch": 1.4524714828897338, | |
| "grad_norm": 0.7091785289078043, | |
| "learning_rate": 4.235556231158765e-06, | |
| "loss": 0.3055, | |
| "step": 764 | |
| }, | |
| { | |
| "epoch": 1.4543726235741445, | |
| "grad_norm": 0.7164965721700899, | |
| "learning_rate": 4.208451601697836e-06, | |
| "loss": 0.3042, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 1.456273764258555, | |
| "grad_norm": 0.6887475989966378, | |
| "learning_rate": 4.181410844420473e-06, | |
| "loss": 0.3064, | |
| "step": 766 | |
| }, | |
| { | |
| "epoch": 1.4581749049429658, | |
| "grad_norm": 0.6876007651215937, | |
| "learning_rate": 4.154434257546095e-06, | |
| "loss": 0.286, | |
| "step": 767 | |
| }, | |
| { | |
| "epoch": 1.4600760456273765, | |
| "grad_norm": 0.6660396112662806, | |
| "learning_rate": 4.127522138586424e-06, | |
| "loss": 0.3028, | |
| "step": 768 | |
| }, | |
| { | |
| "epoch": 1.461977186311787, | |
| "grad_norm": 0.6631324464491161, | |
| "learning_rate": 4.10067478434219e-06, | |
| "loss": 0.2986, | |
| "step": 769 | |
| }, | |
| { | |
| "epoch": 1.4638783269961977, | |
| "grad_norm": 0.6465749963704949, | |
| "learning_rate": 4.073892490899865e-06, | |
| "loss": 0.2953, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 1.4657794676806084, | |
| "grad_norm": 0.7261318398071719, | |
| "learning_rate": 4.047175553628397e-06, | |
| "loss": 0.299, | |
| "step": 771 | |
| }, | |
| { | |
| "epoch": 1.467680608365019, | |
| "grad_norm": 0.7625311389421772, | |
| "learning_rate": 4.020524267175954e-06, | |
| "loss": 0.298, | |
| "step": 772 | |
| }, | |
| { | |
| "epoch": 1.4695817490494296, | |
| "grad_norm": 0.6741612995404818, | |
| "learning_rate": 3.993938925466674e-06, | |
| "loss": 0.292, | |
| "step": 773 | |
| }, | |
| { | |
| "epoch": 1.4714828897338403, | |
| "grad_norm": 0.7113046240771782, | |
| "learning_rate": 3.96741982169742e-06, | |
| "loss": 0.3082, | |
| "step": 774 | |
| }, | |
| { | |
| "epoch": 1.4733840304182508, | |
| "grad_norm": 0.7382452784569906, | |
| "learning_rate": 3.9409672483345465e-06, | |
| "loss": 0.3083, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 1.4752851711026616, | |
| "grad_norm": 0.7625831922163097, | |
| "learning_rate": 3.914581497110684e-06, | |
| "loss": 0.3028, | |
| "step": 776 | |
| }, | |
| { | |
| "epoch": 1.4771863117870723, | |
| "grad_norm": 0.8256403638359667, | |
| "learning_rate": 3.888262859021508e-06, | |
| "loss": 0.2942, | |
| "step": 777 | |
| }, | |
| { | |
| "epoch": 1.4790874524714828, | |
| "grad_norm": 0.6596773558794012, | |
| "learning_rate": 3.862011624322534e-06, | |
| "loss": 0.2873, | |
| "step": 778 | |
| }, | |
| { | |
| "epoch": 1.4809885931558935, | |
| "grad_norm": 0.7260385820992834, | |
| "learning_rate": 3.835828082525925e-06, | |
| "loss": 0.3173, | |
| "step": 779 | |
| }, | |
| { | |
| "epoch": 1.4828897338403042, | |
| "grad_norm": 0.6863507205031991, | |
| "learning_rate": 3.8097125223972864e-06, | |
| "loss": 0.2961, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 1.4847908745247147, | |
| "grad_norm": 0.7190727707096162, | |
| "learning_rate": 3.7836652319524835e-06, | |
| "loss": 0.2961, | |
| "step": 781 | |
| }, | |
| { | |
| "epoch": 1.4866920152091254, | |
| "grad_norm": 0.7652105306079513, | |
| "learning_rate": 3.7576864984544814e-06, | |
| "loss": 0.3055, | |
| "step": 782 | |
| }, | |
| { | |
| "epoch": 1.4885931558935361, | |
| "grad_norm": 0.7665558483232622, | |
| "learning_rate": 3.73177660841015e-06, | |
| "loss": 0.3242, | |
| "step": 783 | |
| }, | |
| { | |
| "epoch": 1.4904942965779466, | |
| "grad_norm": 0.6849131134755368, | |
| "learning_rate": 3.7059358475671225e-06, | |
| "loss": 0.2926, | |
| "step": 784 | |
| }, | |
| { | |
| "epoch": 1.4923954372623573, | |
| "grad_norm": 0.6447789033483946, | |
| "learning_rate": 3.680164500910646e-06, | |
| "loss": 0.2853, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 1.494296577946768, | |
| "grad_norm": 0.675186457779792, | |
| "learning_rate": 3.654462852660423e-06, | |
| "loss": 0.3096, | |
| "step": 786 | |
| }, | |
| { | |
| "epoch": 1.4961977186311788, | |
| "grad_norm": 0.7309911905501041, | |
| "learning_rate": 3.6288311862674885e-06, | |
| "loss": 0.3136, | |
| "step": 787 | |
| }, | |
| { | |
| "epoch": 1.4980988593155893, | |
| "grad_norm": 0.6953152875402182, | |
| "learning_rate": 3.6032697844110896e-06, | |
| "loss": 0.306, | |
| "step": 788 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "grad_norm": 0.663386594815397, | |
| "learning_rate": 3.5777789289955454e-06, | |
| "loss": 0.3029, | |
| "step": 789 | |
| }, | |
| { | |
| "epoch": 1.5019011406844105, | |
| "grad_norm": 0.6432496405098487, | |
| "learning_rate": 3.5523589011471592e-06, | |
| "loss": 0.2773, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 1.5038022813688214, | |
| "grad_norm": 0.675597695364538, | |
| "learning_rate": 3.527009981211119e-06, | |
| "loss": 0.3062, | |
| "step": 791 | |
| }, | |
| { | |
| "epoch": 1.505703422053232, | |
| "grad_norm": 0.6586242520127019, | |
| "learning_rate": 3.5017324487483873e-06, | |
| "loss": 0.2917, | |
| "step": 792 | |
| }, | |
| { | |
| "epoch": 1.5076045627376424, | |
| "grad_norm": 0.6930272300017185, | |
| "learning_rate": 3.47652658253263e-06, | |
| "loss": 0.2946, | |
| "step": 793 | |
| }, | |
| { | |
| "epoch": 1.5095057034220534, | |
| "grad_norm": 0.6967189381203526, | |
| "learning_rate": 3.4513926605471504e-06, | |
| "loss": 0.283, | |
| "step": 794 | |
| }, | |
| { | |
| "epoch": 1.5114068441064639, | |
| "grad_norm": 0.6662009192639498, | |
| "learning_rate": 3.4263309599818017e-06, | |
| "loss": 0.2986, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 1.5133079847908744, | |
| "grad_norm": 0.6863193933185225, | |
| "learning_rate": 3.4013417572299446e-06, | |
| "loss": 0.3049, | |
| "step": 796 | |
| }, | |
| { | |
| "epoch": 1.5152091254752853, | |
| "grad_norm": 0.6669866976205363, | |
| "learning_rate": 3.37642532788541e-06, | |
| "loss": 0.2871, | |
| "step": 797 | |
| }, | |
| { | |
| "epoch": 1.5171102661596958, | |
| "grad_norm": 0.6835919557975013, | |
| "learning_rate": 3.3515819467394184e-06, | |
| "loss": 0.2984, | |
| "step": 798 | |
| }, | |
| { | |
| "epoch": 1.5190114068441065, | |
| "grad_norm": 0.7018632230531058, | |
| "learning_rate": 3.326811887777607e-06, | |
| "loss": 0.302, | |
| "step": 799 | |
| }, | |
| { | |
| "epoch": 1.5209125475285172, | |
| "grad_norm": 0.7205430829164591, | |
| "learning_rate": 3.3021154241769606e-06, | |
| "loss": 0.3022, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.5228136882129277, | |
| "grad_norm": 0.7171970344846872, | |
| "learning_rate": 3.2774928283028153e-06, | |
| "loss": 0.3116, | |
| "step": 801 | |
| }, | |
| { | |
| "epoch": 1.5247148288973384, | |
| "grad_norm": 0.7530171144555614, | |
| "learning_rate": 3.2529443717058693e-06, | |
| "loss": 0.3132, | |
| "step": 802 | |
| }, | |
| { | |
| "epoch": 1.5266159695817492, | |
| "grad_norm": 0.6801930461898661, | |
| "learning_rate": 3.228470325119164e-06, | |
| "loss": 0.2938, | |
| "step": 803 | |
| }, | |
| { | |
| "epoch": 1.5285171102661597, | |
| "grad_norm": 0.6883611804852857, | |
| "learning_rate": 3.20407095845511e-06, | |
| "loss": 0.2929, | |
| "step": 804 | |
| }, | |
| { | |
| "epoch": 1.5304182509505704, | |
| "grad_norm": 0.7542153963625797, | |
| "learning_rate": 3.179746540802506e-06, | |
| "loss": 0.3144, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 1.532319391634981, | |
| "grad_norm": 0.7091137647828326, | |
| "learning_rate": 3.155497340423588e-06, | |
| "loss": 0.3061, | |
| "step": 806 | |
| }, | |
| { | |
| "epoch": 1.5342205323193916, | |
| "grad_norm": 0.6831645146853041, | |
| "learning_rate": 3.1313236247510414e-06, | |
| "loss": 0.2871, | |
| "step": 807 | |
| }, | |
| { | |
| "epoch": 1.5361216730038023, | |
| "grad_norm": 0.747143197700626, | |
| "learning_rate": 3.107225660385077e-06, | |
| "loss": 0.3286, | |
| "step": 808 | |
| }, | |
| { | |
| "epoch": 1.538022813688213, | |
| "grad_norm": 0.6934542801698926, | |
| "learning_rate": 3.0832037130904748e-06, | |
| "loss": 0.3079, | |
| "step": 809 | |
| }, | |
| { | |
| "epoch": 1.5399239543726235, | |
| "grad_norm": 0.6825121918745702, | |
| "learning_rate": 3.0592580477936606e-06, | |
| "loss": 0.2903, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 1.5418250950570342, | |
| "grad_norm": 0.7117260109094474, | |
| "learning_rate": 3.035388928579792e-06, | |
| "loss": 0.312, | |
| "step": 811 | |
| }, | |
| { | |
| "epoch": 1.543726235741445, | |
| "grad_norm": 0.6859508712424827, | |
| "learning_rate": 3.011596618689825e-06, | |
| "loss": 0.2977, | |
| "step": 812 | |
| }, | |
| { | |
| "epoch": 1.5456273764258555, | |
| "grad_norm": 0.635407785791035, | |
| "learning_rate": 2.9878813805176252e-06, | |
| "loss": 0.2869, | |
| "step": 813 | |
| }, | |
| { | |
| "epoch": 1.5475285171102662, | |
| "grad_norm": 0.6883632122716664, | |
| "learning_rate": 2.9642434756070793e-06, | |
| "loss": 0.305, | |
| "step": 814 | |
| }, | |
| { | |
| "epoch": 1.549429657794677, | |
| "grad_norm": 0.6770187205796798, | |
| "learning_rate": 2.940683164649194e-06, | |
| "loss": 0.2906, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 1.5513307984790874, | |
| "grad_norm": 0.6627387313495365, | |
| "learning_rate": 2.9172007074792342e-06, | |
| "loss": 0.288, | |
| "step": 816 | |
| }, | |
| { | |
| "epoch": 1.553231939163498, | |
| "grad_norm": 0.6476134019641687, | |
| "learning_rate": 2.8937963630738517e-06, | |
| "loss": 0.3069, | |
| "step": 817 | |
| }, | |
| { | |
| "epoch": 1.5551330798479088, | |
| "grad_norm": 0.7035478620892421, | |
| "learning_rate": 2.87047038954823e-06, | |
| "loss": 0.3126, | |
| "step": 818 | |
| }, | |
| { | |
| "epoch": 1.5570342205323193, | |
| "grad_norm": 0.6688520113222646, | |
| "learning_rate": 2.8472230441532365e-06, | |
| "loss": 0.3026, | |
| "step": 819 | |
| }, | |
| { | |
| "epoch": 1.55893536121673, | |
| "grad_norm": 0.6731317947602746, | |
| "learning_rate": 2.8240545832725963e-06, | |
| "loss": 0.2955, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 1.5608365019011408, | |
| "grad_norm": 0.6664693897768919, | |
| "learning_rate": 2.8009652624200436e-06, | |
| "loss": 0.3078, | |
| "step": 821 | |
| }, | |
| { | |
| "epoch": 1.5627376425855513, | |
| "grad_norm": 0.6599028937133654, | |
| "learning_rate": 2.7779553362365184e-06, | |
| "loss": 0.2853, | |
| "step": 822 | |
| }, | |
| { | |
| "epoch": 1.564638783269962, | |
| "grad_norm": 0.726953859115326, | |
| "learning_rate": 2.755025058487364e-06, | |
| "loss": 0.3026, | |
| "step": 823 | |
| }, | |
| { | |
| "epoch": 1.5665399239543727, | |
| "grad_norm": 0.6933368084164538, | |
| "learning_rate": 2.7321746820595084e-06, | |
| "loss": 0.2991, | |
| "step": 824 | |
| }, | |
| { | |
| "epoch": 1.5684410646387832, | |
| "grad_norm": 0.717282442640616, | |
| "learning_rate": 2.709404458958693e-06, | |
| "loss": 0.3087, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 1.570342205323194, | |
| "grad_norm": 0.6826538110575737, | |
| "learning_rate": 2.6867146403066833e-06, | |
| "loss": 0.2875, | |
| "step": 826 | |
| }, | |
| { | |
| "epoch": 1.5722433460076046, | |
| "grad_norm": 0.6389443428255136, | |
| "learning_rate": 2.6641054763385044e-06, | |
| "loss": 0.2832, | |
| "step": 827 | |
| }, | |
| { | |
| "epoch": 1.5741444866920151, | |
| "grad_norm": 0.7416146908351132, | |
| "learning_rate": 2.6415772163996845e-06, | |
| "loss": 0.2941, | |
| "step": 828 | |
| }, | |
| { | |
| "epoch": 1.5760456273764258, | |
| "grad_norm": 0.6756284618231428, | |
| "learning_rate": 2.619130108943494e-06, | |
| "loss": 0.291, | |
| "step": 829 | |
| }, | |
| { | |
| "epoch": 1.5779467680608366, | |
| "grad_norm": 0.6617101405637681, | |
| "learning_rate": 2.5967644015282146e-06, | |
| "loss": 0.2815, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 1.579847908745247, | |
| "grad_norm": 0.66166194822998, | |
| "learning_rate": 2.5744803408144026e-06, | |
| "loss": 0.2953, | |
| "step": 831 | |
| }, | |
| { | |
| "epoch": 1.5817490494296578, | |
| "grad_norm": 0.7226683791757725, | |
| "learning_rate": 2.5522781725621814e-06, | |
| "loss": 0.3183, | |
| "step": 832 | |
| }, | |
| { | |
| "epoch": 1.5836501901140685, | |
| "grad_norm": 0.6749525731412888, | |
| "learning_rate": 2.530158141628515e-06, | |
| "loss": 0.2968, | |
| "step": 833 | |
| }, | |
| { | |
| "epoch": 1.585551330798479, | |
| "grad_norm": 0.6608359467019951, | |
| "learning_rate": 2.508120491964512e-06, | |
| "loss": 0.2957, | |
| "step": 834 | |
| }, | |
| { | |
| "epoch": 1.5874524714828897, | |
| "grad_norm": 0.707021290365381, | |
| "learning_rate": 2.486165466612751e-06, | |
| "loss": 0.3063, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 1.5893536121673004, | |
| "grad_norm": 0.7152618474558173, | |
| "learning_rate": 2.464293307704566e-06, | |
| "loss": 0.3032, | |
| "step": 836 | |
| }, | |
| { | |
| "epoch": 1.591254752851711, | |
| "grad_norm": 0.7043458746173418, | |
| "learning_rate": 2.4425042564574186e-06, | |
| "loss": 0.2913, | |
| "step": 837 | |
| }, | |
| { | |
| "epoch": 1.5931558935361216, | |
| "grad_norm": 0.6489619913607211, | |
| "learning_rate": 2.4207985531722034e-06, | |
| "loss": 0.2845, | |
| "step": 838 | |
| }, | |
| { | |
| "epoch": 1.5950570342205324, | |
| "grad_norm": 0.7195150200892559, | |
| "learning_rate": 2.3991764372306113e-06, | |
| "loss": 0.3027, | |
| "step": 839 | |
| }, | |
| { | |
| "epoch": 1.5969581749049429, | |
| "grad_norm": 0.6470582523713176, | |
| "learning_rate": 2.377638147092497e-06, | |
| "loss": 0.2678, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 1.5988593155893536, | |
| "grad_norm": 0.6532397866127092, | |
| "learning_rate": 2.3561839202932344e-06, | |
| "loss": 0.2981, | |
| "step": 841 | |
| }, | |
| { | |
| "epoch": 1.6007604562737643, | |
| "grad_norm": 0.6622491691913368, | |
| "learning_rate": 2.3348139934411008e-06, | |
| "loss": 0.2906, | |
| "step": 842 | |
| }, | |
| { | |
| "epoch": 1.6026615969581748, | |
| "grad_norm": 0.6344689144354533, | |
| "learning_rate": 2.3135286022146785e-06, | |
| "loss": 0.2881, | |
| "step": 843 | |
| }, | |
| { | |
| "epoch": 1.6045627376425855, | |
| "grad_norm": 0.6471652438990747, | |
| "learning_rate": 2.292327981360245e-06, | |
| "loss": 0.2886, | |
| "step": 844 | |
| }, | |
| { | |
| "epoch": 1.6064638783269962, | |
| "grad_norm": 0.681336320299053, | |
| "learning_rate": 2.271212364689176e-06, | |
| "loss": 0.2892, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 1.6083650190114067, | |
| "grad_norm": 0.7114793066735262, | |
| "learning_rate": 2.2501819850753925e-06, | |
| "loss": 0.3085, | |
| "step": 846 | |
| }, | |
| { | |
| "epoch": 1.6102661596958177, | |
| "grad_norm": 0.701858677288113, | |
| "learning_rate": 2.229237074452768e-06, | |
| "loss": 0.2918, | |
| "step": 847 | |
| }, | |
| { | |
| "epoch": 1.6121673003802282, | |
| "grad_norm": 0.6434762105419713, | |
| "learning_rate": 2.2083778638125796e-06, | |
| "loss": 0.2889, | |
| "step": 848 | |
| }, | |
| { | |
| "epoch": 1.6140684410646386, | |
| "grad_norm": 0.6665688385283697, | |
| "learning_rate": 2.1876045832009694e-06, | |
| "loss": 0.2749, | |
| "step": 849 | |
| }, | |
| { | |
| "epoch": 1.6159695817490496, | |
| "grad_norm": 0.6803053919733456, | |
| "learning_rate": 2.16691746171639e-06, | |
| "loss": 0.2957, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.61787072243346, | |
| "grad_norm": 0.6895302895442956, | |
| "learning_rate": 2.1463167275070863e-06, | |
| "loss": 0.3016, | |
| "step": 851 | |
| }, | |
| { | |
| "epoch": 1.6197718631178706, | |
| "grad_norm": 0.6814993088373362, | |
| "learning_rate": 2.125802607768588e-06, | |
| "loss": 0.2932, | |
| "step": 852 | |
| }, | |
| { | |
| "epoch": 1.6216730038022815, | |
| "grad_norm": 0.7824851453741958, | |
| "learning_rate": 2.1053753287411895e-06, | |
| "loss": 0.2803, | |
| "step": 853 | |
| }, | |
| { | |
| "epoch": 1.623574144486692, | |
| "grad_norm": 0.6850340416405972, | |
| "learning_rate": 2.08503511570746e-06, | |
| "loss": 0.2879, | |
| "step": 854 | |
| }, | |
| { | |
| "epoch": 1.6254752851711025, | |
| "grad_norm": 0.6643169981532601, | |
| "learning_rate": 2.064782192989765e-06, | |
| "loss": 0.2826, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 1.6273764258555135, | |
| "grad_norm": 0.7008864652617552, | |
| "learning_rate": 2.0446167839477815e-06, | |
| "loss": 0.2822, | |
| "step": 856 | |
| }, | |
| { | |
| "epoch": 1.629277566539924, | |
| "grad_norm": 0.6810997848187094, | |
| "learning_rate": 2.0245391109760437e-06, | |
| "loss": 0.303, | |
| "step": 857 | |
| }, | |
| { | |
| "epoch": 1.6311787072243344, | |
| "grad_norm": 0.6910108194618777, | |
| "learning_rate": 2.0045493955014915e-06, | |
| "loss": 0.299, | |
| "step": 858 | |
| }, | |
| { | |
| "epoch": 1.6330798479087454, | |
| "grad_norm": 0.679974680089649, | |
| "learning_rate": 1.984647857981017e-06, | |
| "loss": 0.3043, | |
| "step": 859 | |
| }, | |
| { | |
| "epoch": 1.6349809885931559, | |
| "grad_norm": 0.6974310038715719, | |
| "learning_rate": 1.96483471789904e-06, | |
| "loss": 0.3123, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 1.6368821292775664, | |
| "grad_norm": 0.6827283725749244, | |
| "learning_rate": 1.9451101937650963e-06, | |
| "loss": 0.3029, | |
| "step": 861 | |
| }, | |
| { | |
| "epoch": 1.6387832699619773, | |
| "grad_norm": 0.6576433762731538, | |
| "learning_rate": 1.925474503111412e-06, | |
| "loss": 0.2978, | |
| "step": 862 | |
| }, | |
| { | |
| "epoch": 1.6406844106463878, | |
| "grad_norm": 0.6853732717205893, | |
| "learning_rate": 1.905927862490512e-06, | |
| "loss": 0.3025, | |
| "step": 863 | |
| }, | |
| { | |
| "epoch": 1.6425855513307985, | |
| "grad_norm": 0.6945472567838478, | |
| "learning_rate": 1.8864704874728346e-06, | |
| "loss": 0.2962, | |
| "step": 864 | |
| }, | |
| { | |
| "epoch": 1.6444866920152093, | |
| "grad_norm": 0.7166654838113549, | |
| "learning_rate": 1.8671025926443464e-06, | |
| "loss": 0.3077, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 1.6463878326996197, | |
| "grad_norm": 0.6733428457803738, | |
| "learning_rate": 1.8478243916041882e-06, | |
| "loss": 0.2846, | |
| "step": 866 | |
| }, | |
| { | |
| "epoch": 1.6482889733840305, | |
| "grad_norm": 0.6838851349940792, | |
| "learning_rate": 1.828636096962304e-06, | |
| "loss": 0.2895, | |
| "step": 867 | |
| }, | |
| { | |
| "epoch": 1.6501901140684412, | |
| "grad_norm": 0.6747852916863192, | |
| "learning_rate": 1.8095379203371044e-06, | |
| "loss": 0.2928, | |
| "step": 868 | |
| }, | |
| { | |
| "epoch": 1.6520912547528517, | |
| "grad_norm": 0.649165740833038, | |
| "learning_rate": 1.7905300723531393e-06, | |
| "loss": 0.2818, | |
| "step": 869 | |
| }, | |
| { | |
| "epoch": 1.6539923954372624, | |
| "grad_norm": 0.6895770386740175, | |
| "learning_rate": 1.771612762638758e-06, | |
| "loss": 0.2992, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 1.6558935361216731, | |
| "grad_norm": 0.67890361284393, | |
| "learning_rate": 1.7527861998238094e-06, | |
| "loss": 0.2966, | |
| "step": 871 | |
| }, | |
| { | |
| "epoch": 1.6577946768060836, | |
| "grad_norm": 0.6649262528073583, | |
| "learning_rate": 1.7340505915373495e-06, | |
| "loss": 0.2872, | |
| "step": 872 | |
| }, | |
| { | |
| "epoch": 1.6596958174904943, | |
| "grad_norm": 0.6789944816325492, | |
| "learning_rate": 1.7154061444053239e-06, | |
| "loss": 0.305, | |
| "step": 873 | |
| }, | |
| { | |
| "epoch": 1.661596958174905, | |
| "grad_norm": 0.6599976811265336, | |
| "learning_rate": 1.6968530640483126e-06, | |
| "loss": 0.2833, | |
| "step": 874 | |
| }, | |
| { | |
| "epoch": 1.6634980988593155, | |
| "grad_norm": 0.6982608392702473, | |
| "learning_rate": 1.6783915550792652e-06, | |
| "loss": 0.2944, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 1.6653992395437263, | |
| "grad_norm": 0.6953690665691588, | |
| "learning_rate": 1.660021821101222e-06, | |
| "loss": 0.2831, | |
| "step": 876 | |
| }, | |
| { | |
| "epoch": 1.667300380228137, | |
| "grad_norm": 0.6509936323455957, | |
| "learning_rate": 1.6417440647050853e-06, | |
| "loss": 0.2813, | |
| "step": 877 | |
| }, | |
| { | |
| "epoch": 1.6692015209125475, | |
| "grad_norm": 0.6789468371864271, | |
| "learning_rate": 1.6235584874673848e-06, | |
| "loss": 0.2986, | |
| "step": 878 | |
| }, | |
| { | |
| "epoch": 1.6711026615969582, | |
| "grad_norm": 0.663091888832902, | |
| "learning_rate": 1.6054652899480472e-06, | |
| "loss": 0.2729, | |
| "step": 879 | |
| }, | |
| { | |
| "epoch": 1.673003802281369, | |
| "grad_norm": 0.6337057214223821, | |
| "learning_rate": 1.587464671688187e-06, | |
| "loss": 0.2901, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 1.6749049429657794, | |
| "grad_norm": 0.6648400906085427, | |
| "learning_rate": 1.5695568312079156e-06, | |
| "loss": 0.2893, | |
| "step": 881 | |
| }, | |
| { | |
| "epoch": 1.6768060836501901, | |
| "grad_norm": 0.6734114869569092, | |
| "learning_rate": 1.5517419660041277e-06, | |
| "loss": 0.2911, | |
| "step": 882 | |
| }, | |
| { | |
| "epoch": 1.6787072243346008, | |
| "grad_norm": 0.6806449497817229, | |
| "learning_rate": 1.534020272548349e-06, | |
| "loss": 0.2989, | |
| "step": 883 | |
| }, | |
| { | |
| "epoch": 1.6806083650190113, | |
| "grad_norm": 0.7121904502623275, | |
| "learning_rate": 1.5163919462845622e-06, | |
| "loss": 0.2949, | |
| "step": 884 | |
| }, | |
| { | |
| "epoch": 1.682509505703422, | |
| "grad_norm": 0.6913874919963577, | |
| "learning_rate": 1.4988571816270402e-06, | |
| "loss": 0.3182, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 1.6844106463878328, | |
| "grad_norm": 0.7165179072894285, | |
| "learning_rate": 1.4814161719582132e-06, | |
| "loss": 0.2867, | |
| "step": 886 | |
| }, | |
| { | |
| "epoch": 1.6863117870722433, | |
| "grad_norm": 0.6838760414779714, | |
| "learning_rate": 1.4640691096265358e-06, | |
| "loss": 0.3056, | |
| "step": 887 | |
| }, | |
| { | |
| "epoch": 1.688212927756654, | |
| "grad_norm": 0.6912528651637655, | |
| "learning_rate": 1.4468161859443609e-06, | |
| "loss": 0.2962, | |
| "step": 888 | |
| }, | |
| { | |
| "epoch": 1.6901140684410647, | |
| "grad_norm": 0.6596738251646834, | |
| "learning_rate": 1.4296575911858268e-06, | |
| "loss": 0.2854, | |
| "step": 889 | |
| }, | |
| { | |
| "epoch": 1.6920152091254752, | |
| "grad_norm": 0.6861760086062655, | |
| "learning_rate": 1.412593514584777e-06, | |
| "loss": 0.298, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 1.693916349809886, | |
| "grad_norm": 0.7039384198173466, | |
| "learning_rate": 1.3956241443326423e-06, | |
| "loss": 0.301, | |
| "step": 891 | |
| }, | |
| { | |
| "epoch": 1.6958174904942966, | |
| "grad_norm": 0.6724831183019471, | |
| "learning_rate": 1.378749667576399e-06, | |
| "loss": 0.3062, | |
| "step": 892 | |
| }, | |
| { | |
| "epoch": 1.6977186311787071, | |
| "grad_norm": 0.6959857261708096, | |
| "learning_rate": 1.3619702704164783e-06, | |
| "loss": 0.3063, | |
| "step": 893 | |
| }, | |
| { | |
| "epoch": 1.6996197718631179, | |
| "grad_norm": 0.7055107695271625, | |
| "learning_rate": 1.3452861379047289e-06, | |
| "loss": 0.3084, | |
| "step": 894 | |
| }, | |
| { | |
| "epoch": 1.7015209125475286, | |
| "grad_norm": 0.6911522384628693, | |
| "learning_rate": 1.3286974540423747e-06, | |
| "loss": 0.2978, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 1.703422053231939, | |
| "grad_norm": 0.6962048809066319, | |
| "learning_rate": 1.3122044017779768e-06, | |
| "loss": 0.3026, | |
| "step": 896 | |
| }, | |
| { | |
| "epoch": 1.7053231939163498, | |
| "grad_norm": 0.6567439524164248, | |
| "learning_rate": 1.2958071630054214e-06, | |
| "loss": 0.2898, | |
| "step": 897 | |
| }, | |
| { | |
| "epoch": 1.7072243346007605, | |
| "grad_norm": 0.6772372442388603, | |
| "learning_rate": 1.279505918561923e-06, | |
| "loss": 0.3073, | |
| "step": 898 | |
| }, | |
| { | |
| "epoch": 1.709125475285171, | |
| "grad_norm": 0.6952933070936825, | |
| "learning_rate": 1.2633008482260146e-06, | |
| "loss": 0.3063, | |
| "step": 899 | |
| }, | |
| { | |
| "epoch": 1.7110266159695817, | |
| "grad_norm": 0.6669362678136389, | |
| "learning_rate": 1.2471921307155655e-06, | |
| "loss": 0.2911, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.7129277566539924, | |
| "grad_norm": 0.6835505621873457, | |
| "learning_rate": 1.2311799436858275e-06, | |
| "loss": 0.2984, | |
| "step": 901 | |
| }, | |
| { | |
| "epoch": 1.714828897338403, | |
| "grad_norm": 0.6595337518096386, | |
| "learning_rate": 1.2152644637274603e-06, | |
| "loss": 0.2903, | |
| "step": 902 | |
| }, | |
| { | |
| "epoch": 1.7167300380228137, | |
| "grad_norm": 0.6876836084105353, | |
| "learning_rate": 1.1994458663645836e-06, | |
| "loss": 0.3065, | |
| "step": 903 | |
| }, | |
| { | |
| "epoch": 1.7186311787072244, | |
| "grad_norm": 0.6797645923144487, | |
| "learning_rate": 1.1837243260528542e-06, | |
| "loss": 0.3026, | |
| "step": 904 | |
| }, | |
| { | |
| "epoch": 1.7205323193916349, | |
| "grad_norm": 0.6863156817495268, | |
| "learning_rate": 1.168100016177528e-06, | |
| "loss": 0.3022, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 1.7224334600760456, | |
| "grad_norm": 0.7032701879950931, | |
| "learning_rate": 1.1525731090515536e-06, | |
| "loss": 0.2779, | |
| "step": 906 | |
| }, | |
| { | |
| "epoch": 1.7243346007604563, | |
| "grad_norm": 0.6604634486067874, | |
| "learning_rate": 1.137143775913675e-06, | |
| "loss": 0.2964, | |
| "step": 907 | |
| }, | |
| { | |
| "epoch": 1.7262357414448668, | |
| "grad_norm": 0.6472753935963478, | |
| "learning_rate": 1.1218121869265365e-06, | |
| "loss": 0.2958, | |
| "step": 908 | |
| }, | |
| { | |
| "epoch": 1.7281368821292775, | |
| "grad_norm": 0.6115362907901039, | |
| "learning_rate": 1.1065785111748117e-06, | |
| "loss": 0.2708, | |
| "step": 909 | |
| }, | |
| { | |
| "epoch": 1.7300380228136882, | |
| "grad_norm": 0.6823239723838781, | |
| "learning_rate": 1.0914429166633355e-06, | |
| "loss": 0.2823, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 1.7319391634980987, | |
| "grad_norm": 0.671673690011934, | |
| "learning_rate": 1.076405570315252e-06, | |
| "loss": 0.3043, | |
| "step": 911 | |
| }, | |
| { | |
| "epoch": 1.7338403041825095, | |
| "grad_norm": 0.7070551045314329, | |
| "learning_rate": 1.0614666379701732e-06, | |
| "loss": 0.2916, | |
| "step": 912 | |
| }, | |
| { | |
| "epoch": 1.7357414448669202, | |
| "grad_norm": 0.6751178908280784, | |
| "learning_rate": 1.046626284382356e-06, | |
| "loss": 0.2917, | |
| "step": 913 | |
| }, | |
| { | |
| "epoch": 1.7376425855513307, | |
| "grad_norm": 0.6821873546266192, | |
| "learning_rate": 1.0318846732188737e-06, | |
| "loss": 0.2942, | |
| "step": 914 | |
| }, | |
| { | |
| "epoch": 1.7395437262357416, | |
| "grad_norm": 0.6836323014990807, | |
| "learning_rate": 1.017241967057816e-06, | |
| "loss": 0.3098, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 1.741444866920152, | |
| "grad_norm": 0.6528173303653002, | |
| "learning_rate": 1.0026983273865055e-06, | |
| "loss": 0.2977, | |
| "step": 916 | |
| }, | |
| { | |
| "epoch": 1.7433460076045626, | |
| "grad_norm": 0.6351129742940393, | |
| "learning_rate": 9.882539145997027e-07, | |
| "loss": 0.29, | |
| "step": 917 | |
| }, | |
| { | |
| "epoch": 1.7452471482889735, | |
| "grad_norm": 0.6813502457479164, | |
| "learning_rate": 9.739088879978409e-07, | |
| "loss": 0.2984, | |
| "step": 918 | |
| }, | |
| { | |
| "epoch": 1.747148288973384, | |
| "grad_norm": 0.656381646801662, | |
| "learning_rate": 9.59663405785277e-07, | |
| "loss": 0.2848, | |
| "step": 919 | |
| }, | |
| { | |
| "epoch": 1.7490494296577945, | |
| "grad_norm": 0.705232923677699, | |
| "learning_rate": 9.455176250685338e-07, | |
| "loss": 0.3058, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 1.7509505703422055, | |
| "grad_norm": 0.6987330259250887, | |
| "learning_rate": 9.314717018545838e-07, | |
| "loss": 0.297, | |
| "step": 921 | |
| }, | |
| { | |
| "epoch": 1.752851711026616, | |
| "grad_norm": 0.7036699150254239, | |
| "learning_rate": 9.17525791049112e-07, | |
| "loss": 0.2822, | |
| "step": 922 | |
| }, | |
| { | |
| "epoch": 1.7547528517110265, | |
| "grad_norm": 0.6787545519227752, | |
| "learning_rate": 9.036800464548157e-07, | |
| "loss": 0.2849, | |
| "step": 923 | |
| }, | |
| { | |
| "epoch": 1.7566539923954374, | |
| "grad_norm": 0.7340555764977803, | |
| "learning_rate": 8.899346207697135e-07, | |
| "loss": 0.2934, | |
| "step": 924 | |
| }, | |
| { | |
| "epoch": 1.758555133079848, | |
| "grad_norm": 0.7081000888864563, | |
| "learning_rate": 8.762896655854481e-07, | |
| "loss": 0.3148, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 1.7604562737642584, | |
| "grad_norm": 0.6950336787861271, | |
| "learning_rate": 8.627453313856249e-07, | |
| "loss": 0.2939, | |
| "step": 926 | |
| }, | |
| { | |
| "epoch": 1.7623574144486693, | |
| "grad_norm": 0.6949569031992134, | |
| "learning_rate": 8.493017675441495e-07, | |
| "loss": 0.2951, | |
| "step": 927 | |
| }, | |
| { | |
| "epoch": 1.7642585551330798, | |
| "grad_norm": 0.7211441392962942, | |
| "learning_rate": 8.359591223235785e-07, | |
| "loss": 0.304, | |
| "step": 928 | |
| }, | |
| { | |
| "epoch": 1.7661596958174905, | |
| "grad_norm": 0.7366783298045159, | |
| "learning_rate": 8.227175428734868e-07, | |
| "loss": 0.308, | |
| "step": 929 | |
| }, | |
| { | |
| "epoch": 1.7680608365019013, | |
| "grad_norm": 0.6915878915318591, | |
| "learning_rate": 8.095771752288451e-07, | |
| "loss": 0.2904, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 1.7699619771863118, | |
| "grad_norm": 0.6646310464781173, | |
| "learning_rate": 7.965381643084069e-07, | |
| "loss": 0.2773, | |
| "step": 931 | |
| }, | |
| { | |
| "epoch": 1.7718631178707225, | |
| "grad_norm": 0.6712959120150253, | |
| "learning_rate": 7.83600653913108e-07, | |
| "loss": 0.2973, | |
| "step": 932 | |
| }, | |
| { | |
| "epoch": 1.7737642585551332, | |
| "grad_norm": 0.642884332855867, | |
| "learning_rate": 7.707647867244927e-07, | |
| "loss": 0.2774, | |
| "step": 933 | |
| }, | |
| { | |
| "epoch": 1.7756653992395437, | |
| "grad_norm": 0.7209788817545872, | |
| "learning_rate": 7.580307043031232e-07, | |
| "loss": 0.3037, | |
| "step": 934 | |
| }, | |
| { | |
| "epoch": 1.7775665399239544, | |
| "grad_norm": 0.7016736614323559, | |
| "learning_rate": 7.453985470870284e-07, | |
| "loss": 0.2791, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 1.7794676806083651, | |
| "grad_norm": 0.6611102717038619, | |
| "learning_rate": 7.328684543901598e-07, | |
| "loss": 0.2992, | |
| "step": 936 | |
| }, | |
| { | |
| "epoch": 1.7813688212927756, | |
| "grad_norm": 0.6732238735705044, | |
| "learning_rate": 7.204405644008416e-07, | |
| "loss": 0.2832, | |
| "step": 937 | |
| }, | |
| { | |
| "epoch": 1.7832699619771863, | |
| "grad_norm": 0.6737326291645108, | |
| "learning_rate": 7.081150141802518e-07, | |
| "loss": 0.2923, | |
| "step": 938 | |
| }, | |
| { | |
| "epoch": 1.785171102661597, | |
| "grad_norm": 0.6591795416292429, | |
| "learning_rate": 6.958919396609231e-07, | |
| "loss": 0.3007, | |
| "step": 939 | |
| }, | |
| { | |
| "epoch": 1.7870722433460076, | |
| "grad_norm": 0.6719629866280276, | |
| "learning_rate": 6.837714756452241e-07, | |
| "loss": 0.2802, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 1.7889733840304183, | |
| "grad_norm": 0.6639356894636503, | |
| "learning_rate": 6.717537558038845e-07, | |
| "loss": 0.2995, | |
| "step": 941 | |
| }, | |
| { | |
| "epoch": 1.790874524714829, | |
| "grad_norm": 0.6594793805503537, | |
| "learning_rate": 6.598389126745209e-07, | |
| "loss": 0.3106, | |
| "step": 942 | |
| }, | |
| { | |
| "epoch": 1.7927756653992395, | |
| "grad_norm": 0.6754365187795434, | |
| "learning_rate": 6.480270776601682e-07, | |
| "loss": 0.2897, | |
| "step": 943 | |
| }, | |
| { | |
| "epoch": 1.7946768060836502, | |
| "grad_norm": 0.665952018898404, | |
| "learning_rate": 6.36318381027835e-07, | |
| "loss": 0.2852, | |
| "step": 944 | |
| }, | |
| { | |
| "epoch": 1.796577946768061, | |
| "grad_norm": 0.6428017078555177, | |
| "learning_rate": 6.247129519070728e-07, | |
| "loss": 0.2803, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 1.7984790874524714, | |
| "grad_norm": 0.6652873068295456, | |
| "learning_rate": 6.132109182885382e-07, | |
| "loss": 0.2907, | |
| "step": 946 | |
| }, | |
| { | |
| "epoch": 1.8003802281368821, | |
| "grad_norm": 0.6808633269926114, | |
| "learning_rate": 6.018124070225928e-07, | |
| "loss": 0.2834, | |
| "step": 947 | |
| }, | |
| { | |
| "epoch": 1.8022813688212929, | |
| "grad_norm": 0.6576415838227114, | |
| "learning_rate": 5.905175438178979e-07, | |
| "loss": 0.301, | |
| "step": 948 | |
| }, | |
| { | |
| "epoch": 1.8041825095057034, | |
| "grad_norm": 0.6518325809606182, | |
| "learning_rate": 5.793264532400311e-07, | |
| "loss": 0.3015, | |
| "step": 949 | |
| }, | |
| { | |
| "epoch": 1.806083650190114, | |
| "grad_norm": 0.6817978370160456, | |
| "learning_rate": 5.68239258710116e-07, | |
| "loss": 0.2957, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.8079847908745248, | |
| "grad_norm": 0.6963222677836269, | |
| "learning_rate": 5.572560825034523e-07, | |
| "loss": 0.2947, | |
| "step": 951 | |
| }, | |
| { | |
| "epoch": 1.8098859315589353, | |
| "grad_norm": 0.6754434786548771, | |
| "learning_rate": 5.463770457481732e-07, | |
| "loss": 0.2891, | |
| "step": 952 | |
| }, | |
| { | |
| "epoch": 1.811787072243346, | |
| "grad_norm": 0.6692341115649304, | |
| "learning_rate": 5.35602268423906e-07, | |
| "loss": 0.2944, | |
| "step": 953 | |
| }, | |
| { | |
| "epoch": 1.8136882129277567, | |
| "grad_norm": 0.7226461962940662, | |
| "learning_rate": 5.249318693604577e-07, | |
| "loss": 0.3082, | |
| "step": 954 | |
| }, | |
| { | |
| "epoch": 1.8155893536121672, | |
| "grad_norm": 0.7244706729297383, | |
| "learning_rate": 5.143659662364931e-07, | |
| "loss": 0.3039, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 1.817490494296578, | |
| "grad_norm": 0.7161324438774385, | |
| "learning_rate": 5.039046755782417e-07, | |
| "loss": 0.2857, | |
| "step": 956 | |
| }, | |
| { | |
| "epoch": 1.8193916349809887, | |
| "grad_norm": 0.6716404373737771, | |
| "learning_rate": 4.935481127582131e-07, | |
| "loss": 0.2828, | |
| "step": 957 | |
| }, | |
| { | |
| "epoch": 1.8212927756653992, | |
| "grad_norm": 0.7311263576451744, | |
| "learning_rate": 4.83296391993926e-07, | |
| "loss": 0.3099, | |
| "step": 958 | |
| }, | |
| { | |
| "epoch": 1.8231939163498099, | |
| "grad_norm": 0.6585136015360831, | |
| "learning_rate": 4.7314962634664616e-07, | |
| "loss": 0.2911, | |
| "step": 959 | |
| }, | |
| { | |
| "epoch": 1.8250950570342206, | |
| "grad_norm": 0.6789328880679895, | |
| "learning_rate": 4.631079277201389e-07, | |
| "loss": 0.2874, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 1.826996197718631, | |
| "grad_norm": 0.6743239103644939, | |
| "learning_rate": 4.5317140685943726e-07, | |
| "loss": 0.3019, | |
| "step": 961 | |
| }, | |
| { | |
| "epoch": 1.8288973384030418, | |
| "grad_norm": 0.6715060917777644, | |
| "learning_rate": 4.433401733496201e-07, | |
| "loss": 0.2931, | |
| "step": 962 | |
| }, | |
| { | |
| "epoch": 1.8307984790874525, | |
| "grad_norm": 0.7104784901068348, | |
| "learning_rate": 4.3361433561460274e-07, | |
| "loss": 0.3008, | |
| "step": 963 | |
| }, | |
| { | |
| "epoch": 1.832699619771863, | |
| "grad_norm": 0.7211822242998399, | |
| "learning_rate": 4.2399400091594154e-07, | |
| "loss": 0.2783, | |
| "step": 964 | |
| }, | |
| { | |
| "epoch": 1.8346007604562737, | |
| "grad_norm": 0.6389607614808079, | |
| "learning_rate": 4.14479275351648e-07, | |
| "loss": 0.2734, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 1.8365019011406845, | |
| "grad_norm": 0.6731532990647303, | |
| "learning_rate": 4.0507026385502747e-07, | |
| "loss": 0.2824, | |
| "step": 966 | |
| }, | |
| { | |
| "epoch": 1.838403041825095, | |
| "grad_norm": 0.6711783817710473, | |
| "learning_rate": 3.9576707019350903e-07, | |
| "loss": 0.295, | |
| "step": 967 | |
| }, | |
| { | |
| "epoch": 1.8403041825095057, | |
| "grad_norm": 0.729699431059001, | |
| "learning_rate": 3.865697969675164e-07, | |
| "loss": 0.2955, | |
| "step": 968 | |
| }, | |
| { | |
| "epoch": 1.8422053231939164, | |
| "grad_norm": 0.7538535074831761, | |
| "learning_rate": 3.7747854560931996e-07, | |
| "loss": 0.291, | |
| "step": 969 | |
| }, | |
| { | |
| "epoch": 1.8441064638783269, | |
| "grad_norm": 0.6586106232273593, | |
| "learning_rate": 3.684934163819309e-07, | |
| "loss": 0.2918, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 1.8460076045627376, | |
| "grad_norm": 0.6502051906564796, | |
| "learning_rate": 3.596145083779912e-07, | |
| "loss": 0.2899, | |
| "step": 971 | |
| }, | |
| { | |
| "epoch": 1.8479087452471483, | |
| "grad_norm": 0.6875049870257419, | |
| "learning_rate": 3.508419195186774e-07, | |
| "loss": 0.2817, | |
| "step": 972 | |
| }, | |
| { | |
| "epoch": 1.8498098859315588, | |
| "grad_norm": 0.6958405525823462, | |
| "learning_rate": 3.421757465526243e-07, | |
| "loss": 0.2984, | |
| "step": 973 | |
| }, | |
| { | |
| "epoch": 1.8517110266159695, | |
| "grad_norm": 0.7164382055659756, | |
| "learning_rate": 3.33616085054862e-07, | |
| "loss": 0.2942, | |
| "step": 974 | |
| }, | |
| { | |
| "epoch": 1.8536121673003803, | |
| "grad_norm": 0.7566751952531614, | |
| "learning_rate": 3.2516302942574794e-07, | |
| "loss": 0.2934, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 1.8555133079847907, | |
| "grad_norm": 0.7162187315670879, | |
| "learning_rate": 3.1681667288994353e-07, | |
| "loss": 0.2883, | |
| "step": 976 | |
| }, | |
| { | |
| "epoch": 1.8574144486692015, | |
| "grad_norm": 0.6925334864242433, | |
| "learning_rate": 3.0857710749537585e-07, | |
| "loss": 0.2895, | |
| "step": 977 | |
| }, | |
| { | |
| "epoch": 1.8593155893536122, | |
| "grad_norm": 0.6717382132622957, | |
| "learning_rate": 3.0044442411222066e-07, | |
| "loss": 0.282, | |
| "step": 978 | |
| }, | |
| { | |
| "epoch": 1.8612167300380227, | |
| "grad_norm": 0.6767277676375325, | |
| "learning_rate": 2.9241871243190555e-07, | |
| "loss": 0.2965, | |
| "step": 979 | |
| }, | |
| { | |
| "epoch": 1.8631178707224336, | |
| "grad_norm": 0.66575519388906, | |
| "learning_rate": 2.845000609661208e-07, | |
| "loss": 0.2788, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 1.8650190114068441, | |
| "grad_norm": 0.6814181631875712, | |
| "learning_rate": 2.7668855704583997e-07, | |
| "loss": 0.2859, | |
| "step": 981 | |
| }, | |
| { | |
| "epoch": 1.8669201520912546, | |
| "grad_norm": 0.6527478332742671, | |
| "learning_rate": 2.689842868203563e-07, | |
| "loss": 0.2835, | |
| "step": 982 | |
| }, | |
| { | |
| "epoch": 1.8688212927756656, | |
| "grad_norm": 0.7099140845255089, | |
| "learning_rate": 2.6138733525633896e-07, | |
| "loss": 0.289, | |
| "step": 983 | |
| }, | |
| { | |
| "epoch": 1.870722433460076, | |
| "grad_norm": 0.6647136389836994, | |
| "learning_rate": 2.5389778613688744e-07, | |
| "loss": 0.2946, | |
| "step": 984 | |
| }, | |
| { | |
| "epoch": 1.8726235741444865, | |
| "grad_norm": 0.6673575806563161, | |
| "learning_rate": 2.46515722060614e-07, | |
| "loss": 0.3012, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 1.8745247148288975, | |
| "grad_norm": 0.6500216155595889, | |
| "learning_rate": 2.392412244407294e-07, | |
| "loss": 0.2695, | |
| "step": 986 | |
| }, | |
| { | |
| "epoch": 1.876425855513308, | |
| "grad_norm": 0.6815248346754635, | |
| "learning_rate": 2.3207437350414418e-07, | |
| "loss": 0.2906, | |
| "step": 987 | |
| }, | |
| { | |
| "epoch": 1.8783269961977185, | |
| "grad_norm": 0.6912419532025785, | |
| "learning_rate": 2.2501524829059208e-07, | |
| "loss": 0.291, | |
| "step": 988 | |
| }, | |
| { | |
| "epoch": 1.8802281368821294, | |
| "grad_norm": 0.6868829980934273, | |
| "learning_rate": 2.180639266517448e-07, | |
| "loss": 0.2901, | |
| "step": 989 | |
| }, | |
| { | |
| "epoch": 1.88212927756654, | |
| "grad_norm": 0.6868014849754196, | |
| "learning_rate": 2.1122048525036409e-07, | |
| "loss": 0.2972, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 1.8840304182509504, | |
| "grad_norm": 0.6623801650154064, | |
| "learning_rate": 2.0448499955945223e-07, | |
| "loss": 0.2893, | |
| "step": 991 | |
| }, | |
| { | |
| "epoch": 1.8859315589353614, | |
| "grad_norm": 0.7096941969251608, | |
| "learning_rate": 1.9785754386142164e-07, | |
| "loss": 0.3021, | |
| "step": 992 | |
| }, | |
| { | |
| "epoch": 1.8878326996197718, | |
| "grad_norm": 0.6601496029556285, | |
| "learning_rate": 1.9133819124727003e-07, | |
| "loss": 0.3046, | |
| "step": 993 | |
| }, | |
| { | |
| "epoch": 1.8897338403041823, | |
| "grad_norm": 0.6928501522596534, | |
| "learning_rate": 1.8492701361578326e-07, | |
| "loss": 0.291, | |
| "step": 994 | |
| }, | |
| { | |
| "epoch": 1.8916349809885933, | |
| "grad_norm": 0.6738640892365713, | |
| "learning_rate": 1.7862408167273472e-07, | |
| "loss": 0.2972, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 1.8935361216730038, | |
| "grad_norm": 0.6758530654778283, | |
| "learning_rate": 1.724294649301095e-07, | |
| "loss": 0.3096, | |
| "step": 996 | |
| }, | |
| { | |
| "epoch": 1.8954372623574145, | |
| "grad_norm": 0.6795301570152035, | |
| "learning_rate": 1.6634323170533928e-07, | |
| "loss": 0.2908, | |
| "step": 997 | |
| }, | |
| { | |
| "epoch": 1.8973384030418252, | |
| "grad_norm": 0.6412708230903923, | |
| "learning_rate": 1.6036544912054087e-07, | |
| "loss": 0.2924, | |
| "step": 998 | |
| }, | |
| { | |
| "epoch": 1.8992395437262357, | |
| "grad_norm": 0.6555912955090373, | |
| "learning_rate": 1.544961831017855e-07, | |
| "loss": 0.2906, | |
| "step": 999 | |
| }, | |
| { | |
| "epoch": 1.9011406844106464, | |
| "grad_norm": 0.6818543773568423, | |
| "learning_rate": 1.487354983783673e-07, | |
| "loss": 0.2913, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.9030418250950571, | |
| "grad_norm": 0.6317098511265704, | |
| "learning_rate": 1.430834584820895e-07, | |
| "loss": 0.2847, | |
| "step": 1001 | |
| }, | |
| { | |
| "epoch": 1.9049429657794676, | |
| "grad_norm": 0.7189334226904048, | |
| "learning_rate": 1.375401257465625e-07, | |
| "loss": 0.3047, | |
| "step": 1002 | |
| }, | |
| { | |
| "epoch": 1.9068441064638784, | |
| "grad_norm": 0.657552331891348, | |
| "learning_rate": 1.3210556130652031e-07, | |
| "loss": 0.2925, | |
| "step": 1003 | |
| }, | |
| { | |
| "epoch": 1.908745247148289, | |
| "grad_norm": 0.6580875768097073, | |
| "learning_rate": 1.2677982509714415e-07, | |
| "loss": 0.2766, | |
| "step": 1004 | |
| }, | |
| { | |
| "epoch": 1.9106463878326996, | |
| "grad_norm": 0.6972344985396955, | |
| "learning_rate": 1.2156297585339872e-07, | |
| "loss": 0.2866, | |
| "step": 1005 | |
| }, | |
| { | |
| "epoch": 1.9125475285171103, | |
| "grad_norm": 0.6657583345648055, | |
| "learning_rate": 1.1645507110938925e-07, | |
| "loss": 0.2859, | |
| "step": 1006 | |
| }, | |
| { | |
| "epoch": 1.914448669201521, | |
| "grad_norm": 0.6670035297837602, | |
| "learning_rate": 1.1145616719772545e-07, | |
| "loss": 0.2987, | |
| "step": 1007 | |
| }, | |
| { | |
| "epoch": 1.9163498098859315, | |
| "grad_norm": 0.6625505886191134, | |
| "learning_rate": 1.0656631924889749e-07, | |
| "loss": 0.2979, | |
| "step": 1008 | |
| }, | |
| { | |
| "epoch": 1.9182509505703422, | |
| "grad_norm": 0.6765549170908274, | |
| "learning_rate": 1.0178558119067316e-07, | |
| "loss": 0.304, | |
| "step": 1009 | |
| }, | |
| { | |
| "epoch": 1.920152091254753, | |
| "grad_norm": 0.6947365194861944, | |
| "learning_rate": 9.711400574749507e-08, | |
| "loss": 0.2921, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 1.9220532319391634, | |
| "grad_norm": 0.6487311344768585, | |
| "learning_rate": 9.255164443990994e-08, | |
| "loss": 0.2931, | |
| "step": 1011 | |
| }, | |
| { | |
| "epoch": 1.9239543726235742, | |
| "grad_norm": 0.6984880842371495, | |
| "learning_rate": 8.809854758399017e-08, | |
| "loss": 0.3011, | |
| "step": 1012 | |
| }, | |
| { | |
| "epoch": 1.9258555133079849, | |
| "grad_norm": 0.6675294762264681, | |
| "learning_rate": 8.375476429078543e-08, | |
| "loss": 0.3005, | |
| "step": 1013 | |
| }, | |
| { | |
| "epoch": 1.9277566539923954, | |
| "grad_norm": 0.6572759346417789, | |
| "learning_rate": 7.952034246577977e-08, | |
| "loss": 0.2891, | |
| "step": 1014 | |
| }, | |
| { | |
| "epoch": 1.929657794676806, | |
| "grad_norm": 0.6500718504080977, | |
| "learning_rate": 7.539532880836087e-08, | |
| "loss": 0.2929, | |
| "step": 1015 | |
| }, | |
| { | |
| "epoch": 1.9315589353612168, | |
| "grad_norm": 0.6314021337581515, | |
| "learning_rate": 7.137976881130826e-08, | |
| "loss": 0.2929, | |
| "step": 1016 | |
| }, | |
| { | |
| "epoch": 1.9334600760456273, | |
| "grad_norm": 0.6886718970323088, | |
| "learning_rate": 6.747370676028819e-08, | |
| "loss": 0.2993, | |
| "step": 1017 | |
| }, | |
| { | |
| "epoch": 1.935361216730038, | |
| "grad_norm": 0.6937547211312483, | |
| "learning_rate": 6.367718573336845e-08, | |
| "loss": 0.2882, | |
| "step": 1018 | |
| }, | |
| { | |
| "epoch": 1.9372623574144487, | |
| "grad_norm": 0.6954041445203646, | |
| "learning_rate": 5.999024760054095e-08, | |
| "loss": 0.2904, | |
| "step": 1019 | |
| }, | |
| { | |
| "epoch": 1.9391634980988592, | |
| "grad_norm": 0.6572862586396218, | |
| "learning_rate": 5.641293302326323e-08, | |
| "loss": 0.2915, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 1.94106463878327, | |
| "grad_norm": 0.6854251251848709, | |
| "learning_rate": 5.2945281454003236e-08, | |
| "loss": 0.2921, | |
| "step": 1021 | |
| }, | |
| { | |
| "epoch": 1.9429657794676807, | |
| "grad_norm": 0.6423585291302776, | |
| "learning_rate": 4.958733113581415e-08, | |
| "loss": 0.3001, | |
| "step": 1022 | |
| }, | |
| { | |
| "epoch": 1.9448669201520912, | |
| "grad_norm": 0.678075622775272, | |
| "learning_rate": 4.6339119101902475e-08, | |
| "loss": 0.2885, | |
| "step": 1023 | |
| }, | |
| { | |
| "epoch": 1.946768060836502, | |
| "grad_norm": 0.674460240543576, | |
| "learning_rate": 4.320068117522835e-08, | |
| "loss": 0.2915, | |
| "step": 1024 | |
| }, | |
| { | |
| "epoch": 1.9486692015209126, | |
| "grad_norm": 0.652457567447936, | |
| "learning_rate": 4.0172051968101474e-08, | |
| "loss": 0.2832, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 1.950570342205323, | |
| "grad_norm": 0.6397147126044596, | |
| "learning_rate": 3.7253264881809137e-08, | |
| "loss": 0.2901, | |
| "step": 1026 | |
| }, | |
| { | |
| "epoch": 1.9524714828897338, | |
| "grad_norm": 0.6287873760327426, | |
| "learning_rate": 3.4444352106242086e-08, | |
| "loss": 0.2786, | |
| "step": 1027 | |
| }, | |
| { | |
| "epoch": 1.9543726235741445, | |
| "grad_norm": 0.6457260929010515, | |
| "learning_rate": 3.174534461953593e-08, | |
| "loss": 0.2856, | |
| "step": 1028 | |
| }, | |
| { | |
| "epoch": 1.956273764258555, | |
| "grad_norm": 0.6555189495485545, | |
| "learning_rate": 2.915627218774142e-08, | |
| "loss": 0.2861, | |
| "step": 1029 | |
| }, | |
| { | |
| "epoch": 1.9581749049429658, | |
| "grad_norm": 0.7242582011745291, | |
| "learning_rate": 2.667716336448356e-08, | |
| "loss": 0.3098, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 1.9600760456273765, | |
| "grad_norm": 0.6677402414930224, | |
| "learning_rate": 2.430804549065302e-08, | |
| "loss": 0.2958, | |
| "step": 1031 | |
| }, | |
| { | |
| "epoch": 1.961977186311787, | |
| "grad_norm": 0.6762512774148852, | |
| "learning_rate": 2.2048944694104123e-08, | |
| "loss": 0.2944, | |
| "step": 1032 | |
| }, | |
| { | |
| "epoch": 1.9638783269961977, | |
| "grad_norm": 0.7307960234898214, | |
| "learning_rate": 1.989988588936509e-08, | |
| "loss": 0.3095, | |
| "step": 1033 | |
| }, | |
| { | |
| "epoch": 1.9657794676806084, | |
| "grad_norm": 0.7069615691374648, | |
| "learning_rate": 1.7860892777367133e-08, | |
| "loss": 0.2865, | |
| "step": 1034 | |
| }, | |
| { | |
| "epoch": 1.967680608365019, | |
| "grad_norm": 0.6349795744185669, | |
| "learning_rate": 1.5931987845176912e-08, | |
| "loss": 0.2702, | |
| "step": 1035 | |
| }, | |
| { | |
| "epoch": 1.9695817490494296, | |
| "grad_norm": 0.6512755987024023, | |
| "learning_rate": 1.411319236575337e-08, | |
| "loss": 0.2862, | |
| "step": 1036 | |
| }, | |
| { | |
| "epoch": 1.9714828897338403, | |
| "grad_norm": 0.7141562913785444, | |
| "learning_rate": 1.2404526397711281e-08, | |
| "loss": 0.299, | |
| "step": 1037 | |
| }, | |
| { | |
| "epoch": 1.9733840304182508, | |
| "grad_norm": 0.6528174592269812, | |
| "learning_rate": 1.0806008785100297e-08, | |
| "loss": 0.2695, | |
| "step": 1038 | |
| }, | |
| { | |
| "epoch": 1.9752851711026616, | |
| "grad_norm": 0.7470505123681149, | |
| "learning_rate": 9.317657157197347e-09, | |
| "loss": 0.3072, | |
| "step": 1039 | |
| }, | |
| { | |
| "epoch": 1.9771863117870723, | |
| "grad_norm": 0.7176889173743611, | |
| "learning_rate": 7.93948792831234e-09, | |
| "loss": 0.2939, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 1.9790874524714828, | |
| "grad_norm": 0.6794862298531197, | |
| "learning_rate": 6.671516297606095e-09, | |
| "loss": 0.2875, | |
| "step": 1041 | |
| }, | |
| { | |
| "epoch": 1.9809885931558935, | |
| "grad_norm": 0.688928738993111, | |
| "learning_rate": 5.513756248924917e-09, | |
| "loss": 0.292, | |
| "step": 1042 | |
| }, | |
| { | |
| "epoch": 1.9828897338403042, | |
| "grad_norm": 0.7024316388036554, | |
| "learning_rate": 4.466220550641831e-09, | |
| "loss": 0.2901, | |
| "step": 1043 | |
| }, | |
| { | |
| "epoch": 1.9847908745247147, | |
| "grad_norm": 0.6988774411715154, | |
| "learning_rate": 3.528920755523357e-09, | |
| "loss": 0.2981, | |
| "step": 1044 | |
| }, | |
| { | |
| "epoch": 1.9866920152091256, | |
| "grad_norm": 0.7052888557832583, | |
| "learning_rate": 2.701867200592956e-09, | |
| "loss": 0.289, | |
| "step": 1045 | |
| }, | |
| { | |
| "epoch": 1.9885931558935361, | |
| "grad_norm": 0.6497681881360204, | |
| "learning_rate": 1.9850690070266633e-09, | |
| "loss": 0.2878, | |
| "step": 1046 | |
| }, | |
| { | |
| "epoch": 1.9904942965779466, | |
| "grad_norm": 0.6767942631840376, | |
| "learning_rate": 1.378534080042071e-09, | |
| "loss": 0.2933, | |
| "step": 1047 | |
| }, | |
| { | |
| "epoch": 1.9923954372623576, | |
| "grad_norm": 0.6578703652998189, | |
| "learning_rate": 8.822691088195001e-10, | |
| "loss": 0.2802, | |
| "step": 1048 | |
| }, | |
| { | |
| "epoch": 1.994296577946768, | |
| "grad_norm": 0.6793374142080242, | |
| "learning_rate": 4.962795664265052e-10, | |
| "loss": 0.2799, | |
| "step": 1049 | |
| }, | |
| { | |
| "epoch": 1.9961977186311786, | |
| "grad_norm": 0.6422987240197928, | |
| "learning_rate": 2.2056970975459223e-10, | |
| "loss": 0.2849, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.9980988593155895, | |
| "grad_norm": 0.7141345348779329, | |
| "learning_rate": 5.514257947369928e-11, | |
| "loss": 0.3059, | |
| "step": 1051 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "grad_norm": 0.65229679246691, | |
| "learning_rate": 0.0, | |
| "loss": 0.2935, | |
| "step": 1052 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "step": 1052, | |
| "total_flos": 116557120143360.0, | |
| "train_loss": 0.3777812981956812, | |
| "train_runtime": 3566.5914, | |
| "train_samples_per_second": 37.755, | |
| "train_steps_per_second": 0.295 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 1052, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 2, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 116557120143360.0, | |
| "train_batch_size": 2, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |