| { | |
| "best_metric": 1.5042036771774292, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-150", | |
| "epoch": 3.009009009009009, | |
| "eval_steps": 50, | |
| "global_step": 167, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.018018018018018018, | |
| "grad_norm": 0.20094779133796692, | |
| "learning_rate": 1e-05, | |
| "loss": 1.7238, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.018018018018018018, | |
| "eval_loss": 2.4445245265960693, | |
| "eval_runtime": 1.9421, | |
| "eval_samples_per_second": 48.402, | |
| "eval_steps_per_second": 12.358, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.036036036036036036, | |
| "grad_norm": 0.23159636557102203, | |
| "learning_rate": 2e-05, | |
| "loss": 2.3157, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.05405405405405406, | |
| "grad_norm": 0.17576082050800323, | |
| "learning_rate": 3e-05, | |
| "loss": 1.889, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.07207207207207207, | |
| "grad_norm": 0.18563885986804962, | |
| "learning_rate": 4e-05, | |
| "loss": 1.8117, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.09009009009009009, | |
| "grad_norm": 0.2837836742401123, | |
| "learning_rate": 5e-05, | |
| "loss": 2.5249, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.10810810810810811, | |
| "grad_norm": 0.5448350310325623, | |
| "learning_rate": 6e-05, | |
| "loss": 2.5175, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.12612612612612611, | |
| "grad_norm": 0.5622051358222961, | |
| "learning_rate": 7e-05, | |
| "loss": 2.6268, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.14414414414414414, | |
| "grad_norm": 0.5050106048583984, | |
| "learning_rate": 8e-05, | |
| "loss": 2.5345, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.16216216216216217, | |
| "grad_norm": 0.3459974229335785, | |
| "learning_rate": 9e-05, | |
| "loss": 2.6653, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.18018018018018017, | |
| "grad_norm": 0.2534170150756836, | |
| "learning_rate": 0.0001, | |
| "loss": 1.9698, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.1981981981981982, | |
| "grad_norm": 0.2756398916244507, | |
| "learning_rate": 9.998999018714263e-05, | |
| "loss": 1.7193, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.21621621621621623, | |
| "grad_norm": 0.3350951075553894, | |
| "learning_rate": 9.995996475642466e-05, | |
| "loss": 2.133, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.23423423423423423, | |
| "grad_norm": 0.5594307780265808, | |
| "learning_rate": 9.990993572980378e-05, | |
| "loss": 3.0643, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.25225225225225223, | |
| "grad_norm": 0.13822466135025024, | |
| "learning_rate": 9.983992313852774e-05, | |
| "loss": 1.7842, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.2702702702702703, | |
| "grad_norm": 0.14877331256866455, | |
| "learning_rate": 9.974995501511404e-05, | |
| "loss": 2.2152, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.2882882882882883, | |
| "grad_norm": 0.13884077966213226, | |
| "learning_rate": 9.964006738212575e-05, | |
| "loss": 1.8158, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.3063063063063063, | |
| "grad_norm": 0.17055541276931763, | |
| "learning_rate": 9.951030423774859e-05, | |
| "loss": 2.0037, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.32432432432432434, | |
| "grad_norm": 0.44902870059013367, | |
| "learning_rate": 9.936071753817415e-05, | |
| "loss": 2.1393, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.34234234234234234, | |
| "grad_norm": 0.4291950762271881, | |
| "learning_rate": 9.919136717679722e-05, | |
| "loss": 2.232, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.36036036036036034, | |
| "grad_norm": 0.46815672516822815, | |
| "learning_rate": 9.900232096023477e-05, | |
| "loss": 2.1575, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.3783783783783784, | |
| "grad_norm": 0.35431531071662903, | |
| "learning_rate": 9.879365458117678e-05, | |
| "loss": 2.1213, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.3963963963963964, | |
| "grad_norm": 0.2607969045639038, | |
| "learning_rate": 9.856545158807938e-05, | |
| "loss": 2.4743, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.4144144144144144, | |
| "grad_norm": 0.20863136649131775, | |
| "learning_rate": 9.831780335171279e-05, | |
| "loss": 1.7214, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.43243243243243246, | |
| "grad_norm": 0.31846967339515686, | |
| "learning_rate": 9.805080902857699e-05, | |
| "loss": 2.2591, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.45045045045045046, | |
| "grad_norm": 0.34938788414001465, | |
| "learning_rate": 9.776457552120033e-05, | |
| "loss": 2.2996, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.46846846846846846, | |
| "grad_norm": 0.6246058940887451, | |
| "learning_rate": 9.745921743533653e-05, | |
| "loss": 2.1906, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.4864864864864865, | |
| "grad_norm": 0.14733730256557465, | |
| "learning_rate": 9.713485703407731e-05, | |
| "loss": 1.6917, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.5045045045045045, | |
| "grad_norm": 0.2119920402765274, | |
| "learning_rate": 9.679162418889931e-05, | |
| "loss": 2.1286, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.5225225225225225, | |
| "grad_norm": 0.20382221043109894, | |
| "learning_rate": 9.642965632766436e-05, | |
| "loss": 1.889, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.5405405405405406, | |
| "grad_norm": 0.22297801077365875, | |
| "learning_rate": 9.604909837959455e-05, | |
| "loss": 2.1375, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.5585585585585585, | |
| "grad_norm": 0.29066169261932373, | |
| "learning_rate": 9.565010271724352e-05, | |
| "loss": 1.6588, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.5765765765765766, | |
| "grad_norm": 0.4789801836013794, | |
| "learning_rate": 9.523282909548773e-05, | |
| "loss": 1.7155, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.5945945945945946, | |
| "grad_norm": 0.5113940834999084, | |
| "learning_rate": 9.47974445875617e-05, | |
| "loss": 1.68, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.6126126126126126, | |
| "grad_norm": 0.5598991513252258, | |
| "learning_rate": 9.434412351816328e-05, | |
| "loss": 1.5835, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.6306306306306306, | |
| "grad_norm": 0.23446883261203766, | |
| "learning_rate": 9.387304739365523e-05, | |
| "loss": 2.0415, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.6486486486486487, | |
| "grad_norm": 0.23417960107326508, | |
| "learning_rate": 9.338440482939146e-05, | |
| "loss": 1.9709, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.6666666666666666, | |
| "grad_norm": 0.2393627017736435, | |
| "learning_rate": 9.287839147419686e-05, | |
| "loss": 1.809, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.6846846846846847, | |
| "grad_norm": 0.37448129057884216, | |
| "learning_rate": 9.2355209932031e-05, | |
| "loss": 2.3969, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.7027027027027027, | |
| "grad_norm": 0.5388697385787964, | |
| "learning_rate": 9.181506968086697e-05, | |
| "loss": 2.8039, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.7207207207207207, | |
| "grad_norm": 0.13199251890182495, | |
| "learning_rate": 9.125818698881798e-05, | |
| "loss": 1.6472, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.7387387387387387, | |
| "grad_norm": 0.21171270310878754, | |
| "learning_rate": 9.068478482754532e-05, | |
| "loss": 2.5439, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.7567567567567568, | |
| "grad_norm": 0.19285422563552856, | |
| "learning_rate": 9.0095092782982e-05, | |
| "loss": 1.7294, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.7747747747747747, | |
| "grad_norm": 0.19233770668506622, | |
| "learning_rate": 8.948934696340843e-05, | |
| "loss": 1.8547, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.7927927927927928, | |
| "grad_norm": 2.065861225128174, | |
| "learning_rate": 8.886778990491631e-05, | |
| "loss": 1.4244, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.8108108108108109, | |
| "grad_norm": 1.3515313863754272, | |
| "learning_rate": 8.823067047429907e-05, | |
| "loss": 1.3373, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.8288288288288288, | |
| "grad_norm": 0.7247907519340515, | |
| "learning_rate": 8.757824376940746e-05, | |
| "loss": 1.1633, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.8468468468468469, | |
| "grad_norm": 0.5490639209747314, | |
| "learning_rate": 8.691077101701024e-05, | |
| "loss": 1.3324, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.8648648648648649, | |
| "grad_norm": 0.2311074584722519, | |
| "learning_rate": 8.622851946820095e-05, | |
| "loss": 2.2242, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.8828828828828829, | |
| "grad_norm": 0.20029251277446747, | |
| "learning_rate": 8.553176229139261e-05, | |
| "loss": 1.7394, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.9009009009009009, | |
| "grad_norm": 0.27592265605926514, | |
| "learning_rate": 8.482077846294308e-05, | |
| "loss": 2.1438, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.9009009009009009, | |
| "eval_loss": 1.783769965171814, | |
| "eval_runtime": 1.9447, | |
| "eval_samples_per_second": 48.337, | |
| "eval_steps_per_second": 12.341, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.918918918918919, | |
| "grad_norm": 0.3021310269832611, | |
| "learning_rate": 8.409585265545509e-05, | |
| "loss": 2.0702, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.9369369369369369, | |
| "grad_norm": 0.5802788734436035, | |
| "learning_rate": 8.335727512379534e-05, | |
| "loss": 2.8902, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.954954954954955, | |
| "grad_norm": 0.1312423199415207, | |
| "learning_rate": 8.260534158887876e-05, | |
| "loss": 1.7881, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.972972972972973, | |
| "grad_norm": 0.742950439453125, | |
| "learning_rate": 8.184035311926396e-05, | |
| "loss": 1.1818, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.990990990990991, | |
| "grad_norm": 0.3108900189399719, | |
| "learning_rate": 8.106261601060772e-05, | |
| "loss": 1.7755, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 1.009009009009009, | |
| "grad_norm": 0.5375937223434448, | |
| "learning_rate": 8.027244166302642e-05, | |
| "loss": 3.1039, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 1.027027027027027, | |
| "grad_norm": 0.21149364113807678, | |
| "learning_rate": 7.947014645641379e-05, | |
| "loss": 2.5015, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 1.045045045045045, | |
| "grad_norm": 0.1894480586051941, | |
| "learning_rate": 7.865605162376486e-05, | |
| "loss": 1.9007, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 1.063063063063063, | |
| "grad_norm": 0.20430035889148712, | |
| "learning_rate": 7.783048312255653e-05, | |
| "loss": 1.7878, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 1.0810810810810811, | |
| "grad_norm": 0.2826579511165619, | |
| "learning_rate": 7.699377150423672e-05, | |
| "loss": 1.5133, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 1.0990990990990992, | |
| "grad_norm": 0.7269498705863953, | |
| "learning_rate": 7.614625178187402e-05, | |
| "loss": 0.8433, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 1.117117117117117, | |
| "grad_norm": 0.6260727047920227, | |
| "learning_rate": 7.528826329602099e-05, | |
| "loss": 0.8916, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 1.135135135135135, | |
| "grad_norm": 0.5703617334365845, | |
| "learning_rate": 7.442014957884472e-05, | |
| "loss": 0.8438, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 1.1531531531531531, | |
| "grad_norm": 0.27178001403808594, | |
| "learning_rate": 7.354225821657914e-05, | |
| "loss": 2.1417, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 1.1711711711711712, | |
| "grad_norm": 0.23363694548606873, | |
| "learning_rate": 7.265494071035401e-05, | |
| "loss": 1.7944, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 1.1891891891891893, | |
| "grad_norm": 0.2847091257572174, | |
| "learning_rate": 7.175855233545668e-05, | |
| "loss": 1.751, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 1.2072072072072073, | |
| "grad_norm": 0.3543185293674469, | |
| "learning_rate": 7.085345199908235e-05, | |
| "loss": 2.0152, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 1.2252252252252251, | |
| "grad_norm": 0.5327386856079102, | |
| "learning_rate": 6.994000209663036e-05, | |
| "loss": 2.6897, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 1.2432432432432432, | |
| "grad_norm": 0.2581247091293335, | |
| "learning_rate": 6.901856836660386e-05, | |
| "loss": 1.2071, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 1.2612612612612613, | |
| "grad_norm": 0.18185599148273468, | |
| "learning_rate": 6.808951974417078e-05, | |
| "loss": 2.24, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 1.2792792792792793, | |
| "grad_norm": 0.21230140328407288, | |
| "learning_rate": 6.715322821344494e-05, | |
| "loss": 2.0791, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 1.2972972972972974, | |
| "grad_norm": 0.20122762024402618, | |
| "learning_rate": 6.621006865854644e-05, | |
| "loss": 1.6265, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 1.3153153153153152, | |
| "grad_norm": 0.22769460082054138, | |
| "learning_rate": 6.526041871350086e-05, | |
| "loss": 1.5291, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 1.3333333333333333, | |
| "grad_norm": 1.8807597160339355, | |
| "learning_rate": 6.43046586110374e-05, | |
| "loss": 0.7391, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 1.3513513513513513, | |
| "grad_norm": 1.10353422164917, | |
| "learning_rate": 6.334317103034652e-05, | |
| "loss": 0.6971, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 1.3693693693693694, | |
| "grad_norm": 0.6892307996749878, | |
| "learning_rate": 6.237634094385813e-05, | |
| "loss": 0.7755, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 1.3873873873873874, | |
| "grad_norm": 0.28367915749549866, | |
| "learning_rate": 6.140455546310148e-05, | |
| "loss": 2.0102, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 1.4054054054054055, | |
| "grad_norm": 0.354563444852829, | |
| "learning_rate": 6.042820368370854e-05, | |
| "loss": 2.0273, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 1.4234234234234235, | |
| "grad_norm": 0.2600463926792145, | |
| "learning_rate": 5.944767652962309e-05, | |
| "loss": 1.6145, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 1.4414414414414414, | |
| "grad_norm": 0.29232466220855713, | |
| "learning_rate": 5.8463366596577706e-05, | |
| "loss": 2.1392, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 1.4594594594594594, | |
| "grad_norm": 0.3904821276664734, | |
| "learning_rate": 5.747566799490132e-05, | |
| "loss": 2.3455, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 1.4774774774774775, | |
| "grad_norm": 0.3034067153930664, | |
| "learning_rate": 5.648497619172042e-05, | |
| "loss": 1.6627, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 1.4954954954954955, | |
| "grad_norm": 0.18810759484767914, | |
| "learning_rate": 5.549168785261698e-05, | |
| "loss": 2.3614, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 1.5135135135135136, | |
| "grad_norm": 0.15777312219142914, | |
| "learning_rate": 5.4496200682806495e-05, | |
| "loss": 1.697, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 1.5315315315315314, | |
| "grad_norm": 0.19415871798992157, | |
| "learning_rate": 5.3498913267899864e-05, | |
| "loss": 1.9134, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 1.5495495495495497, | |
| "grad_norm": 0.17604829370975494, | |
| "learning_rate": 5.250022491431259e-05, | |
| "loss": 1.8209, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 1.5675675675675675, | |
| "grad_norm": 1.056093692779541, | |
| "learning_rate": 5.150053548938557e-05, | |
| "loss": 0.6133, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 1.5855855855855856, | |
| "grad_norm": 0.8092790246009827, | |
| "learning_rate": 5.0500245261281175e-05, | |
| "loss": 0.7028, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 1.6036036036036037, | |
| "grad_norm": 0.6385818123817444, | |
| "learning_rate": 4.949975473871884e-05, | |
| "loss": 0.4955, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 1.6216216216216215, | |
| "grad_norm": 0.2813832759857178, | |
| "learning_rate": 4.849946451061443e-05, | |
| "loss": 1.7059, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 1.6396396396396398, | |
| "grad_norm": 0.25375688076019287, | |
| "learning_rate": 4.749977508568742e-05, | |
| "loss": 2.027, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 1.6576576576576576, | |
| "grad_norm": 0.28173476457595825, | |
| "learning_rate": 4.650108673210015e-05, | |
| "loss": 1.5781, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 1.6756756756756757, | |
| "grad_norm": 0.37018927931785583, | |
| "learning_rate": 4.550379931719351e-05, | |
| "loss": 1.9611, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 1.6936936936936937, | |
| "grad_norm": 0.44017812609672546, | |
| "learning_rate": 4.4508312147383036e-05, | |
| "loss": 1.9964, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 1.7117117117117115, | |
| "grad_norm": 0.27065154910087585, | |
| "learning_rate": 4.3515023808279586e-05, | |
| "loss": 1.7931, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 1.7297297297297298, | |
| "grad_norm": 0.1628153771162033, | |
| "learning_rate": 4.252433200509869e-05, | |
| "loss": 2.0541, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 1.7477477477477477, | |
| "grad_norm": 0.15087255835533142, | |
| "learning_rate": 4.1536633403422306e-05, | |
| "loss": 1.5446, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 1.7657657657657657, | |
| "grad_norm": 0.19169510900974274, | |
| "learning_rate": 4.0552323470376916e-05, | |
| "loss": 2.1779, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 1.7837837837837838, | |
| "grad_norm": 0.17723213136196136, | |
| "learning_rate": 3.9571796316291476e-05, | |
| "loss": 1.5049, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 1.8018018018018018, | |
| "grad_norm": 0.5622679591178894, | |
| "learning_rate": 3.859544453689853e-05, | |
| "loss": 0.538, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.8018018018018018, | |
| "eval_loss": 1.5607163906097412, | |
| "eval_runtime": 1.8915, | |
| "eval_samples_per_second": 49.696, | |
| "eval_steps_per_second": 12.688, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.8198198198198199, | |
| "grad_norm": 0.459534227848053, | |
| "learning_rate": 3.762365905614187e-05, | |
| "loss": 0.6408, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 1.8378378378378377, | |
| "grad_norm": 0.45416250824928284, | |
| "learning_rate": 3.665682896965349e-05, | |
| "loss": 0.6204, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 1.855855855855856, | |
| "grad_norm": 0.3494967818260193, | |
| "learning_rate": 3.5695341388962614e-05, | |
| "loss": 0.8394, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 1.8738738738738738, | |
| "grad_norm": 0.20583558082580566, | |
| "learning_rate": 3.473958128649915e-05, | |
| "loss": 1.7962, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 1.8918918918918919, | |
| "grad_norm": 0.2505161762237549, | |
| "learning_rate": 3.378993134145356e-05, | |
| "loss": 1.8203, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 1.90990990990991, | |
| "grad_norm": 0.2894545793533325, | |
| "learning_rate": 3.284677178655507e-05, | |
| "loss": 1.7578, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 1.9279279279279278, | |
| "grad_norm": 0.36985740065574646, | |
| "learning_rate": 3.1910480255829237e-05, | |
| "loss": 2.3144, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 1.945945945945946, | |
| "grad_norm": 0.29496514797210693, | |
| "learning_rate": 3.098143163339615e-05, | |
| "loss": 2.2301, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 1.9639639639639639, | |
| "grad_norm": 0.13244111835956573, | |
| "learning_rate": 3.0059997903369656e-05, | |
| "loss": 1.2973, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 1.981981981981982, | |
| "grad_norm": 0.1934548020362854, | |
| "learning_rate": 2.914654800091768e-05, | |
| "loss": 1.0715, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "grad_norm": 0.7730455994606018, | |
| "learning_rate": 2.824144766454333e-05, | |
| "loss": 2.955, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 2.018018018018018, | |
| "grad_norm": 0.14186394214630127, | |
| "learning_rate": 2.7345059289646008e-05, | |
| "loss": 1.6136, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 2.036036036036036, | |
| "grad_norm": 0.1974637508392334, | |
| "learning_rate": 2.6457741783420886e-05, | |
| "loss": 2.1953, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 2.054054054054054, | |
| "grad_norm": 0.15466494858264923, | |
| "learning_rate": 2.5579850421155293e-05, | |
| "loss": 1.5703, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 2.0720720720720722, | |
| "grad_norm": 0.17066383361816406, | |
| "learning_rate": 2.4711736703979018e-05, | |
| "loss": 1.8892, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 2.09009009009009, | |
| "grad_norm": 0.18141521513462067, | |
| "learning_rate": 2.3853748218126e-05, | |
| "loss": 0.7031, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 2.108108108108108, | |
| "grad_norm": 0.1847659945487976, | |
| "learning_rate": 2.3006228495763295e-05, | |
| "loss": 0.425, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 2.126126126126126, | |
| "grad_norm": 0.18299931287765503, | |
| "learning_rate": 2.2169516877443485e-05, | |
| "loss": 0.4302, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 2.144144144144144, | |
| "grad_norm": 0.1804618388414383, | |
| "learning_rate": 2.1343948376235144e-05, | |
| "loss": 1.5657, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 2.1621621621621623, | |
| "grad_norm": 0.22961750626564026, | |
| "learning_rate": 2.052985354358622e-05, | |
| "loss": 1.9153, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 2.18018018018018, | |
| "grad_norm": 0.24420714378356934, | |
| "learning_rate": 1.9727558336973595e-05, | |
| "loss": 1.726, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 2.1981981981981984, | |
| "grad_norm": 0.27104926109313965, | |
| "learning_rate": 1.8937383989392294e-05, | |
| "loss": 1.7018, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 2.2162162162162162, | |
| "grad_norm": 0.37170183658599854, | |
| "learning_rate": 1.8159646880736036e-05, | |
| "loss": 1.9319, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 2.234234234234234, | |
| "grad_norm": 0.4491060674190521, | |
| "learning_rate": 1.739465841112125e-05, | |
| "loss": 2.0411, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 2.2522522522522523, | |
| "grad_norm": 0.1254836916923523, | |
| "learning_rate": 1.664272487620466e-05, | |
| "loss": 1.4904, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 2.27027027027027, | |
| "grad_norm": 0.15240132808685303, | |
| "learning_rate": 1.590414734454493e-05, | |
| "loss": 2.1468, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 2.2882882882882885, | |
| "grad_norm": 0.16004954278469086, | |
| "learning_rate": 1.517922153705692e-05, | |
| "loss": 1.8539, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 2.3063063063063063, | |
| "grad_norm": 0.18988263607025146, | |
| "learning_rate": 1.4468237708607397e-05, | |
| "loss": 1.7977, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 2.3243243243243246, | |
| "grad_norm": 0.18853111565113068, | |
| "learning_rate": 1.3771480531799052e-05, | |
| "loss": 1.6049, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 2.3423423423423424, | |
| "grad_norm": 0.1556941568851471, | |
| "learning_rate": 1.308922898298977e-05, | |
| "loss": 0.3974, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 2.3603603603603602, | |
| "grad_norm": 0.17364652454853058, | |
| "learning_rate": 1.2421756230592534e-05, | |
| "loss": 0.4047, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 2.3783783783783785, | |
| "grad_norm": 0.15375109016895294, | |
| "learning_rate": 1.1769329525700935e-05, | |
| "loss": 0.443, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 2.3963963963963963, | |
| "grad_norm": 0.2470708191394806, | |
| "learning_rate": 1.1132210095083694e-05, | |
| "loss": 1.5159, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 2.4144144144144146, | |
| "grad_norm": 0.21014344692230225, | |
| "learning_rate": 1.0510653036591583e-05, | |
| "loss": 1.8404, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 2.4324324324324325, | |
| "grad_norm": 0.23266132175922394, | |
| "learning_rate": 9.904907217018e-06, | |
| "loss": 1.3737, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 2.4504504504504503, | |
| "grad_norm": 0.32797127962112427, | |
| "learning_rate": 9.31521517245469e-06, | |
| "loss": 2.1399, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 2.4684684684684686, | |
| "grad_norm": 0.4745734930038452, | |
| "learning_rate": 8.741813011182014e-06, | |
| "loss": 2.2424, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 2.4864864864864864, | |
| "grad_norm": 0.13079120218753815, | |
| "learning_rate": 8.18493031913305e-06, | |
| "loss": 1.5224, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 2.5045045045045047, | |
| "grad_norm": 0.1728202998638153, | |
| "learning_rate": 7.644790067969005e-06, | |
| "loss": 2.0656, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 2.5225225225225225, | |
| "grad_norm": 0.20250830054283142, | |
| "learning_rate": 7.1216085258031414e-06, | |
| "loss": 2.0862, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 2.5405405405405403, | |
| "grad_norm": 0.16197475790977478, | |
| "learning_rate": 6.6155951706085405e-06, | |
| "loss": 1.4935, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 2.5585585585585586, | |
| "grad_norm": 0.15914876759052277, | |
| "learning_rate": 6.1269526063447765e-06, | |
| "loss": 0.5048, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 2.5765765765765765, | |
| "grad_norm": 0.18777939677238464, | |
| "learning_rate": 5.6558764818367195e-06, | |
| "loss": 0.6094, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 2.5945945945945947, | |
| "grad_norm": 0.14933425188064575, | |
| "learning_rate": 5.2025554124383095e-06, | |
| "loss": 0.4805, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 2.6126126126126126, | |
| "grad_norm": 0.1998123824596405, | |
| "learning_rate": 4.767170904512292e-06, | |
| "loss": 1.3946, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 2.6306306306306304, | |
| "grad_norm": 0.270035982131958, | |
| "learning_rate": 4.349897282756487e-06, | |
| "loss": 2.0762, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 2.6486486486486487, | |
| "grad_norm": 0.22045235335826874, | |
| "learning_rate": 3.95090162040545e-06, | |
| "loss": 1.8445, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 2.6666666666666665, | |
| "grad_norm": 0.25569671392440796, | |
| "learning_rate": 3.570343672335641e-06, | |
| "loss": 1.7808, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 2.684684684684685, | |
| "grad_norm": 0.3833853602409363, | |
| "learning_rate": 3.2083758111006945e-06, | |
| "loss": 1.7097, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 2.7027027027027026, | |
| "grad_norm": 0.46845442056655884, | |
| "learning_rate": 2.86514296592269e-06, | |
| "loss": 2.2938, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 2.7027027027027026, | |
| "eval_loss": 1.5042036771774292, | |
| "eval_runtime": 1.8931, | |
| "eval_samples_per_second": 49.654, | |
| "eval_steps_per_second": 12.678, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 2.7207207207207205, | |
| "grad_norm": 0.1713583618402481, | |
| "learning_rate": 2.54078256466348e-06, | |
| "loss": 1.8336, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 2.7387387387387387, | |
| "grad_norm": 0.17491984367370605, | |
| "learning_rate": 2.2354244787996748e-06, | |
| "loss": 2.1273, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 2.756756756756757, | |
| "grad_norm": 0.1518978625535965, | |
| "learning_rate": 1.9491909714230204e-06, | |
| "loss": 1.8808, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 2.774774774774775, | |
| "grad_norm": 0.16476936638355255, | |
| "learning_rate": 1.6821966482872264e-06, | |
| "loss": 1.7178, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 2.7927927927927927, | |
| "grad_norm": 0.16660767793655396, | |
| "learning_rate": 1.434548411920622e-06, | |
| "loss": 1.1728, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 2.810810810810811, | |
| "grad_norm": 0.15310387313365936, | |
| "learning_rate": 1.206345418823235e-06, | |
| "loss": 0.4533, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 2.828828828828829, | |
| "grad_norm": 0.16932179033756256, | |
| "learning_rate": 9.976790397652315e-07, | |
| "loss": 0.4293, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 2.846846846846847, | |
| "grad_norm": 0.2117478847503662, | |
| "learning_rate": 8.086328232027873e-07, | |
| "loss": 0.5155, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 2.864864864864865, | |
| "grad_norm": 0.22566819190979004, | |
| "learning_rate": 6.392824618258519e-07, | |
| "loss": 1.8127, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 2.8828828828828827, | |
| "grad_norm": 0.2261633276939392, | |
| "learning_rate": 4.896957622514298e-07, | |
| "loss": 1.6093, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 2.900900900900901, | |
| "grad_norm": 0.2446056753396988, | |
| "learning_rate": 3.599326178742535e-07, | |
| "loss": 1.5191, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 2.918918918918919, | |
| "grad_norm": 0.3451971411705017, | |
| "learning_rate": 2.500449848859776e-07, | |
| "loss": 1.9343, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 2.936936936936937, | |
| "grad_norm": 0.46758314967155457, | |
| "learning_rate": 1.6007686147225254e-07, | |
| "loss": 2.3823, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 2.954954954954955, | |
| "grad_norm": 0.15292246639728546, | |
| "learning_rate": 9.006427019622176e-08, | |
| "loss": 1.9701, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 2.972972972972973, | |
| "grad_norm": 0.1602296382188797, | |
| "learning_rate": 4.0035243575342605e-08, | |
| "loss": 0.7246, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 2.990990990990991, | |
| "grad_norm": 0.2040543258190155, | |
| "learning_rate": 1.0009812857370016e-08, | |
| "loss": 1.3962, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 3.009009009009009, | |
| "grad_norm": 0.4154304265975952, | |
| "learning_rate": 0.0, | |
| "loss": 2.7024, | |
| "step": 167 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 167, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 4, | |
| "save_steps": 50, | |
| "stateful_callbacks": { | |
| "EarlyStoppingCallback": { | |
| "args": { | |
| "early_stopping_patience": 5, | |
| "early_stopping_threshold": 0.0 | |
| }, | |
| "attributes": { | |
| "early_stopping_patience_counter": 0 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1.268381537599488e+16, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |