| { | |
| "best_metric": 2.39079213142395, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.25157232704402516, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0012578616352201257, | |
| "grad_norm": 0.7913956642150879, | |
| "learning_rate": 1e-05, | |
| "loss": 2.246, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0012578616352201257, | |
| "eval_loss": 2.8562700748443604, | |
| "eval_runtime": 28.0145, | |
| "eval_samples_per_second": 47.797, | |
| "eval_steps_per_second": 11.958, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0025157232704402514, | |
| "grad_norm": 0.9761772155761719, | |
| "learning_rate": 2e-05, | |
| "loss": 2.6848, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0037735849056603774, | |
| "grad_norm": 0.9434037208557129, | |
| "learning_rate": 3e-05, | |
| "loss": 2.7153, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.005031446540880503, | |
| "grad_norm": 1.0020833015441895, | |
| "learning_rate": 4e-05, | |
| "loss": 2.7106, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.006289308176100629, | |
| "grad_norm": 0.9618245363235474, | |
| "learning_rate": 5e-05, | |
| "loss": 2.797, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.007547169811320755, | |
| "grad_norm": 0.9292449355125427, | |
| "learning_rate": 6e-05, | |
| "loss": 2.596, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.00880503144654088, | |
| "grad_norm": 0.9059561491012573, | |
| "learning_rate": 7e-05, | |
| "loss": 2.6076, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.010062893081761006, | |
| "grad_norm": 0.9207968711853027, | |
| "learning_rate": 8e-05, | |
| "loss": 2.6015, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.011320754716981131, | |
| "grad_norm": 1.0058094263076782, | |
| "learning_rate": 9e-05, | |
| "loss": 2.5945, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.012578616352201259, | |
| "grad_norm": 1.0165884494781494, | |
| "learning_rate": 0.0001, | |
| "loss": 2.6799, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.013836477987421384, | |
| "grad_norm": 1.1318511962890625, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 2.7071, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.01509433962264151, | |
| "grad_norm": 1.0073941946029663, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 2.669, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.016352201257861635, | |
| "grad_norm": 0.9791858196258545, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 2.6599, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.01761006289308176, | |
| "grad_norm": 1.0005615949630737, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 2.7752, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.018867924528301886, | |
| "grad_norm": 0.9608151316642761, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 2.6789, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.02012578616352201, | |
| "grad_norm": 1.0178604125976562, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 2.6226, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.021383647798742137, | |
| "grad_norm": 0.9900417327880859, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 2.6544, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.022641509433962263, | |
| "grad_norm": 1.0411388874053955, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 2.8262, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.02389937106918239, | |
| "grad_norm": 1.0442683696746826, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 2.9196, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.025157232704402517, | |
| "grad_norm": 1.0308101177215576, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 2.7692, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.026415094339622643, | |
| "grad_norm": 1.0633000135421753, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 2.5357, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.027672955974842768, | |
| "grad_norm": 1.1763873100280762, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 2.706, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.028930817610062894, | |
| "grad_norm": 1.11016845703125, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 2.7779, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.03018867924528302, | |
| "grad_norm": 1.1306737661361694, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 2.6919, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.031446540880503145, | |
| "grad_norm": 1.1097077131271362, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 2.6992, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.03270440251572327, | |
| "grad_norm": 1.0761488676071167, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 2.61, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.033962264150943396, | |
| "grad_norm": 1.1367909908294678, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 2.6774, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.03522012578616352, | |
| "grad_norm": 1.1249269247055054, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 2.7028, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.03647798742138365, | |
| "grad_norm": 1.250557541847229, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 2.8417, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.03773584905660377, | |
| "grad_norm": 1.1810945272445679, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 2.7136, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.0389937106918239, | |
| "grad_norm": 1.18496572971344, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 2.7719, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.04025157232704402, | |
| "grad_norm": 1.229445457458496, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 2.6405, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.04150943396226415, | |
| "grad_norm": 1.1902095079421997, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 2.5579, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.042767295597484274, | |
| "grad_norm": 1.2509702444076538, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 2.6424, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.0440251572327044, | |
| "grad_norm": 1.359423279762268, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 2.6986, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.045283018867924525, | |
| "grad_norm": 1.287873387336731, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 2.5385, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.04654088050314465, | |
| "grad_norm": 1.3196231126785278, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 2.5856, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.04779874213836478, | |
| "grad_norm": 1.3013463020324707, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 2.4781, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.04905660377358491, | |
| "grad_norm": 1.345083475112915, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 2.5005, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.050314465408805034, | |
| "grad_norm": 1.314312219619751, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 2.463, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.05157232704402516, | |
| "grad_norm": 1.3623231649398804, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 2.6055, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.052830188679245285, | |
| "grad_norm": 1.5225359201431274, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 2.4176, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.05408805031446541, | |
| "grad_norm": 1.4751486778259277, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 2.378, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.055345911949685536, | |
| "grad_norm": 1.4943875074386597, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 2.36, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.05660377358490566, | |
| "grad_norm": 1.460222601890564, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 2.1474, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.05786163522012579, | |
| "grad_norm": 1.6100987195968628, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 2.4179, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.05911949685534591, | |
| "grad_norm": 1.765604853630066, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 2.327, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.06037735849056604, | |
| "grad_norm": 1.8875750303268433, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 2.3311, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.061635220125786164, | |
| "grad_norm": 2.4009406566619873, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 2.3198, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.06289308176100629, | |
| "grad_norm": 3.120755434036255, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 2.2676, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.06289308176100629, | |
| "eval_loss": 2.6739397048950195, | |
| "eval_runtime": 28.1676, | |
| "eval_samples_per_second": 47.537, | |
| "eval_steps_per_second": 11.893, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.06415094339622641, | |
| "grad_norm": 1.7542524337768555, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 2.24, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.06540880503144654, | |
| "grad_norm": 2.223454713821411, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 2.7143, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.06666666666666667, | |
| "grad_norm": 1.419851541519165, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 2.4013, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.06792452830188679, | |
| "grad_norm": 1.2127619981765747, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 2.6547, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.06918238993710692, | |
| "grad_norm": 0.9795339107513428, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 2.4785, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.07044025157232704, | |
| "grad_norm": 0.9852433800697327, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 2.5533, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.07169811320754717, | |
| "grad_norm": 0.975907027721405, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 2.7356, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.0729559748427673, | |
| "grad_norm": 0.93094402551651, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 2.5297, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.07421383647798742, | |
| "grad_norm": 0.9323387145996094, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 2.5081, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.07547169811320754, | |
| "grad_norm": 0.9696832299232483, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 2.6677, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.07672955974842767, | |
| "grad_norm": 0.9423898458480835, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 2.6656, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.0779874213836478, | |
| "grad_norm": 0.9334185123443604, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 2.5017, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.07924528301886792, | |
| "grad_norm": 0.9014331102371216, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 2.4658, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.08050314465408805, | |
| "grad_norm": 0.9938144087791443, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 2.497, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.08176100628930817, | |
| "grad_norm": 0.9624847769737244, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 2.6022, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.0830188679245283, | |
| "grad_norm": 0.9886726140975952, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 2.6842, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.08427672955974842, | |
| "grad_norm": 0.9895218014717102, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 2.6567, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.08553459119496855, | |
| "grad_norm": 1.0864946842193604, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 2.5935, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.08679245283018867, | |
| "grad_norm": 0.9743802547454834, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 2.5159, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.0880503144654088, | |
| "grad_norm": 0.9576653838157654, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 2.5324, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.08930817610062892, | |
| "grad_norm": 0.9808172583580017, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 2.6208, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.09056603773584905, | |
| "grad_norm": 1.0251048803329468, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 2.5564, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.09182389937106918, | |
| "grad_norm": 1.056553602218628, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 2.6373, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.0930817610062893, | |
| "grad_norm": 1.038964033126831, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 2.6226, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.09433962264150944, | |
| "grad_norm": 1.035335659980774, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 2.5811, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.09559748427672957, | |
| "grad_norm": 1.0560981035232544, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 2.5483, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.09685534591194969, | |
| "grad_norm": 1.0325617790222168, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 2.4451, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.09811320754716982, | |
| "grad_norm": 1.144875407218933, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 2.528, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.09937106918238994, | |
| "grad_norm": 1.1012463569641113, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 2.4928, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.10062893081761007, | |
| "grad_norm": 1.1191120147705078, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 2.6184, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.1018867924528302, | |
| "grad_norm": 1.1596407890319824, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 2.6848, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.10314465408805032, | |
| "grad_norm": 1.1232153177261353, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 2.4866, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.10440251572327044, | |
| "grad_norm": 1.1583571434020996, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 2.5794, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.10566037735849057, | |
| "grad_norm": 1.1157124042510986, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 2.3932, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.1069182389937107, | |
| "grad_norm": 1.1909675598144531, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 2.5739, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.10817610062893082, | |
| "grad_norm": 1.1610687971115112, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 2.4917, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.10943396226415095, | |
| "grad_norm": 1.3111274242401123, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 2.6641, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.11069182389937107, | |
| "grad_norm": 1.1908925771713257, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 2.565, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.1119496855345912, | |
| "grad_norm": 1.293189525604248, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 2.4645, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.11320754716981132, | |
| "grad_norm": 1.3886587619781494, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 2.615, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.11446540880503145, | |
| "grad_norm": 1.3037738800048828, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 2.4346, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.11572327044025157, | |
| "grad_norm": 1.3293190002441406, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 2.3868, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.1169811320754717, | |
| "grad_norm": 1.3965297937393188, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 2.4574, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.11823899371069183, | |
| "grad_norm": 1.4165204763412476, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 2.2459, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.11949685534591195, | |
| "grad_norm": 1.514732003211975, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 2.4935, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.12075471698113208, | |
| "grad_norm": 1.5237711668014526, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 2.4015, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.1220125786163522, | |
| "grad_norm": 1.6351544857025146, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 2.164, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.12327044025157233, | |
| "grad_norm": 1.7510908842086792, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 2.1507, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.12452830188679245, | |
| "grad_norm": 1.9573068618774414, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 1.9199, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.12578616352201258, | |
| "grad_norm": 2.961660385131836, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 2.381, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.12578616352201258, | |
| "eval_loss": 2.5099501609802246, | |
| "eval_runtime": 28.181, | |
| "eval_samples_per_second": 47.514, | |
| "eval_steps_per_second": 11.887, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.1270440251572327, | |
| "grad_norm": 1.1740059852600098, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 2.2193, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.12830188679245283, | |
| "grad_norm": 1.3514665365219116, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 2.343, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.12955974842767295, | |
| "grad_norm": 1.19695246219635, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 2.7012, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.13081761006289308, | |
| "grad_norm": 1.1020835638046265, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 2.4956, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.1320754716981132, | |
| "grad_norm": 1.0215140581130981, | |
| "learning_rate": 5e-05, | |
| "loss": 2.5141, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.13333333333333333, | |
| "grad_norm": 0.8773401379585266, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 2.441, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.13459119496855346, | |
| "grad_norm": 0.9116614460945129, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 2.3812, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.13584905660377358, | |
| "grad_norm": 0.9063419699668884, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 2.4601, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.1371069182389937, | |
| "grad_norm": 0.8941348791122437, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 2.4365, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.13836477987421383, | |
| "grad_norm": 0.9765293002128601, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 2.5013, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.13962264150943396, | |
| "grad_norm": 0.889190137386322, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 2.5586, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.14088050314465408, | |
| "grad_norm": 0.9321250319480896, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 2.5583, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.1421383647798742, | |
| "grad_norm": 0.9468171000480652, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 2.4434, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.14339622641509434, | |
| "grad_norm": 1.0092283487319946, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 2.6614, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.14465408805031446, | |
| "grad_norm": 1.027503490447998, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 2.752, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.1459119496855346, | |
| "grad_norm": 0.9917012453079224, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 2.6123, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.1471698113207547, | |
| "grad_norm": 1.0165959596633911, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 2.4123, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.14842767295597484, | |
| "grad_norm": 1.0201990604400635, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 2.5684, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.14968553459119496, | |
| "grad_norm": 0.9859070181846619, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 2.486, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.1509433962264151, | |
| "grad_norm": 0.9886462688446045, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 2.4565, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.15220125786163521, | |
| "grad_norm": 0.978788435459137, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 2.5479, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.15345911949685534, | |
| "grad_norm": 1.0328288078308105, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 2.5884, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.15471698113207547, | |
| "grad_norm": 1.0253801345825195, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 2.4175, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.1559748427672956, | |
| "grad_norm": 1.044564962387085, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 2.4906, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.15723270440251572, | |
| "grad_norm": 1.0834729671478271, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 2.5781, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.15849056603773584, | |
| "grad_norm": 1.0950733423233032, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 2.601, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.15974842767295597, | |
| "grad_norm": 1.0863381624221802, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 2.4629, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.1610062893081761, | |
| "grad_norm": 1.1020711660385132, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 2.417, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.16226415094339622, | |
| "grad_norm": 1.078052043914795, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 2.5443, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.16352201257861634, | |
| "grad_norm": 1.1078234910964966, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 2.5401, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.16477987421383647, | |
| "grad_norm": 1.119960069656372, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 2.3636, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.1660377358490566, | |
| "grad_norm": 1.1632513999938965, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 2.4013, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.16729559748427672, | |
| "grad_norm": 1.1259607076644897, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 2.4083, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.16855345911949685, | |
| "grad_norm": 1.2020472288131714, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 2.4778, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.16981132075471697, | |
| "grad_norm": 1.2250616550445557, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 2.5305, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.1710691823899371, | |
| "grad_norm": 1.1925936937332153, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 2.4507, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.17232704402515722, | |
| "grad_norm": 1.1623207330703735, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 2.3651, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.17358490566037735, | |
| "grad_norm": 1.2259721755981445, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 2.2519, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.17484276729559747, | |
| "grad_norm": 1.2976479530334473, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 2.5588, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.1761006289308176, | |
| "grad_norm": 1.241849660873413, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 2.2486, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.17735849056603772, | |
| "grad_norm": 1.3135217428207397, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 2.239, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.17861635220125785, | |
| "grad_norm": 1.3822765350341797, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 2.3502, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.17987421383647798, | |
| "grad_norm": 1.3332746028900146, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 2.1912, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.1811320754716981, | |
| "grad_norm": 1.4259880781173706, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 2.1525, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.18238993710691823, | |
| "grad_norm": 1.4124729633331299, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 2.1363, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.18364779874213835, | |
| "grad_norm": 1.5314245223999023, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 2.0689, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.18490566037735848, | |
| "grad_norm": 1.66856849193573, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 2.1705, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.1861635220125786, | |
| "grad_norm": 1.9275692701339722, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 2.1954, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.18742138364779873, | |
| "grad_norm": 2.003782033920288, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 1.9537, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.18867924528301888, | |
| "grad_norm": 2.8870654106140137, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 1.9617, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.18867924528301888, | |
| "eval_loss": 2.40899658203125, | |
| "eval_runtime": 28.1784, | |
| "eval_samples_per_second": 47.519, | |
| "eval_steps_per_second": 11.889, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.189937106918239, | |
| "grad_norm": 0.5950325131416321, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 1.9524, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.19119496855345913, | |
| "grad_norm": 0.8324146866798401, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 2.3964, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.19245283018867926, | |
| "grad_norm": 0.8736923933029175, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 2.5999, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.19371069182389938, | |
| "grad_norm": 0.8897731304168701, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 2.5278, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.1949685534591195, | |
| "grad_norm": 1.0104871988296509, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 2.4467, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.19622641509433963, | |
| "grad_norm": 0.9391211867332458, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 2.4912, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.19748427672955976, | |
| "grad_norm": 0.9052959084510803, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 2.4792, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.19874213836477989, | |
| "grad_norm": 0.9391781687736511, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 2.4833, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "grad_norm": 0.9394570589065552, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 2.598, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.20125786163522014, | |
| "grad_norm": 0.9689228534698486, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 2.5279, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.20251572327044026, | |
| "grad_norm": 0.9413164854049683, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 2.3978, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.2037735849056604, | |
| "grad_norm": 0.9433810114860535, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 2.4886, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.2050314465408805, | |
| "grad_norm": 1.0142652988433838, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 2.4882, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.20628930817610064, | |
| "grad_norm": 0.9539534449577332, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 2.6021, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.20754716981132076, | |
| "grad_norm": 0.972380518913269, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 2.5516, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.2088050314465409, | |
| "grad_norm": 0.9923897981643677, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 2.5301, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.21006289308176102, | |
| "grad_norm": 0.931374728679657, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 2.4134, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.21132075471698114, | |
| "grad_norm": 0.966039776802063, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 2.614, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.21257861635220127, | |
| "grad_norm": 1.0072507858276367, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 2.579, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.2138364779874214, | |
| "grad_norm": 1.0375057458877563, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 2.5472, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.21509433962264152, | |
| "grad_norm": 1.0220199823379517, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 2.5135, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.21635220125786164, | |
| "grad_norm": 0.9835240840911865, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 2.5222, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.21761006289308177, | |
| "grad_norm": 1.0312256813049316, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 2.5593, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.2188679245283019, | |
| "grad_norm": 1.0404565334320068, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 2.4988, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.22012578616352202, | |
| "grad_norm": 1.0345498323440552, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 2.5087, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.22138364779874214, | |
| "grad_norm": 1.079869270324707, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 2.6527, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.22264150943396227, | |
| "grad_norm": 1.1313115358352661, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 2.5757, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.2238993710691824, | |
| "grad_norm": 1.142509937286377, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 2.5999, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.22515723270440252, | |
| "grad_norm": 1.0770502090454102, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 2.5459, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.22641509433962265, | |
| "grad_norm": 1.1246929168701172, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 2.4993, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.22767295597484277, | |
| "grad_norm": 1.1188902854919434, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 2.4931, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.2289308176100629, | |
| "grad_norm": 1.1235661506652832, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 2.5169, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.23018867924528302, | |
| "grad_norm": 1.1644644737243652, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 2.5066, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.23144654088050315, | |
| "grad_norm": 1.2271788120269775, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 2.4485, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.23270440251572327, | |
| "grad_norm": 1.150336742401123, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 2.3984, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.2339622641509434, | |
| "grad_norm": 1.2218152284622192, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 2.5356, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.23522012578616353, | |
| "grad_norm": 1.2468903064727783, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 2.3837, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.23647798742138365, | |
| "grad_norm": 1.229461908340454, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 2.3331, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.23773584905660378, | |
| "grad_norm": 1.299018383026123, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 2.512, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.2389937106918239, | |
| "grad_norm": 1.2981739044189453, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 2.4421, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.24025157232704403, | |
| "grad_norm": 1.316055178642273, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 2.1508, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.24150943396226415, | |
| "grad_norm": 1.3228431940078735, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 2.3501, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.24276729559748428, | |
| "grad_norm": 1.3977642059326172, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 2.3785, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.2440251572327044, | |
| "grad_norm": 1.387685775756836, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 2.2292, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.24528301886792453, | |
| "grad_norm": 1.4316163063049316, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 2.1844, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.24654088050314465, | |
| "grad_norm": 1.569111943244934, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 2.2505, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.24779874213836478, | |
| "grad_norm": 1.5726760625839233, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 1.9243, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.2490566037735849, | |
| "grad_norm": 1.8344026803970337, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 2.0061, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.25031446540880503, | |
| "grad_norm": 2.0768260955810547, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 1.8445, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.25157232704402516, | |
| "grad_norm": 2.4009296894073486, | |
| "learning_rate": 0.0, | |
| "loss": 1.831, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.25157232704402516, | |
| "eval_loss": 2.39079213142395, | |
| "eval_runtime": 28.1716, | |
| "eval_samples_per_second": 47.53, | |
| "eval_steps_per_second": 11.891, | |
| "step": 200 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 200, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "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": 4.018876949987328e+16, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |