| { | |
| "best_metric": 0.15011541545391083, | |
| "best_model_checkpoint": "whisper-base-full-data-v2/checkpoint-60000", | |
| "epoch": 20.003133322888925, | |
| "global_step": 63840, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 5.8514, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 1.1194, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 6e-06, | |
| "loss": 0.6536, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 0.5617, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1e-05, | |
| "loss": 0.5055, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.2e-05, | |
| "loss": 0.4732, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.4000000000000001e-05, | |
| "loss": 0.451, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.6000000000000003e-05, | |
| "loss": 0.4286, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.8e-05, | |
| "loss": 0.4165, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 2e-05, | |
| "loss": 0.4006, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 2.2000000000000003e-05, | |
| "loss": 0.3873, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 2.4e-05, | |
| "loss": 0.3765, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 2.6000000000000002e-05, | |
| "loss": 0.3687, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 2.8000000000000003e-05, | |
| "loss": 0.3629, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 3e-05, | |
| "loss": 0.3514, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 3.2000000000000005e-05, | |
| "loss": 0.3483, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 3.4000000000000007e-05, | |
| "loss": 0.3438, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 3.6e-05, | |
| "loss": 0.342, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 3.8e-05, | |
| "loss": 0.3345, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 4e-05, | |
| "loss": 0.332, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 4.2e-05, | |
| "loss": 0.3234, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.4000000000000006e-05, | |
| "loss": 0.3212, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 4.600000000000001e-05, | |
| "loss": 0.3195, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 4.8e-05, | |
| "loss": 0.3251, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 5e-05, | |
| "loss": 0.3167, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 5.2000000000000004e-05, | |
| "loss": 0.3212, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 5.4000000000000005e-05, | |
| "loss": 0.3148, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 5.6000000000000006e-05, | |
| "loss": 0.3159, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 5.8e-05, | |
| "loss": 0.3127, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 6e-05, | |
| "loss": 0.3093, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 6.2e-05, | |
| "loss": 0.3088, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 6.400000000000001e-05, | |
| "loss": 0.309, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 6.6e-05, | |
| "loss": 0.3021, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 6.800000000000001e-05, | |
| "loss": 0.2986, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 7e-05, | |
| "loss": 0.2962, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 7.2e-05, | |
| "loss": 0.299, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 7.4e-05, | |
| "loss": 0.2978, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 7.6e-05, | |
| "loss": 0.3008, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 7.800000000000001e-05, | |
| "loss": 0.3001, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 8e-05, | |
| "loss": 0.2974, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 8.2e-05, | |
| "loss": 0.2993, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 8.4e-05, | |
| "loss": 0.2958, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 8.6e-05, | |
| "loss": 0.2959, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 8.800000000000001e-05, | |
| "loss": 0.2964, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 9e-05, | |
| "loss": 0.301, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 9.200000000000001e-05, | |
| "loss": 0.3022, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 9.4e-05, | |
| "loss": 0.3044, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 9.6e-05, | |
| "loss": 0.3003, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 9.8e-05, | |
| "loss": 0.2975, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 0.0001, | |
| "loss": 0.2986, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "eval_loss": 0.3998117744922638, | |
| "eval_runtime": 149.8792, | |
| "eval_samples_per_second": 106.359, | |
| "eval_steps_per_second": 1.668, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 9.983004758667574e-05, | |
| "loss": 0.298, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 9.966009517335146e-05, | |
| "loss": 0.2964, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 9.94901427600272e-05, | |
| "loss": 0.2929, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 9.932019034670292e-05, | |
| "loss": 0.2932, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 9.915023793337866e-05, | |
| "loss": 0.2899, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 9.89802855200544e-05, | |
| "loss": 0.2853, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 9.881033310673012e-05, | |
| "loss": 0.2818, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 9.864038069340585e-05, | |
| "loss": 0.2825, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 9.847042828008159e-05, | |
| "loss": 0.2807, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 9.830047586675731e-05, | |
| "loss": 0.2728, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 9.813052345343305e-05, | |
| "loss": 0.2767, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 9.796057104010877e-05, | |
| "loss": 0.2708, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 9.779061862678451e-05, | |
| "loss": 0.2672, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 9.762066621346023e-05, | |
| "loss": 0.2698, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 9.745071380013597e-05, | |
| "loss": 0.2579, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 9.72807613868117e-05, | |
| "loss": 0.253, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 9.711080897348744e-05, | |
| "loss": 0.2551, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 9.694085656016316e-05, | |
| "loss": 0.2522, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 9.67709041468389e-05, | |
| "loss": 0.2475, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 9.660095173351462e-05, | |
| "loss": 0.2515, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 9.643099932019036e-05, | |
| "loss": 0.2449, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 9.626104690686608e-05, | |
| "loss": 0.2508, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 9.609109449354181e-05, | |
| "loss": 0.244, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 9.592114208021754e-05, | |
| "loss": 0.2452, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 9.575118966689327e-05, | |
| "loss": 0.2403, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 9.5581237253569e-05, | |
| "loss": 0.2423, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 9.541128484024475e-05, | |
| "loss": 0.2409, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 9.524133242692047e-05, | |
| "loss": 0.2378, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 9.50713800135962e-05, | |
| "loss": 0.2345, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 9.490142760027193e-05, | |
| "loss": 0.2368, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 9.473147518694766e-05, | |
| "loss": 0.2366, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 9.456152277362339e-05, | |
| "loss": 0.2333, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 9.439157036029912e-05, | |
| "loss": 0.2277, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 9.422161794697484e-05, | |
| "loss": 0.232, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 9.405166553365058e-05, | |
| "loss": 0.2276, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 9.38817131203263e-05, | |
| "loss": 0.2296, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 9.371176070700204e-05, | |
| "loss": 0.2248, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 9.354180829367778e-05, | |
| "loss": 0.2251, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 9.337185588035351e-05, | |
| "loss": 0.2255, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 9.320190346702923e-05, | |
| "loss": 0.2272, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 9.303195105370497e-05, | |
| "loss": 0.2241, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 9.28619986403807e-05, | |
| "loss": 0.2237, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 9.269204622705643e-05, | |
| "loss": 0.221, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 9.252209381373215e-05, | |
| "loss": 0.2185, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 9.235214140040789e-05, | |
| "loss": 0.216, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 9.218218898708361e-05, | |
| "loss": 0.2148, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 9.201223657375935e-05, | |
| "loss": 0.2064, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 9.184228416043508e-05, | |
| "loss": 0.2058, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 9.167233174711082e-05, | |
| "loss": 0.2059, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 9.150237933378654e-05, | |
| "loss": 0.2037, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "eval_loss": 0.3050629198551178, | |
| "eval_runtime": 143.7999, | |
| "eval_samples_per_second": 110.855, | |
| "eval_steps_per_second": 1.739, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 9.133242692046228e-05, | |
| "loss": 0.2057, | |
| "step": 10100 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 9.1162474507138e-05, | |
| "loss": 0.2023, | |
| "step": 10200 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "learning_rate": 9.099252209381374e-05, | |
| "loss": 0.2054, | |
| "step": 10300 | |
| }, | |
| { | |
| "epoch": 3.26, | |
| "learning_rate": 9.082256968048946e-05, | |
| "loss": 0.2016, | |
| "step": 10400 | |
| }, | |
| { | |
| "epoch": 3.29, | |
| "learning_rate": 9.06526172671652e-05, | |
| "loss": 0.2028, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 3.32, | |
| "learning_rate": 9.048266485384093e-05, | |
| "loss": 0.2036, | |
| "step": 10600 | |
| }, | |
| { | |
| "epoch": 3.35, | |
| "learning_rate": 9.031271244051665e-05, | |
| "loss": 0.2044, | |
| "step": 10700 | |
| }, | |
| { | |
| "epoch": 3.38, | |
| "learning_rate": 9.014276002719239e-05, | |
| "loss": 0.202, | |
| "step": 10800 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "learning_rate": 8.997280761386813e-05, | |
| "loss": 0.2032, | |
| "step": 10900 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "learning_rate": 8.980285520054385e-05, | |
| "loss": 0.2012, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 3.48, | |
| "learning_rate": 8.963290278721959e-05, | |
| "loss": 0.1976, | |
| "step": 11100 | |
| }, | |
| { | |
| "epoch": 3.51, | |
| "learning_rate": 8.946295037389532e-05, | |
| "loss": 0.1978, | |
| "step": 11200 | |
| }, | |
| { | |
| "epoch": 3.54, | |
| "learning_rate": 8.929299796057104e-05, | |
| "loss": 0.1954, | |
| "step": 11300 | |
| }, | |
| { | |
| "epoch": 3.57, | |
| "learning_rate": 8.912304554724678e-05, | |
| "loss": 0.1943, | |
| "step": 11400 | |
| }, | |
| { | |
| "epoch": 3.6, | |
| "learning_rate": 8.89530931339225e-05, | |
| "loss": 0.1938, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 3.63, | |
| "learning_rate": 8.878314072059824e-05, | |
| "loss": 0.1963, | |
| "step": 11600 | |
| }, | |
| { | |
| "epoch": 3.67, | |
| "learning_rate": 8.861318830727396e-05, | |
| "loss": 0.1947, | |
| "step": 11700 | |
| }, | |
| { | |
| "epoch": 3.7, | |
| "learning_rate": 8.84432358939497e-05, | |
| "loss": 0.194, | |
| "step": 11800 | |
| }, | |
| { | |
| "epoch": 3.73, | |
| "learning_rate": 8.827328348062542e-05, | |
| "loss": 0.1915, | |
| "step": 11900 | |
| }, | |
| { | |
| "epoch": 3.76, | |
| "learning_rate": 8.810333106730117e-05, | |
| "loss": 0.1937, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 3.79, | |
| "learning_rate": 8.79333786539769e-05, | |
| "loss": 0.1919, | |
| "step": 12100 | |
| }, | |
| { | |
| "epoch": 3.82, | |
| "learning_rate": 8.776342624065263e-05, | |
| "loss": 0.1926, | |
| "step": 12200 | |
| }, | |
| { | |
| "epoch": 3.85, | |
| "learning_rate": 8.759347382732835e-05, | |
| "loss": 0.1931, | |
| "step": 12300 | |
| }, | |
| { | |
| "epoch": 3.89, | |
| "learning_rate": 8.742352141400409e-05, | |
| "loss": 0.1901, | |
| "step": 12400 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "learning_rate": 8.725356900067981e-05, | |
| "loss": 0.191, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 3.95, | |
| "learning_rate": 8.708361658735555e-05, | |
| "loss": 0.188, | |
| "step": 12600 | |
| }, | |
| { | |
| "epoch": 3.98, | |
| "learning_rate": 8.691366417403127e-05, | |
| "loss": 0.189, | |
| "step": 12700 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 8.6743711760707e-05, | |
| "loss": 0.1825, | |
| "step": 12800 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 8.657375934738273e-05, | |
| "loss": 0.1802, | |
| "step": 12900 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 8.640380693405848e-05, | |
| "loss": 0.1797, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 8.62338545207342e-05, | |
| "loss": 0.1778, | |
| "step": 13100 | |
| }, | |
| { | |
| "epoch": 4.14, | |
| "learning_rate": 8.606390210740994e-05, | |
| "loss": 0.1773, | |
| "step": 13200 | |
| }, | |
| { | |
| "epoch": 4.17, | |
| "learning_rate": 8.589394969408566e-05, | |
| "loss": 0.1781, | |
| "step": 13300 | |
| }, | |
| { | |
| "epoch": 4.2, | |
| "learning_rate": 8.57239972807614e-05, | |
| "loss": 0.1771, | |
| "step": 13400 | |
| }, | |
| { | |
| "epoch": 4.23, | |
| "learning_rate": 8.555404486743712e-05, | |
| "loss": 0.1765, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 4.26, | |
| "learning_rate": 8.538409245411285e-05, | |
| "loss": 0.1761, | |
| "step": 13600 | |
| }, | |
| { | |
| "epoch": 4.29, | |
| "learning_rate": 8.521414004078858e-05, | |
| "loss": 0.1759, | |
| "step": 13700 | |
| }, | |
| { | |
| "epoch": 4.32, | |
| "learning_rate": 8.504418762746431e-05, | |
| "loss": 0.1736, | |
| "step": 13800 | |
| }, | |
| { | |
| "epoch": 4.36, | |
| "learning_rate": 8.487423521414004e-05, | |
| "loss": 0.1765, | |
| "step": 13900 | |
| }, | |
| { | |
| "epoch": 4.39, | |
| "learning_rate": 8.470428280081577e-05, | |
| "loss": 0.1764, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 4.42, | |
| "learning_rate": 8.453433038749151e-05, | |
| "loss": 0.176, | |
| "step": 14100 | |
| }, | |
| { | |
| "epoch": 4.45, | |
| "learning_rate": 8.436437797416724e-05, | |
| "loss": 0.1726, | |
| "step": 14200 | |
| }, | |
| { | |
| "epoch": 4.48, | |
| "learning_rate": 8.419442556084297e-05, | |
| "loss": 0.1735, | |
| "step": 14300 | |
| }, | |
| { | |
| "epoch": 4.51, | |
| "learning_rate": 8.40244731475187e-05, | |
| "loss": 0.1749, | |
| "step": 14400 | |
| }, | |
| { | |
| "epoch": 4.54, | |
| "learning_rate": 8.385452073419443e-05, | |
| "loss": 0.1717, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 4.57, | |
| "learning_rate": 8.368456832087016e-05, | |
| "loss": 0.174, | |
| "step": 14600 | |
| }, | |
| { | |
| "epoch": 4.61, | |
| "learning_rate": 8.351461590754588e-05, | |
| "loss": 0.1744, | |
| "step": 14700 | |
| }, | |
| { | |
| "epoch": 4.64, | |
| "learning_rate": 8.334466349422162e-05, | |
| "loss": 0.1705, | |
| "step": 14800 | |
| }, | |
| { | |
| "epoch": 4.67, | |
| "learning_rate": 8.317471108089734e-05, | |
| "loss": 0.1703, | |
| "step": 14900 | |
| }, | |
| { | |
| "epoch": 4.7, | |
| "learning_rate": 8.300475866757308e-05, | |
| "loss": 0.1683, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 4.7, | |
| "eval_loss": 0.26461148262023926, | |
| "eval_runtime": 161.5353, | |
| "eval_samples_per_second": 98.684, | |
| "eval_steps_per_second": 1.548, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 4.73, | |
| "learning_rate": 8.28348062542488e-05, | |
| "loss": 0.1721, | |
| "step": 15100 | |
| }, | |
| { | |
| "epoch": 4.76, | |
| "learning_rate": 8.266485384092455e-05, | |
| "loss": 0.169, | |
| "step": 15200 | |
| }, | |
| { | |
| "epoch": 4.79, | |
| "learning_rate": 8.249490142760027e-05, | |
| "loss": 0.1716, | |
| "step": 15300 | |
| }, | |
| { | |
| "epoch": 4.83, | |
| "learning_rate": 8.232494901427601e-05, | |
| "loss": 0.1655, | |
| "step": 15400 | |
| }, | |
| { | |
| "epoch": 4.86, | |
| "learning_rate": 8.215499660095173e-05, | |
| "loss": 0.1709, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 4.89, | |
| "learning_rate": 8.198504418762747e-05, | |
| "loss": 0.1684, | |
| "step": 15600 | |
| }, | |
| { | |
| "epoch": 4.92, | |
| "learning_rate": 8.181509177430319e-05, | |
| "loss": 0.1694, | |
| "step": 15700 | |
| }, | |
| { | |
| "epoch": 4.95, | |
| "learning_rate": 8.164513936097893e-05, | |
| "loss": 0.1658, | |
| "step": 15800 | |
| }, | |
| { | |
| "epoch": 4.98, | |
| "learning_rate": 8.147518694765466e-05, | |
| "loss": 0.1672, | |
| "step": 15900 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 8.130523453433039e-05, | |
| "loss": 0.1628, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 5.04, | |
| "learning_rate": 8.113528212100612e-05, | |
| "loss": 0.1548, | |
| "step": 16100 | |
| }, | |
| { | |
| "epoch": 5.08, | |
| "learning_rate": 8.096532970768186e-05, | |
| "loss": 0.1581, | |
| "step": 16200 | |
| }, | |
| { | |
| "epoch": 5.11, | |
| "learning_rate": 8.079537729435758e-05, | |
| "loss": 0.1564, | |
| "step": 16300 | |
| }, | |
| { | |
| "epoch": 5.14, | |
| "learning_rate": 8.062542488103332e-05, | |
| "loss": 0.1568, | |
| "step": 16400 | |
| }, | |
| { | |
| "epoch": 5.17, | |
| "learning_rate": 8.045547246770904e-05, | |
| "loss": 0.1577, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 5.2, | |
| "learning_rate": 8.028552005438478e-05, | |
| "loss": 0.1562, | |
| "step": 16600 | |
| }, | |
| { | |
| "epoch": 5.23, | |
| "learning_rate": 8.011556764106051e-05, | |
| "loss": 0.1572, | |
| "step": 16700 | |
| }, | |
| { | |
| "epoch": 5.26, | |
| "learning_rate": 7.994561522773624e-05, | |
| "loss": 0.1584, | |
| "step": 16800 | |
| }, | |
| { | |
| "epoch": 5.3, | |
| "learning_rate": 7.977566281441197e-05, | |
| "loss": 0.1576, | |
| "step": 16900 | |
| }, | |
| { | |
| "epoch": 5.33, | |
| "learning_rate": 7.96057104010877e-05, | |
| "loss": 0.1615, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 5.36, | |
| "learning_rate": 7.943575798776343e-05, | |
| "loss": 0.1553, | |
| "step": 17100 | |
| }, | |
| { | |
| "epoch": 5.39, | |
| "learning_rate": 7.926580557443915e-05, | |
| "loss": 0.1564, | |
| "step": 17200 | |
| }, | |
| { | |
| "epoch": 5.42, | |
| "learning_rate": 7.90958531611149e-05, | |
| "loss": 0.1557, | |
| "step": 17300 | |
| }, | |
| { | |
| "epoch": 5.45, | |
| "learning_rate": 7.892590074779063e-05, | |
| "loss": 0.1559, | |
| "step": 17400 | |
| }, | |
| { | |
| "epoch": 5.48, | |
| "learning_rate": 7.875594833446636e-05, | |
| "loss": 0.1555, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 5.51, | |
| "learning_rate": 7.858599592114208e-05, | |
| "loss": 0.1552, | |
| "step": 17600 | |
| }, | |
| { | |
| "epoch": 5.55, | |
| "learning_rate": 7.841604350781782e-05, | |
| "loss": 0.1521, | |
| "step": 17700 | |
| }, | |
| { | |
| "epoch": 5.58, | |
| "learning_rate": 7.824609109449354e-05, | |
| "loss": 0.1539, | |
| "step": 17800 | |
| }, | |
| { | |
| "epoch": 5.61, | |
| "learning_rate": 7.807613868116928e-05, | |
| "loss": 0.1546, | |
| "step": 17900 | |
| }, | |
| { | |
| "epoch": 5.64, | |
| "learning_rate": 7.7906186267845e-05, | |
| "loss": 0.1535, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 5.67, | |
| "learning_rate": 7.773623385452074e-05, | |
| "loss": 0.1509, | |
| "step": 18100 | |
| }, | |
| { | |
| "epoch": 5.7, | |
| "learning_rate": 7.756628144119646e-05, | |
| "loss": 0.1541, | |
| "step": 18200 | |
| }, | |
| { | |
| "epoch": 5.73, | |
| "learning_rate": 7.73963290278722e-05, | |
| "loss": 0.1532, | |
| "step": 18300 | |
| }, | |
| { | |
| "epoch": 5.77, | |
| "learning_rate": 7.722637661454793e-05, | |
| "loss": 0.1522, | |
| "step": 18400 | |
| }, | |
| { | |
| "epoch": 5.8, | |
| "learning_rate": 7.705642420122367e-05, | |
| "loss": 0.1523, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 5.83, | |
| "learning_rate": 7.688647178789939e-05, | |
| "loss": 0.1552, | |
| "step": 18600 | |
| }, | |
| { | |
| "epoch": 5.86, | |
| "learning_rate": 7.671651937457513e-05, | |
| "loss": 0.1493, | |
| "step": 18700 | |
| }, | |
| { | |
| "epoch": 5.89, | |
| "learning_rate": 7.654656696125085e-05, | |
| "loss": 0.148, | |
| "step": 18800 | |
| }, | |
| { | |
| "epoch": 5.92, | |
| "learning_rate": 7.637661454792659e-05, | |
| "loss": 0.1521, | |
| "step": 18900 | |
| }, | |
| { | |
| "epoch": 5.95, | |
| "learning_rate": 7.620666213460231e-05, | |
| "loss": 0.1523, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 5.98, | |
| "learning_rate": 7.603670972127805e-05, | |
| "loss": 0.1507, | |
| "step": 19100 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "learning_rate": 7.586675730795377e-05, | |
| "loss": 0.1461, | |
| "step": 19200 | |
| }, | |
| { | |
| "epoch": 6.05, | |
| "learning_rate": 7.56968048946295e-05, | |
| "loss": 0.1415, | |
| "step": 19300 | |
| }, | |
| { | |
| "epoch": 6.08, | |
| "learning_rate": 7.552685248130524e-05, | |
| "loss": 0.1432, | |
| "step": 19400 | |
| }, | |
| { | |
| "epoch": 6.11, | |
| "learning_rate": 7.535690006798098e-05, | |
| "loss": 0.141, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 6.14, | |
| "learning_rate": 7.51869476546567e-05, | |
| "loss": 0.1399, | |
| "step": 19600 | |
| }, | |
| { | |
| "epoch": 6.17, | |
| "learning_rate": 7.501699524133244e-05, | |
| "loss": 0.1387, | |
| "step": 19700 | |
| }, | |
| { | |
| "epoch": 6.2, | |
| "learning_rate": 7.484704282800816e-05, | |
| "loss": 0.1434, | |
| "step": 19800 | |
| }, | |
| { | |
| "epoch": 6.24, | |
| "learning_rate": 7.46770904146839e-05, | |
| "loss": 0.1416, | |
| "step": 19900 | |
| }, | |
| { | |
| "epoch": 6.27, | |
| "learning_rate": 7.450713800135962e-05, | |
| "loss": 0.1426, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 6.27, | |
| "eval_loss": 0.2383899986743927, | |
| "eval_runtime": 149.4041, | |
| "eval_samples_per_second": 106.697, | |
| "eval_steps_per_second": 1.673, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 6.3, | |
| "learning_rate": 7.433718558803535e-05, | |
| "loss": 0.1388, | |
| "step": 20100 | |
| }, | |
| { | |
| "epoch": 6.33, | |
| "learning_rate": 7.416723317471108e-05, | |
| "loss": 0.1387, | |
| "step": 20200 | |
| }, | |
| { | |
| "epoch": 6.36, | |
| "learning_rate": 7.399728076138681e-05, | |
| "loss": 0.1436, | |
| "step": 20300 | |
| }, | |
| { | |
| "epoch": 6.39, | |
| "learning_rate": 7.382732834806254e-05, | |
| "loss": 0.1391, | |
| "step": 20400 | |
| }, | |
| { | |
| "epoch": 6.42, | |
| "learning_rate": 7.365737593473828e-05, | |
| "loss": 0.1395, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 6.45, | |
| "learning_rate": 7.348742352141401e-05, | |
| "loss": 0.1406, | |
| "step": 20600 | |
| }, | |
| { | |
| "epoch": 6.49, | |
| "learning_rate": 7.331747110808974e-05, | |
| "loss": 0.1396, | |
| "step": 20700 | |
| }, | |
| { | |
| "epoch": 6.52, | |
| "learning_rate": 7.314751869476547e-05, | |
| "loss": 0.1406, | |
| "step": 20800 | |
| }, | |
| { | |
| "epoch": 6.55, | |
| "learning_rate": 7.29775662814412e-05, | |
| "loss": 0.1414, | |
| "step": 20900 | |
| }, | |
| { | |
| "epoch": 6.58, | |
| "learning_rate": 7.280761386811693e-05, | |
| "loss": 0.14, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 6.61, | |
| "learning_rate": 7.263766145479266e-05, | |
| "loss": 0.137, | |
| "step": 21100 | |
| }, | |
| { | |
| "epoch": 6.64, | |
| "learning_rate": 7.246770904146838e-05, | |
| "loss": 0.1372, | |
| "step": 21200 | |
| }, | |
| { | |
| "epoch": 6.67, | |
| "learning_rate": 7.229775662814412e-05, | |
| "loss": 0.1424, | |
| "step": 21300 | |
| }, | |
| { | |
| "epoch": 6.71, | |
| "learning_rate": 7.212780421481986e-05, | |
| "loss": 0.1373, | |
| "step": 21400 | |
| }, | |
| { | |
| "epoch": 6.74, | |
| "learning_rate": 7.195785180149558e-05, | |
| "loss": 0.1377, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 6.77, | |
| "learning_rate": 7.178789938817131e-05, | |
| "loss": 0.1375, | |
| "step": 21600 | |
| }, | |
| { | |
| "epoch": 6.8, | |
| "learning_rate": 7.161794697484705e-05, | |
| "loss": 0.1372, | |
| "step": 21700 | |
| }, | |
| { | |
| "epoch": 6.83, | |
| "learning_rate": 7.144799456152277e-05, | |
| "loss": 0.1373, | |
| "step": 21800 | |
| }, | |
| { | |
| "epoch": 6.86, | |
| "learning_rate": 7.127804214819851e-05, | |
| "loss": 0.1406, | |
| "step": 21900 | |
| }, | |
| { | |
| "epoch": 6.89, | |
| "learning_rate": 7.110808973487423e-05, | |
| "loss": 0.1366, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 6.92, | |
| "learning_rate": 7.093813732154997e-05, | |
| "loss": 0.1365, | |
| "step": 22100 | |
| }, | |
| { | |
| "epoch": 6.96, | |
| "learning_rate": 7.07681849082257e-05, | |
| "loss": 0.1356, | |
| "step": 22200 | |
| }, | |
| { | |
| "epoch": 6.99, | |
| "learning_rate": 7.059823249490143e-05, | |
| "loss": 0.1406, | |
| "step": 22300 | |
| }, | |
| { | |
| "epoch": 7.02, | |
| "learning_rate": 7.042828008157716e-05, | |
| "loss": 0.132, | |
| "step": 22400 | |
| }, | |
| { | |
| "epoch": 7.05, | |
| "learning_rate": 7.025832766825289e-05, | |
| "loss": 0.127, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 7.08, | |
| "learning_rate": 7.008837525492862e-05, | |
| "loss": 0.1289, | |
| "step": 22600 | |
| }, | |
| { | |
| "epoch": 7.11, | |
| "learning_rate": 6.991842284160436e-05, | |
| "loss": 0.1254, | |
| "step": 22700 | |
| }, | |
| { | |
| "epoch": 7.14, | |
| "learning_rate": 6.97484704282801e-05, | |
| "loss": 0.1286, | |
| "step": 22800 | |
| }, | |
| { | |
| "epoch": 7.18, | |
| "learning_rate": 6.957851801495582e-05, | |
| "loss": 0.1308, | |
| "step": 22900 | |
| }, | |
| { | |
| "epoch": 7.21, | |
| "learning_rate": 6.940856560163155e-05, | |
| "loss": 0.1252, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 7.24, | |
| "learning_rate": 6.923861318830728e-05, | |
| "loss": 0.1263, | |
| "step": 23100 | |
| }, | |
| { | |
| "epoch": 7.27, | |
| "learning_rate": 6.906866077498301e-05, | |
| "loss": 0.1287, | |
| "step": 23200 | |
| }, | |
| { | |
| "epoch": 7.3, | |
| "learning_rate": 6.889870836165873e-05, | |
| "loss": 0.129, | |
| "step": 23300 | |
| }, | |
| { | |
| "epoch": 7.33, | |
| "learning_rate": 6.872875594833447e-05, | |
| "loss": 0.1271, | |
| "step": 23400 | |
| }, | |
| { | |
| "epoch": 7.36, | |
| "learning_rate": 6.85588035350102e-05, | |
| "loss": 0.1258, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 7.39, | |
| "learning_rate": 6.838885112168593e-05, | |
| "loss": 0.1293, | |
| "step": 23600 | |
| }, | |
| { | |
| "epoch": 7.43, | |
| "learning_rate": 6.821889870836167e-05, | |
| "loss": 0.1268, | |
| "step": 23700 | |
| }, | |
| { | |
| "epoch": 7.46, | |
| "learning_rate": 6.80489462950374e-05, | |
| "loss": 0.1285, | |
| "step": 23800 | |
| }, | |
| { | |
| "epoch": 7.49, | |
| "learning_rate": 6.787899388171312e-05, | |
| "loss": 0.1263, | |
| "step": 23900 | |
| }, | |
| { | |
| "epoch": 7.52, | |
| "learning_rate": 6.770904146838886e-05, | |
| "loss": 0.1267, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 7.55, | |
| "learning_rate": 6.753908905506458e-05, | |
| "loss": 0.126, | |
| "step": 24100 | |
| }, | |
| { | |
| "epoch": 7.58, | |
| "learning_rate": 6.736913664174032e-05, | |
| "loss": 0.127, | |
| "step": 24200 | |
| }, | |
| { | |
| "epoch": 7.61, | |
| "learning_rate": 6.719918422841604e-05, | |
| "loss": 0.1267, | |
| "step": 24300 | |
| }, | |
| { | |
| "epoch": 7.65, | |
| "learning_rate": 6.702923181509178e-05, | |
| "loss": 0.1267, | |
| "step": 24400 | |
| }, | |
| { | |
| "epoch": 7.68, | |
| "learning_rate": 6.68592794017675e-05, | |
| "loss": 0.1267, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 7.71, | |
| "learning_rate": 6.668932698844324e-05, | |
| "loss": 0.1257, | |
| "step": 24600 | |
| }, | |
| { | |
| "epoch": 7.74, | |
| "learning_rate": 6.651937457511896e-05, | |
| "loss": 0.1269, | |
| "step": 24700 | |
| }, | |
| { | |
| "epoch": 7.77, | |
| "learning_rate": 6.634942216179471e-05, | |
| "loss": 0.1273, | |
| "step": 24800 | |
| }, | |
| { | |
| "epoch": 7.8, | |
| "learning_rate": 6.617946974847043e-05, | |
| "loss": 0.1273, | |
| "step": 24900 | |
| }, | |
| { | |
| "epoch": 7.83, | |
| "learning_rate": 6.600951733514617e-05, | |
| "loss": 0.1265, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 7.83, | |
| "eval_loss": 0.21857744455337524, | |
| "eval_runtime": 151.4029, | |
| "eval_samples_per_second": 105.289, | |
| "eval_steps_per_second": 1.651, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 7.86, | |
| "learning_rate": 6.583956492182189e-05, | |
| "loss": 0.1278, | |
| "step": 25100 | |
| }, | |
| { | |
| "epoch": 7.9, | |
| "learning_rate": 6.566961250849763e-05, | |
| "loss": 0.1266, | |
| "step": 25200 | |
| }, | |
| { | |
| "epoch": 7.93, | |
| "learning_rate": 6.549966009517335e-05, | |
| "loss": 0.1271, | |
| "step": 25300 | |
| }, | |
| { | |
| "epoch": 7.96, | |
| "learning_rate": 6.532970768184909e-05, | |
| "loss": 0.1258, | |
| "step": 25400 | |
| }, | |
| { | |
| "epoch": 7.99, | |
| "learning_rate": 6.515975526852481e-05, | |
| "loss": 0.1233, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 8.02, | |
| "learning_rate": 6.498980285520054e-05, | |
| "loss": 0.1189, | |
| "step": 25600 | |
| }, | |
| { | |
| "epoch": 8.05, | |
| "learning_rate": 6.481985044187627e-05, | |
| "loss": 0.1172, | |
| "step": 25700 | |
| }, | |
| { | |
| "epoch": 8.08, | |
| "learning_rate": 6.464989802855202e-05, | |
| "loss": 0.1175, | |
| "step": 25800 | |
| }, | |
| { | |
| "epoch": 8.12, | |
| "learning_rate": 6.447994561522774e-05, | |
| "loss": 0.1178, | |
| "step": 25900 | |
| }, | |
| { | |
| "epoch": 8.15, | |
| "learning_rate": 6.430999320190348e-05, | |
| "loss": 0.1159, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 8.18, | |
| "learning_rate": 6.41400407885792e-05, | |
| "loss": 0.1145, | |
| "step": 26100 | |
| }, | |
| { | |
| "epoch": 8.21, | |
| "learning_rate": 6.397008837525493e-05, | |
| "loss": 0.1172, | |
| "step": 26200 | |
| }, | |
| { | |
| "epoch": 8.24, | |
| "learning_rate": 6.380013596193066e-05, | |
| "loss": 0.1157, | |
| "step": 26300 | |
| }, | |
| { | |
| "epoch": 8.27, | |
| "learning_rate": 6.36301835486064e-05, | |
| "loss": 0.1165, | |
| "step": 26400 | |
| }, | |
| { | |
| "epoch": 8.3, | |
| "learning_rate": 6.346023113528212e-05, | |
| "loss": 0.1178, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 8.33, | |
| "learning_rate": 6.329027872195785e-05, | |
| "loss": 0.1181, | |
| "step": 26600 | |
| }, | |
| { | |
| "epoch": 8.37, | |
| "learning_rate": 6.312032630863358e-05, | |
| "loss": 0.1142, | |
| "step": 26700 | |
| }, | |
| { | |
| "epoch": 8.4, | |
| "learning_rate": 6.295037389530931e-05, | |
| "loss": 0.1158, | |
| "step": 26800 | |
| }, | |
| { | |
| "epoch": 8.43, | |
| "learning_rate": 6.278042148198505e-05, | |
| "loss": 0.114, | |
| "step": 26900 | |
| }, | |
| { | |
| "epoch": 8.46, | |
| "learning_rate": 6.261046906866078e-05, | |
| "loss": 0.1176, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 8.49, | |
| "learning_rate": 6.24405166553365e-05, | |
| "loss": 0.1149, | |
| "step": 27100 | |
| }, | |
| { | |
| "epoch": 8.52, | |
| "learning_rate": 6.227056424201224e-05, | |
| "loss": 0.1172, | |
| "step": 27200 | |
| }, | |
| { | |
| "epoch": 8.55, | |
| "learning_rate": 6.210061182868797e-05, | |
| "loss": 0.1149, | |
| "step": 27300 | |
| }, | |
| { | |
| "epoch": 8.59, | |
| "learning_rate": 6.19306594153637e-05, | |
| "loss": 0.116, | |
| "step": 27400 | |
| }, | |
| { | |
| "epoch": 8.62, | |
| "learning_rate": 6.176070700203944e-05, | |
| "loss": 0.1162, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 8.65, | |
| "learning_rate": 6.159075458871516e-05, | |
| "loss": 0.1132, | |
| "step": 27600 | |
| }, | |
| { | |
| "epoch": 8.68, | |
| "learning_rate": 6.14208021753909e-05, | |
| "loss": 0.1136, | |
| "step": 27700 | |
| }, | |
| { | |
| "epoch": 8.71, | |
| "learning_rate": 6.125084976206662e-05, | |
| "loss": 0.114, | |
| "step": 27800 | |
| }, | |
| { | |
| "epoch": 8.74, | |
| "learning_rate": 6.108089734874235e-05, | |
| "loss": 0.1139, | |
| "step": 27900 | |
| }, | |
| { | |
| "epoch": 8.77, | |
| "learning_rate": 6.091094493541809e-05, | |
| "loss": 0.1158, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 8.8, | |
| "learning_rate": 6.074099252209382e-05, | |
| "loss": 0.1137, | |
| "step": 28100 | |
| }, | |
| { | |
| "epoch": 8.84, | |
| "learning_rate": 6.057104010876955e-05, | |
| "loss": 0.1138, | |
| "step": 28200 | |
| }, | |
| { | |
| "epoch": 8.87, | |
| "learning_rate": 6.040108769544528e-05, | |
| "loss": 0.1151, | |
| "step": 28300 | |
| }, | |
| { | |
| "epoch": 8.9, | |
| "learning_rate": 6.023113528212101e-05, | |
| "loss": 0.1127, | |
| "step": 28400 | |
| }, | |
| { | |
| "epoch": 8.93, | |
| "learning_rate": 6.006118286879674e-05, | |
| "loss": 0.1152, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 8.96, | |
| "learning_rate": 5.989123045547247e-05, | |
| "loss": 0.1138, | |
| "step": 28600 | |
| }, | |
| { | |
| "epoch": 8.99, | |
| "learning_rate": 5.97212780421482e-05, | |
| "loss": 0.115, | |
| "step": 28700 | |
| }, | |
| { | |
| "epoch": 9.02, | |
| "learning_rate": 5.9551325628823926e-05, | |
| "loss": 0.1093, | |
| "step": 28800 | |
| }, | |
| { | |
| "epoch": 9.06, | |
| "learning_rate": 5.9381373215499656e-05, | |
| "loss": 0.1058, | |
| "step": 28900 | |
| }, | |
| { | |
| "epoch": 9.09, | |
| "learning_rate": 5.92114208021754e-05, | |
| "loss": 0.1068, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 9.12, | |
| "learning_rate": 5.904146838885113e-05, | |
| "loss": 0.1055, | |
| "step": 29100 | |
| }, | |
| { | |
| "epoch": 9.15, | |
| "learning_rate": 5.887151597552686e-05, | |
| "loss": 0.1074, | |
| "step": 29200 | |
| }, | |
| { | |
| "epoch": 9.18, | |
| "learning_rate": 5.870156356220259e-05, | |
| "loss": 0.1059, | |
| "step": 29300 | |
| }, | |
| { | |
| "epoch": 9.21, | |
| "learning_rate": 5.8531611148878316e-05, | |
| "loss": 0.1041, | |
| "step": 29400 | |
| }, | |
| { | |
| "epoch": 9.24, | |
| "learning_rate": 5.8361658735554046e-05, | |
| "loss": 0.1086, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 9.27, | |
| "learning_rate": 5.8191706322229775e-05, | |
| "loss": 0.1094, | |
| "step": 29600 | |
| }, | |
| { | |
| "epoch": 9.31, | |
| "learning_rate": 5.8021753908905504e-05, | |
| "loss": 0.106, | |
| "step": 29700 | |
| }, | |
| { | |
| "epoch": 9.34, | |
| "learning_rate": 5.7851801495581234e-05, | |
| "loss": 0.1086, | |
| "step": 29800 | |
| }, | |
| { | |
| "epoch": 9.37, | |
| "learning_rate": 5.768184908225697e-05, | |
| "loss": 0.1093, | |
| "step": 29900 | |
| }, | |
| { | |
| "epoch": 9.4, | |
| "learning_rate": 5.75118966689327e-05, | |
| "loss": 0.1043, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 9.4, | |
| "eval_loss": 0.20128852128982544, | |
| "eval_runtime": 146.1611, | |
| "eval_samples_per_second": 109.065, | |
| "eval_steps_per_second": 1.71, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 9.43, | |
| "learning_rate": 5.7341944255608436e-05, | |
| "loss": 0.1058, | |
| "step": 30100 | |
| }, | |
| { | |
| "epoch": 9.46, | |
| "learning_rate": 5.7171991842284165e-05, | |
| "loss": 0.1064, | |
| "step": 30200 | |
| }, | |
| { | |
| "epoch": 9.49, | |
| "learning_rate": 5.7002039428959894e-05, | |
| "loss": 0.1066, | |
| "step": 30300 | |
| }, | |
| { | |
| "epoch": 9.53, | |
| "learning_rate": 5.6832087015635624e-05, | |
| "loss": 0.1042, | |
| "step": 30400 | |
| }, | |
| { | |
| "epoch": 9.56, | |
| "learning_rate": 5.666213460231135e-05, | |
| "loss": 0.1046, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 9.59, | |
| "learning_rate": 5.649218218898709e-05, | |
| "loss": 0.1047, | |
| "step": 30600 | |
| }, | |
| { | |
| "epoch": 9.62, | |
| "learning_rate": 5.632222977566282e-05, | |
| "loss": 0.1045, | |
| "step": 30700 | |
| }, | |
| { | |
| "epoch": 9.65, | |
| "learning_rate": 5.615227736233855e-05, | |
| "loss": 0.1052, | |
| "step": 30800 | |
| }, | |
| { | |
| "epoch": 9.68, | |
| "learning_rate": 5.598232494901428e-05, | |
| "loss": 0.1083, | |
| "step": 30900 | |
| }, | |
| { | |
| "epoch": 9.71, | |
| "learning_rate": 5.581237253569001e-05, | |
| "loss": 0.1071, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 9.74, | |
| "learning_rate": 5.5642420122365736e-05, | |
| "loss": 0.1041, | |
| "step": 31100 | |
| }, | |
| { | |
| "epoch": 9.78, | |
| "learning_rate": 5.547246770904147e-05, | |
| "loss": 0.1039, | |
| "step": 31200 | |
| }, | |
| { | |
| "epoch": 9.81, | |
| "learning_rate": 5.530251529571721e-05, | |
| "loss": 0.1046, | |
| "step": 31300 | |
| }, | |
| { | |
| "epoch": 9.84, | |
| "learning_rate": 5.513256288239294e-05, | |
| "loss": 0.104, | |
| "step": 31400 | |
| }, | |
| { | |
| "epoch": 9.87, | |
| "learning_rate": 5.496261046906867e-05, | |
| "loss": 0.1052, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 9.9, | |
| "learning_rate": 5.47926580557444e-05, | |
| "loss": 0.1033, | |
| "step": 31600 | |
| }, | |
| { | |
| "epoch": 9.93, | |
| "learning_rate": 5.4622705642420126e-05, | |
| "loss": 0.1048, | |
| "step": 31700 | |
| }, | |
| { | |
| "epoch": 9.96, | |
| "learning_rate": 5.4452753229095856e-05, | |
| "loss": 0.1041, | |
| "step": 31800 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "learning_rate": 5.4282800815771585e-05, | |
| "loss": 0.1036, | |
| "step": 31900 | |
| }, | |
| { | |
| "epoch": 10.03, | |
| "learning_rate": 5.4112848402447314e-05, | |
| "loss": 0.0968, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 10.06, | |
| "learning_rate": 5.3942895989123044e-05, | |
| "loss": 0.0981, | |
| "step": 32100 | |
| }, | |
| { | |
| "epoch": 10.09, | |
| "learning_rate": 5.377294357579879e-05, | |
| "loss": 0.0979, | |
| "step": 32200 | |
| }, | |
| { | |
| "epoch": 10.12, | |
| "learning_rate": 5.3602991162474516e-05, | |
| "loss": 0.0956, | |
| "step": 32300 | |
| }, | |
| { | |
| "epoch": 10.15, | |
| "learning_rate": 5.3433038749150246e-05, | |
| "loss": 0.0972, | |
| "step": 32400 | |
| }, | |
| { | |
| "epoch": 10.18, | |
| "learning_rate": 5.3263086335825975e-05, | |
| "loss": 0.0979, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 10.21, | |
| "learning_rate": 5.3093133922501704e-05, | |
| "loss": 0.096, | |
| "step": 32600 | |
| }, | |
| { | |
| "epoch": 10.25, | |
| "learning_rate": 5.2923181509177434e-05, | |
| "loss": 0.0964, | |
| "step": 32700 | |
| }, | |
| { | |
| "epoch": 10.28, | |
| "learning_rate": 5.275322909585316e-05, | |
| "loss": 0.0973, | |
| "step": 32800 | |
| }, | |
| { | |
| "epoch": 10.31, | |
| "learning_rate": 5.258327668252889e-05, | |
| "loss": 0.0955, | |
| "step": 32900 | |
| }, | |
| { | |
| "epoch": 10.34, | |
| "learning_rate": 5.241332426920462e-05, | |
| "loss": 0.1002, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 10.37, | |
| "learning_rate": 5.224337185588035e-05, | |
| "loss": 0.096, | |
| "step": 33100 | |
| }, | |
| { | |
| "epoch": 10.4, | |
| "learning_rate": 5.207341944255608e-05, | |
| "loss": 0.0964, | |
| "step": 33200 | |
| }, | |
| { | |
| "epoch": 10.43, | |
| "learning_rate": 5.1903467029231824e-05, | |
| "loss": 0.0975, | |
| "step": 33300 | |
| }, | |
| { | |
| "epoch": 10.47, | |
| "learning_rate": 5.173351461590755e-05, | |
| "loss": 0.0961, | |
| "step": 33400 | |
| }, | |
| { | |
| "epoch": 10.5, | |
| "learning_rate": 5.156356220258328e-05, | |
| "loss": 0.0987, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 10.53, | |
| "learning_rate": 5.139360978925901e-05, | |
| "loss": 0.0965, | |
| "step": 33600 | |
| }, | |
| { | |
| "epoch": 10.56, | |
| "learning_rate": 5.122365737593474e-05, | |
| "loss": 0.0969, | |
| "step": 33700 | |
| }, | |
| { | |
| "epoch": 10.59, | |
| "learning_rate": 5.105370496261047e-05, | |
| "loss": 0.0936, | |
| "step": 33800 | |
| }, | |
| { | |
| "epoch": 10.62, | |
| "learning_rate": 5.08837525492862e-05, | |
| "loss": 0.0955, | |
| "step": 33900 | |
| }, | |
| { | |
| "epoch": 10.65, | |
| "learning_rate": 5.071380013596193e-05, | |
| "loss": 0.0985, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 10.68, | |
| "learning_rate": 5.054384772263766e-05, | |
| "loss": 0.0948, | |
| "step": 34100 | |
| }, | |
| { | |
| "epoch": 10.72, | |
| "learning_rate": 5.037389530931339e-05, | |
| "loss": 0.0979, | |
| "step": 34200 | |
| }, | |
| { | |
| "epoch": 10.75, | |
| "learning_rate": 5.020394289598912e-05, | |
| "loss": 0.0954, | |
| "step": 34300 | |
| }, | |
| { | |
| "epoch": 10.78, | |
| "learning_rate": 5.003399048266486e-05, | |
| "loss": 0.0965, | |
| "step": 34400 | |
| }, | |
| { | |
| "epoch": 10.81, | |
| "learning_rate": 4.986403806934058e-05, | |
| "loss": 0.0942, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 10.84, | |
| "learning_rate": 4.969408565601632e-05, | |
| "loss": 0.0961, | |
| "step": 34600 | |
| }, | |
| { | |
| "epoch": 10.87, | |
| "learning_rate": 4.952413324269205e-05, | |
| "loss": 0.0963, | |
| "step": 34700 | |
| }, | |
| { | |
| "epoch": 10.9, | |
| "learning_rate": 4.935418082936778e-05, | |
| "loss": 0.0955, | |
| "step": 34800 | |
| }, | |
| { | |
| "epoch": 10.94, | |
| "learning_rate": 4.918422841604351e-05, | |
| "loss": 0.095, | |
| "step": 34900 | |
| }, | |
| { | |
| "epoch": 10.97, | |
| "learning_rate": 4.901427600271924e-05, | |
| "loss": 0.0971, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 10.97, | |
| "eval_loss": 0.18939317762851715, | |
| "eval_runtime": 145.9174, | |
| "eval_samples_per_second": 109.247, | |
| "eval_steps_per_second": 1.713, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "learning_rate": 4.8844323589394966e-05, | |
| "loss": 0.0965, | |
| "step": 35100 | |
| }, | |
| { | |
| "epoch": 11.03, | |
| "learning_rate": 4.86743711760707e-05, | |
| "loss": 0.0898, | |
| "step": 35200 | |
| }, | |
| { | |
| "epoch": 11.06, | |
| "learning_rate": 4.850441876274643e-05, | |
| "loss": 0.087, | |
| "step": 35300 | |
| }, | |
| { | |
| "epoch": 11.09, | |
| "learning_rate": 4.833446634942216e-05, | |
| "loss": 0.0893, | |
| "step": 35400 | |
| }, | |
| { | |
| "epoch": 11.12, | |
| "learning_rate": 4.816451393609789e-05, | |
| "loss": 0.0892, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 11.15, | |
| "learning_rate": 4.799456152277362e-05, | |
| "loss": 0.0873, | |
| "step": 35600 | |
| }, | |
| { | |
| "epoch": 11.19, | |
| "learning_rate": 4.7824609109449356e-05, | |
| "loss": 0.0892, | |
| "step": 35700 | |
| }, | |
| { | |
| "epoch": 11.22, | |
| "learning_rate": 4.7654656696125086e-05, | |
| "loss": 0.0884, | |
| "step": 35800 | |
| }, | |
| { | |
| "epoch": 11.25, | |
| "learning_rate": 4.7484704282800815e-05, | |
| "loss": 0.0887, | |
| "step": 35900 | |
| }, | |
| { | |
| "epoch": 11.28, | |
| "learning_rate": 4.731475186947655e-05, | |
| "loss": 0.0874, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 11.31, | |
| "learning_rate": 4.714479945615228e-05, | |
| "loss": 0.0893, | |
| "step": 36100 | |
| }, | |
| { | |
| "epoch": 11.34, | |
| "learning_rate": 4.697484704282801e-05, | |
| "loss": 0.0884, | |
| "step": 36200 | |
| }, | |
| { | |
| "epoch": 11.37, | |
| "learning_rate": 4.680489462950374e-05, | |
| "loss": 0.0885, | |
| "step": 36300 | |
| }, | |
| { | |
| "epoch": 11.41, | |
| "learning_rate": 4.6634942216179476e-05, | |
| "loss": 0.0892, | |
| "step": 36400 | |
| }, | |
| { | |
| "epoch": 11.44, | |
| "learning_rate": 4.6464989802855205e-05, | |
| "loss": 0.088, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 11.47, | |
| "learning_rate": 4.6295037389530934e-05, | |
| "loss": 0.0886, | |
| "step": 36600 | |
| }, | |
| { | |
| "epoch": 11.5, | |
| "learning_rate": 4.6125084976206664e-05, | |
| "loss": 0.0878, | |
| "step": 36700 | |
| }, | |
| { | |
| "epoch": 11.53, | |
| "learning_rate": 4.59551325628824e-05, | |
| "loss": 0.0896, | |
| "step": 36800 | |
| }, | |
| { | |
| "epoch": 11.56, | |
| "learning_rate": 4.578518014955813e-05, | |
| "loss": 0.0895, | |
| "step": 36900 | |
| }, | |
| { | |
| "epoch": 11.59, | |
| "learning_rate": 4.561522773623386e-05, | |
| "loss": 0.086, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 11.62, | |
| "learning_rate": 4.544527532290959e-05, | |
| "loss": 0.0918, | |
| "step": 37100 | |
| }, | |
| { | |
| "epoch": 11.66, | |
| "learning_rate": 4.527532290958532e-05, | |
| "loss": 0.0874, | |
| "step": 37200 | |
| }, | |
| { | |
| "epoch": 11.69, | |
| "learning_rate": 4.5105370496261054e-05, | |
| "loss": 0.0878, | |
| "step": 37300 | |
| }, | |
| { | |
| "epoch": 11.72, | |
| "learning_rate": 4.493541808293678e-05, | |
| "loss": 0.0873, | |
| "step": 37400 | |
| }, | |
| { | |
| "epoch": 11.75, | |
| "learning_rate": 4.476546566961251e-05, | |
| "loss": 0.0884, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 11.78, | |
| "learning_rate": 4.459551325628824e-05, | |
| "loss": 0.0878, | |
| "step": 37600 | |
| }, | |
| { | |
| "epoch": 11.81, | |
| "learning_rate": 4.442556084296397e-05, | |
| "loss": 0.0879, | |
| "step": 37700 | |
| }, | |
| { | |
| "epoch": 11.84, | |
| "learning_rate": 4.425560842963971e-05, | |
| "loss": 0.0866, | |
| "step": 37800 | |
| }, | |
| { | |
| "epoch": 11.88, | |
| "learning_rate": 4.408565601631544e-05, | |
| "loss": 0.0869, | |
| "step": 37900 | |
| }, | |
| { | |
| "epoch": 11.91, | |
| "learning_rate": 4.3915703602991166e-05, | |
| "loss": 0.088, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 11.94, | |
| "learning_rate": 4.3745751189666896e-05, | |
| "loss": 0.0865, | |
| "step": 38100 | |
| }, | |
| { | |
| "epoch": 11.97, | |
| "learning_rate": 4.3575798776342625e-05, | |
| "loss": 0.0903, | |
| "step": 38200 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "learning_rate": 4.3405846363018354e-05, | |
| "loss": 0.0864, | |
| "step": 38300 | |
| }, | |
| { | |
| "epoch": 12.03, | |
| "learning_rate": 4.323589394969409e-05, | |
| "loss": 0.0804, | |
| "step": 38400 | |
| }, | |
| { | |
| "epoch": 12.06, | |
| "learning_rate": 4.306594153636982e-05, | |
| "loss": 0.0799, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 12.09, | |
| "learning_rate": 4.289598912304555e-05, | |
| "loss": 0.0815, | |
| "step": 38600 | |
| }, | |
| { | |
| "epoch": 12.13, | |
| "learning_rate": 4.272603670972128e-05, | |
| "loss": 0.0831, | |
| "step": 38700 | |
| }, | |
| { | |
| "epoch": 12.16, | |
| "learning_rate": 4.255608429639701e-05, | |
| "loss": 0.083, | |
| "step": 38800 | |
| }, | |
| { | |
| "epoch": 12.19, | |
| "learning_rate": 4.2386131883072744e-05, | |
| "loss": 0.0797, | |
| "step": 38900 | |
| }, | |
| { | |
| "epoch": 12.22, | |
| "learning_rate": 4.2216179469748474e-05, | |
| "loss": 0.0819, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 12.25, | |
| "learning_rate": 4.20462270564242e-05, | |
| "loss": 0.0822, | |
| "step": 39100 | |
| }, | |
| { | |
| "epoch": 12.28, | |
| "learning_rate": 4.187627464309993e-05, | |
| "loss": 0.0812, | |
| "step": 39200 | |
| }, | |
| { | |
| "epoch": 12.31, | |
| "learning_rate": 4.170632222977566e-05, | |
| "loss": 0.0819, | |
| "step": 39300 | |
| }, | |
| { | |
| "epoch": 12.35, | |
| "learning_rate": 4.15363698164514e-05, | |
| "loss": 0.0795, | |
| "step": 39400 | |
| }, | |
| { | |
| "epoch": 12.38, | |
| "learning_rate": 4.136641740312713e-05, | |
| "loss": 0.0783, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 12.41, | |
| "learning_rate": 4.119646498980286e-05, | |
| "loss": 0.081, | |
| "step": 39600 | |
| }, | |
| { | |
| "epoch": 12.44, | |
| "learning_rate": 4.1026512576478586e-05, | |
| "loss": 0.0811, | |
| "step": 39700 | |
| }, | |
| { | |
| "epoch": 12.47, | |
| "learning_rate": 4.0856560163154316e-05, | |
| "loss": 0.0797, | |
| "step": 39800 | |
| }, | |
| { | |
| "epoch": 12.5, | |
| "learning_rate": 4.0686607749830045e-05, | |
| "loss": 0.0798, | |
| "step": 39900 | |
| }, | |
| { | |
| "epoch": 12.53, | |
| "learning_rate": 4.051665533650578e-05, | |
| "loss": 0.0801, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 12.53, | |
| "eval_loss": 0.1791016012430191, | |
| "eval_runtime": 152.3374, | |
| "eval_samples_per_second": 104.643, | |
| "eval_steps_per_second": 1.641, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 12.56, | |
| "learning_rate": 4.034670292318151e-05, | |
| "loss": 0.0815, | |
| "step": 40100 | |
| }, | |
| { | |
| "epoch": 12.6, | |
| "learning_rate": 4.017675050985724e-05, | |
| "loss": 0.0807, | |
| "step": 40200 | |
| }, | |
| { | |
| "epoch": 12.63, | |
| "learning_rate": 4.000679809653297e-05, | |
| "loss": 0.0789, | |
| "step": 40300 | |
| }, | |
| { | |
| "epoch": 12.66, | |
| "learning_rate": 3.98368456832087e-05, | |
| "loss": 0.0793, | |
| "step": 40400 | |
| }, | |
| { | |
| "epoch": 12.69, | |
| "learning_rate": 3.9666893269884435e-05, | |
| "loss": 0.0793, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 12.72, | |
| "learning_rate": 3.9496940856560164e-05, | |
| "loss": 0.0775, | |
| "step": 40600 | |
| }, | |
| { | |
| "epoch": 12.75, | |
| "learning_rate": 3.9326988443235894e-05, | |
| "loss": 0.0802, | |
| "step": 40700 | |
| }, | |
| { | |
| "epoch": 12.78, | |
| "learning_rate": 3.915703602991162e-05, | |
| "loss": 0.0814, | |
| "step": 40800 | |
| }, | |
| { | |
| "epoch": 12.82, | |
| "learning_rate": 3.898708361658735e-05, | |
| "loss": 0.0809, | |
| "step": 40900 | |
| }, | |
| { | |
| "epoch": 12.85, | |
| "learning_rate": 3.881713120326309e-05, | |
| "loss": 0.0786, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 12.88, | |
| "learning_rate": 3.864717878993882e-05, | |
| "loss": 0.0794, | |
| "step": 41100 | |
| }, | |
| { | |
| "epoch": 12.91, | |
| "learning_rate": 3.847722637661455e-05, | |
| "loss": 0.0795, | |
| "step": 41200 | |
| }, | |
| { | |
| "epoch": 12.94, | |
| "learning_rate": 3.830727396329028e-05, | |
| "loss": 0.0805, | |
| "step": 41300 | |
| }, | |
| { | |
| "epoch": 12.97, | |
| "learning_rate": 3.8137321549966006e-05, | |
| "loss": 0.0777, | |
| "step": 41400 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "learning_rate": 3.796736913664174e-05, | |
| "loss": 0.0797, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 13.03, | |
| "learning_rate": 3.779741672331747e-05, | |
| "loss": 0.074, | |
| "step": 41600 | |
| }, | |
| { | |
| "epoch": 13.07, | |
| "learning_rate": 3.76274643099932e-05, | |
| "loss": 0.0734, | |
| "step": 41700 | |
| }, | |
| { | |
| "epoch": 13.1, | |
| "learning_rate": 3.745751189666894e-05, | |
| "loss": 0.0741, | |
| "step": 41800 | |
| }, | |
| { | |
| "epoch": 13.13, | |
| "learning_rate": 3.728755948334467e-05, | |
| "loss": 0.0732, | |
| "step": 41900 | |
| }, | |
| { | |
| "epoch": 13.16, | |
| "learning_rate": 3.7117607070020396e-05, | |
| "loss": 0.0733, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 13.19, | |
| "learning_rate": 3.6947654656696126e-05, | |
| "loss": 0.0728, | |
| "step": 42100 | |
| }, | |
| { | |
| "epoch": 13.22, | |
| "learning_rate": 3.677770224337186e-05, | |
| "loss": 0.0742, | |
| "step": 42200 | |
| }, | |
| { | |
| "epoch": 13.25, | |
| "learning_rate": 3.660774983004759e-05, | |
| "loss": 0.0741, | |
| "step": 42300 | |
| }, | |
| { | |
| "epoch": 13.29, | |
| "learning_rate": 3.643779741672332e-05, | |
| "loss": 0.0746, | |
| "step": 42400 | |
| }, | |
| { | |
| "epoch": 13.32, | |
| "learning_rate": 3.626784500339905e-05, | |
| "loss": 0.0721, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 13.35, | |
| "learning_rate": 3.6097892590074786e-05, | |
| "loss": 0.0755, | |
| "step": 42600 | |
| }, | |
| { | |
| "epoch": 13.38, | |
| "learning_rate": 3.5927940176750516e-05, | |
| "loss": 0.0752, | |
| "step": 42700 | |
| }, | |
| { | |
| "epoch": 13.41, | |
| "learning_rate": 3.5757987763426245e-05, | |
| "loss": 0.0728, | |
| "step": 42800 | |
| }, | |
| { | |
| "epoch": 13.44, | |
| "learning_rate": 3.5588035350101974e-05, | |
| "loss": 0.0724, | |
| "step": 42900 | |
| }, | |
| { | |
| "epoch": 13.47, | |
| "learning_rate": 3.5418082936777704e-05, | |
| "loss": 0.0722, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 13.5, | |
| "learning_rate": 3.524813052345343e-05, | |
| "loss": 0.0746, | |
| "step": 43100 | |
| }, | |
| { | |
| "epoch": 13.54, | |
| "learning_rate": 3.507817811012917e-05, | |
| "loss": 0.0715, | |
| "step": 43200 | |
| }, | |
| { | |
| "epoch": 13.57, | |
| "learning_rate": 3.49082256968049e-05, | |
| "loss": 0.0722, | |
| "step": 43300 | |
| }, | |
| { | |
| "epoch": 13.6, | |
| "learning_rate": 3.473827328348063e-05, | |
| "loss": 0.0745, | |
| "step": 43400 | |
| }, | |
| { | |
| "epoch": 13.63, | |
| "learning_rate": 3.456832087015636e-05, | |
| "loss": 0.0738, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 13.66, | |
| "learning_rate": 3.439836845683209e-05, | |
| "loss": 0.0733, | |
| "step": 43600 | |
| }, | |
| { | |
| "epoch": 13.69, | |
| "learning_rate": 3.422841604350782e-05, | |
| "loss": 0.0718, | |
| "step": 43700 | |
| }, | |
| { | |
| "epoch": 13.72, | |
| "learning_rate": 3.405846363018355e-05, | |
| "loss": 0.0729, | |
| "step": 43800 | |
| }, | |
| { | |
| "epoch": 13.76, | |
| "learning_rate": 3.388851121685928e-05, | |
| "loss": 0.0713, | |
| "step": 43900 | |
| }, | |
| { | |
| "epoch": 13.79, | |
| "learning_rate": 3.371855880353501e-05, | |
| "loss": 0.0726, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 13.82, | |
| "learning_rate": 3.354860639021074e-05, | |
| "loss": 0.073, | |
| "step": 44100 | |
| }, | |
| { | |
| "epoch": 13.85, | |
| "learning_rate": 3.337865397688648e-05, | |
| "loss": 0.0715, | |
| "step": 44200 | |
| }, | |
| { | |
| "epoch": 13.88, | |
| "learning_rate": 3.3208701563562206e-05, | |
| "loss": 0.0733, | |
| "step": 44300 | |
| }, | |
| { | |
| "epoch": 13.91, | |
| "learning_rate": 3.3038749150237936e-05, | |
| "loss": 0.0756, | |
| "step": 44400 | |
| }, | |
| { | |
| "epoch": 13.94, | |
| "learning_rate": 3.2868796736913665e-05, | |
| "loss": 0.0708, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 13.97, | |
| "learning_rate": 3.2698844323589395e-05, | |
| "loss": 0.0728, | |
| "step": 44600 | |
| }, | |
| { | |
| "epoch": 14.01, | |
| "learning_rate": 3.2528891910265124e-05, | |
| "loss": 0.071, | |
| "step": 44700 | |
| }, | |
| { | |
| "epoch": 14.04, | |
| "learning_rate": 3.235893949694086e-05, | |
| "loss": 0.067, | |
| "step": 44800 | |
| }, | |
| { | |
| "epoch": 14.07, | |
| "learning_rate": 3.218898708361659e-05, | |
| "loss": 0.0667, | |
| "step": 44900 | |
| }, | |
| { | |
| "epoch": 14.1, | |
| "learning_rate": 3.201903467029232e-05, | |
| "loss": 0.0654, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 14.1, | |
| "eval_loss": 0.1702832728624344, | |
| "eval_runtime": 155.4539, | |
| "eval_samples_per_second": 102.545, | |
| "eval_steps_per_second": 1.608, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 14.13, | |
| "learning_rate": 3.184908225696805e-05, | |
| "loss": 0.0665, | |
| "step": 45100 | |
| }, | |
| { | |
| "epoch": 14.16, | |
| "learning_rate": 3.167912984364378e-05, | |
| "loss": 0.066, | |
| "step": 45200 | |
| }, | |
| { | |
| "epoch": 14.19, | |
| "learning_rate": 3.1509177430319514e-05, | |
| "loss": 0.0663, | |
| "step": 45300 | |
| }, | |
| { | |
| "epoch": 14.23, | |
| "learning_rate": 3.133922501699524e-05, | |
| "loss": 0.0658, | |
| "step": 45400 | |
| }, | |
| { | |
| "epoch": 14.26, | |
| "learning_rate": 3.116927260367097e-05, | |
| "loss": 0.0663, | |
| "step": 45500 | |
| }, | |
| { | |
| "epoch": 14.29, | |
| "learning_rate": 3.09993201903467e-05, | |
| "loss": 0.0664, | |
| "step": 45600 | |
| }, | |
| { | |
| "epoch": 14.32, | |
| "learning_rate": 3.082936777702243e-05, | |
| "loss": 0.0679, | |
| "step": 45700 | |
| }, | |
| { | |
| "epoch": 14.35, | |
| "learning_rate": 3.065941536369817e-05, | |
| "loss": 0.0666, | |
| "step": 45800 | |
| }, | |
| { | |
| "epoch": 14.38, | |
| "learning_rate": 3.0489462950373897e-05, | |
| "loss": 0.066, | |
| "step": 45900 | |
| }, | |
| { | |
| "epoch": 14.41, | |
| "learning_rate": 3.0319510537049626e-05, | |
| "loss": 0.0668, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 14.44, | |
| "learning_rate": 3.014955812372536e-05, | |
| "loss": 0.0678, | |
| "step": 46100 | |
| }, | |
| { | |
| "epoch": 14.48, | |
| "learning_rate": 2.997960571040109e-05, | |
| "loss": 0.0675, | |
| "step": 46200 | |
| }, | |
| { | |
| "epoch": 14.51, | |
| "learning_rate": 2.9809653297076818e-05, | |
| "loss": 0.066, | |
| "step": 46300 | |
| }, | |
| { | |
| "epoch": 14.54, | |
| "learning_rate": 2.9639700883752554e-05, | |
| "loss": 0.0663, | |
| "step": 46400 | |
| }, | |
| { | |
| "epoch": 14.57, | |
| "learning_rate": 2.9469748470428284e-05, | |
| "loss": 0.0669, | |
| "step": 46500 | |
| }, | |
| { | |
| "epoch": 14.6, | |
| "learning_rate": 2.9299796057104013e-05, | |
| "loss": 0.0656, | |
| "step": 46600 | |
| }, | |
| { | |
| "epoch": 14.63, | |
| "learning_rate": 2.9129843643779742e-05, | |
| "loss": 0.0638, | |
| "step": 46700 | |
| }, | |
| { | |
| "epoch": 14.66, | |
| "learning_rate": 2.8959891230455472e-05, | |
| "loss": 0.0669, | |
| "step": 46800 | |
| }, | |
| { | |
| "epoch": 14.7, | |
| "learning_rate": 2.8789938817131208e-05, | |
| "loss": 0.0638, | |
| "step": 46900 | |
| }, | |
| { | |
| "epoch": 14.73, | |
| "learning_rate": 2.8619986403806937e-05, | |
| "loss": 0.0655, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 14.76, | |
| "learning_rate": 2.8450033990482667e-05, | |
| "loss": 0.0672, | |
| "step": 47100 | |
| }, | |
| { | |
| "epoch": 14.79, | |
| "learning_rate": 2.8280081577158396e-05, | |
| "loss": 0.0655, | |
| "step": 47200 | |
| }, | |
| { | |
| "epoch": 14.82, | |
| "learning_rate": 2.8110129163834125e-05, | |
| "loss": 0.066, | |
| "step": 47300 | |
| }, | |
| { | |
| "epoch": 14.85, | |
| "learning_rate": 2.794017675050986e-05, | |
| "loss": 0.065, | |
| "step": 47400 | |
| }, | |
| { | |
| "epoch": 14.88, | |
| "learning_rate": 2.777022433718559e-05, | |
| "loss": 0.0655, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 14.91, | |
| "learning_rate": 2.760027192386132e-05, | |
| "loss": 0.066, | |
| "step": 47600 | |
| }, | |
| { | |
| "epoch": 14.95, | |
| "learning_rate": 2.743031951053705e-05, | |
| "loss": 0.0633, | |
| "step": 47700 | |
| }, | |
| { | |
| "epoch": 14.98, | |
| "learning_rate": 2.726036709721278e-05, | |
| "loss": 0.0657, | |
| "step": 47800 | |
| }, | |
| { | |
| "epoch": 15.01, | |
| "learning_rate": 2.709041468388851e-05, | |
| "loss": 0.0641, | |
| "step": 47900 | |
| }, | |
| { | |
| "epoch": 15.04, | |
| "learning_rate": 2.6920462270564245e-05, | |
| "loss": 0.0593, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 15.07, | |
| "learning_rate": 2.6750509857239974e-05, | |
| "loss": 0.0595, | |
| "step": 48100 | |
| }, | |
| { | |
| "epoch": 15.1, | |
| "learning_rate": 2.6580557443915704e-05, | |
| "loss": 0.0601, | |
| "step": 48200 | |
| }, | |
| { | |
| "epoch": 15.13, | |
| "learning_rate": 2.6410605030591433e-05, | |
| "loss": 0.0592, | |
| "step": 48300 | |
| }, | |
| { | |
| "epoch": 15.17, | |
| "learning_rate": 2.6240652617267162e-05, | |
| "loss": 0.062, | |
| "step": 48400 | |
| }, | |
| { | |
| "epoch": 15.2, | |
| "learning_rate": 2.60707002039429e-05, | |
| "loss": 0.059, | |
| "step": 48500 | |
| }, | |
| { | |
| "epoch": 15.23, | |
| "learning_rate": 2.5900747790618628e-05, | |
| "loss": 0.0602, | |
| "step": 48600 | |
| }, | |
| { | |
| "epoch": 15.26, | |
| "learning_rate": 2.5730795377294357e-05, | |
| "loss": 0.0613, | |
| "step": 48700 | |
| }, | |
| { | |
| "epoch": 15.29, | |
| "learning_rate": 2.556084296397009e-05, | |
| "loss": 0.0596, | |
| "step": 48800 | |
| }, | |
| { | |
| "epoch": 15.32, | |
| "learning_rate": 2.539089055064582e-05, | |
| "loss": 0.0615, | |
| "step": 48900 | |
| }, | |
| { | |
| "epoch": 15.35, | |
| "learning_rate": 2.5220938137321552e-05, | |
| "loss": 0.0592, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 15.38, | |
| "learning_rate": 2.5050985723997282e-05, | |
| "loss": 0.0587, | |
| "step": 49100 | |
| }, | |
| { | |
| "epoch": 15.42, | |
| "learning_rate": 2.4881033310673014e-05, | |
| "loss": 0.0595, | |
| "step": 49200 | |
| }, | |
| { | |
| "epoch": 15.45, | |
| "learning_rate": 2.4711080897348744e-05, | |
| "loss": 0.0585, | |
| "step": 49300 | |
| }, | |
| { | |
| "epoch": 15.48, | |
| "learning_rate": 2.4541128484024477e-05, | |
| "loss": 0.0603, | |
| "step": 49400 | |
| }, | |
| { | |
| "epoch": 15.51, | |
| "learning_rate": 2.4371176070700206e-05, | |
| "loss": 0.0605, | |
| "step": 49500 | |
| }, | |
| { | |
| "epoch": 15.54, | |
| "learning_rate": 2.4201223657375935e-05, | |
| "loss": 0.0604, | |
| "step": 49600 | |
| }, | |
| { | |
| "epoch": 15.57, | |
| "learning_rate": 2.4031271244051668e-05, | |
| "loss": 0.0583, | |
| "step": 49700 | |
| }, | |
| { | |
| "epoch": 15.6, | |
| "learning_rate": 2.3861318830727398e-05, | |
| "loss": 0.0569, | |
| "step": 49800 | |
| }, | |
| { | |
| "epoch": 15.64, | |
| "learning_rate": 2.3691366417403127e-05, | |
| "loss": 0.0601, | |
| "step": 49900 | |
| }, | |
| { | |
| "epoch": 15.67, | |
| "learning_rate": 2.352141400407886e-05, | |
| "loss": 0.0583, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 15.67, | |
| "eval_loss": 0.16135616600513458, | |
| "eval_runtime": 147.0571, | |
| "eval_samples_per_second": 108.4, | |
| "eval_steps_per_second": 1.7, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 15.7, | |
| "learning_rate": 2.335146159075459e-05, | |
| "loss": 0.0596, | |
| "step": 50100 | |
| }, | |
| { | |
| "epoch": 15.73, | |
| "learning_rate": 2.3181509177430322e-05, | |
| "loss": 0.059, | |
| "step": 50200 | |
| }, | |
| { | |
| "epoch": 15.76, | |
| "learning_rate": 2.301155676410605e-05, | |
| "loss": 0.0568, | |
| "step": 50300 | |
| }, | |
| { | |
| "epoch": 15.79, | |
| "learning_rate": 2.284160435078178e-05, | |
| "loss": 0.0605, | |
| "step": 50400 | |
| }, | |
| { | |
| "epoch": 15.82, | |
| "learning_rate": 2.2671651937457514e-05, | |
| "loss": 0.0589, | |
| "step": 50500 | |
| }, | |
| { | |
| "epoch": 15.85, | |
| "learning_rate": 2.2501699524133243e-05, | |
| "loss": 0.0579, | |
| "step": 50600 | |
| }, | |
| { | |
| "epoch": 15.89, | |
| "learning_rate": 2.2331747110808972e-05, | |
| "loss": 0.059, | |
| "step": 50700 | |
| }, | |
| { | |
| "epoch": 15.92, | |
| "learning_rate": 2.2161794697484705e-05, | |
| "loss": 0.0585, | |
| "step": 50800 | |
| }, | |
| { | |
| "epoch": 15.95, | |
| "learning_rate": 2.1991842284160435e-05, | |
| "loss": 0.0593, | |
| "step": 50900 | |
| }, | |
| { | |
| "epoch": 15.98, | |
| "learning_rate": 2.1821889870836167e-05, | |
| "loss": 0.0591, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 16.01, | |
| "learning_rate": 2.1651937457511897e-05, | |
| "loss": 0.0567, | |
| "step": 51100 | |
| }, | |
| { | |
| "epoch": 16.04, | |
| "learning_rate": 2.1481985044187626e-05, | |
| "loss": 0.056, | |
| "step": 51200 | |
| }, | |
| { | |
| "epoch": 16.07, | |
| "learning_rate": 2.131203263086336e-05, | |
| "loss": 0.0547, | |
| "step": 51300 | |
| }, | |
| { | |
| "epoch": 16.11, | |
| "learning_rate": 2.1142080217539088e-05, | |
| "loss": 0.0543, | |
| "step": 51400 | |
| }, | |
| { | |
| "epoch": 16.14, | |
| "learning_rate": 2.097212780421482e-05, | |
| "loss": 0.0545, | |
| "step": 51500 | |
| }, | |
| { | |
| "epoch": 16.17, | |
| "learning_rate": 2.080217539089055e-05, | |
| "loss": 0.0546, | |
| "step": 51600 | |
| }, | |
| { | |
| "epoch": 16.2, | |
| "learning_rate": 2.0632222977566283e-05, | |
| "loss": 0.0533, | |
| "step": 51700 | |
| }, | |
| { | |
| "epoch": 16.23, | |
| "learning_rate": 2.0462270564242013e-05, | |
| "loss": 0.0529, | |
| "step": 51800 | |
| }, | |
| { | |
| "epoch": 16.26, | |
| "learning_rate": 2.0292318150917745e-05, | |
| "loss": 0.0548, | |
| "step": 51900 | |
| }, | |
| { | |
| "epoch": 16.29, | |
| "learning_rate": 2.0122365737593475e-05, | |
| "loss": 0.0527, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 16.32, | |
| "learning_rate": 1.9952413324269208e-05, | |
| "loss": 0.0517, | |
| "step": 52100 | |
| }, | |
| { | |
| "epoch": 16.36, | |
| "learning_rate": 1.9782460910944937e-05, | |
| "loss": 0.0544, | |
| "step": 52200 | |
| }, | |
| { | |
| "epoch": 16.39, | |
| "learning_rate": 1.9612508497620666e-05, | |
| "loss": 0.0543, | |
| "step": 52300 | |
| }, | |
| { | |
| "epoch": 16.42, | |
| "learning_rate": 1.94425560842964e-05, | |
| "loss": 0.0538, | |
| "step": 52400 | |
| }, | |
| { | |
| "epoch": 16.45, | |
| "learning_rate": 1.927260367097213e-05, | |
| "loss": 0.0525, | |
| "step": 52500 | |
| }, | |
| { | |
| "epoch": 16.48, | |
| "learning_rate": 1.910265125764786e-05, | |
| "loss": 0.0534, | |
| "step": 52600 | |
| }, | |
| { | |
| "epoch": 16.51, | |
| "learning_rate": 1.893269884432359e-05, | |
| "loss": 0.0552, | |
| "step": 52700 | |
| }, | |
| { | |
| "epoch": 16.54, | |
| "learning_rate": 1.876274643099932e-05, | |
| "loss": 0.0554, | |
| "step": 52800 | |
| }, | |
| { | |
| "epoch": 16.58, | |
| "learning_rate": 1.8592794017675053e-05, | |
| "loss": 0.0525, | |
| "step": 52900 | |
| }, | |
| { | |
| "epoch": 16.61, | |
| "learning_rate": 1.8422841604350782e-05, | |
| "loss": 0.0536, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 16.64, | |
| "learning_rate": 1.8252889191026512e-05, | |
| "loss": 0.0539, | |
| "step": 53100 | |
| }, | |
| { | |
| "epoch": 16.67, | |
| "learning_rate": 1.8082936777702245e-05, | |
| "loss": 0.0525, | |
| "step": 53200 | |
| }, | |
| { | |
| "epoch": 16.7, | |
| "learning_rate": 1.7912984364377974e-05, | |
| "loss": 0.0524, | |
| "step": 53300 | |
| }, | |
| { | |
| "epoch": 16.73, | |
| "learning_rate": 1.7743031951053707e-05, | |
| "loss": 0.0534, | |
| "step": 53400 | |
| }, | |
| { | |
| "epoch": 16.76, | |
| "learning_rate": 1.7573079537729436e-05, | |
| "loss": 0.0517, | |
| "step": 53500 | |
| }, | |
| { | |
| "epoch": 16.79, | |
| "learning_rate": 1.7403127124405166e-05, | |
| "loss": 0.0524, | |
| "step": 53600 | |
| }, | |
| { | |
| "epoch": 16.83, | |
| "learning_rate": 1.7233174711080898e-05, | |
| "loss": 0.0519, | |
| "step": 53700 | |
| }, | |
| { | |
| "epoch": 16.86, | |
| "learning_rate": 1.7063222297756628e-05, | |
| "loss": 0.0513, | |
| "step": 53800 | |
| }, | |
| { | |
| "epoch": 16.89, | |
| "learning_rate": 1.6893269884432357e-05, | |
| "loss": 0.0531, | |
| "step": 53900 | |
| }, | |
| { | |
| "epoch": 16.92, | |
| "learning_rate": 1.672331747110809e-05, | |
| "loss": 0.052, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 16.95, | |
| "learning_rate": 1.655336505778382e-05, | |
| "loss": 0.0526, | |
| "step": 54100 | |
| }, | |
| { | |
| "epoch": 16.98, | |
| "learning_rate": 1.6383412644459552e-05, | |
| "loss": 0.0536, | |
| "step": 54200 | |
| }, | |
| { | |
| "epoch": 17.01, | |
| "learning_rate": 1.621346023113528e-05, | |
| "loss": 0.0512, | |
| "step": 54300 | |
| }, | |
| { | |
| "epoch": 17.05, | |
| "learning_rate": 1.6043507817811014e-05, | |
| "loss": 0.0507, | |
| "step": 54400 | |
| }, | |
| { | |
| "epoch": 17.08, | |
| "learning_rate": 1.5873555404486744e-05, | |
| "loss": 0.0478, | |
| "step": 54500 | |
| }, | |
| { | |
| "epoch": 17.11, | |
| "learning_rate": 1.5703602991162476e-05, | |
| "loss": 0.0481, | |
| "step": 54600 | |
| }, | |
| { | |
| "epoch": 17.14, | |
| "learning_rate": 1.5533650577838206e-05, | |
| "loss": 0.0487, | |
| "step": 54700 | |
| }, | |
| { | |
| "epoch": 17.17, | |
| "learning_rate": 1.536369816451394e-05, | |
| "loss": 0.0475, | |
| "step": 54800 | |
| }, | |
| { | |
| "epoch": 17.2, | |
| "learning_rate": 1.5193745751189666e-05, | |
| "loss": 0.0485, | |
| "step": 54900 | |
| }, | |
| { | |
| "epoch": 17.23, | |
| "learning_rate": 1.5023793337865399e-05, | |
| "loss": 0.0471, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 17.23, | |
| "eval_loss": 0.15533138811588287, | |
| "eval_runtime": 149.8141, | |
| "eval_samples_per_second": 106.405, | |
| "eval_steps_per_second": 1.669, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 17.26, | |
| "learning_rate": 1.4853840924541128e-05, | |
| "loss": 0.0494, | |
| "step": 55100 | |
| }, | |
| { | |
| "epoch": 17.3, | |
| "learning_rate": 1.468388851121686e-05, | |
| "loss": 0.049, | |
| "step": 55200 | |
| }, | |
| { | |
| "epoch": 17.33, | |
| "learning_rate": 1.451393609789259e-05, | |
| "loss": 0.05, | |
| "step": 55300 | |
| }, | |
| { | |
| "epoch": 17.36, | |
| "learning_rate": 1.4343983684568322e-05, | |
| "loss": 0.049, | |
| "step": 55400 | |
| }, | |
| { | |
| "epoch": 17.39, | |
| "learning_rate": 1.4174031271244055e-05, | |
| "loss": 0.0479, | |
| "step": 55500 | |
| }, | |
| { | |
| "epoch": 17.42, | |
| "learning_rate": 1.4004078857919784e-05, | |
| "loss": 0.0469, | |
| "step": 55600 | |
| }, | |
| { | |
| "epoch": 17.45, | |
| "learning_rate": 1.3834126444595513e-05, | |
| "loss": 0.0506, | |
| "step": 55700 | |
| }, | |
| { | |
| "epoch": 17.48, | |
| "learning_rate": 1.3664174031271246e-05, | |
| "loss": 0.0469, | |
| "step": 55800 | |
| }, | |
| { | |
| "epoch": 17.52, | |
| "learning_rate": 1.3494221617946976e-05, | |
| "loss": 0.0473, | |
| "step": 55900 | |
| }, | |
| { | |
| "epoch": 17.55, | |
| "learning_rate": 1.3324269204622705e-05, | |
| "loss": 0.0478, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 17.58, | |
| "learning_rate": 1.3154316791298438e-05, | |
| "loss": 0.0477, | |
| "step": 56100 | |
| }, | |
| { | |
| "epoch": 17.61, | |
| "learning_rate": 1.2984364377974167e-05, | |
| "loss": 0.0476, | |
| "step": 56200 | |
| }, | |
| { | |
| "epoch": 17.64, | |
| "learning_rate": 1.28144119646499e-05, | |
| "loss": 0.0489, | |
| "step": 56300 | |
| }, | |
| { | |
| "epoch": 17.67, | |
| "learning_rate": 1.264445955132563e-05, | |
| "loss": 0.0455, | |
| "step": 56400 | |
| }, | |
| { | |
| "epoch": 17.7, | |
| "learning_rate": 1.247450713800136e-05, | |
| "loss": 0.0495, | |
| "step": 56500 | |
| }, | |
| { | |
| "epoch": 17.73, | |
| "learning_rate": 1.2304554724677091e-05, | |
| "loss": 0.048, | |
| "step": 56600 | |
| }, | |
| { | |
| "epoch": 17.77, | |
| "learning_rate": 1.2134602311352823e-05, | |
| "loss": 0.0477, | |
| "step": 56700 | |
| }, | |
| { | |
| "epoch": 17.8, | |
| "learning_rate": 1.1964649898028554e-05, | |
| "loss": 0.0463, | |
| "step": 56800 | |
| }, | |
| { | |
| "epoch": 17.83, | |
| "learning_rate": 1.1794697484704285e-05, | |
| "loss": 0.0467, | |
| "step": 56900 | |
| }, | |
| { | |
| "epoch": 17.86, | |
| "learning_rate": 1.1624745071380014e-05, | |
| "loss": 0.047, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 17.89, | |
| "learning_rate": 1.1454792658055745e-05, | |
| "loss": 0.0468, | |
| "step": 57100 | |
| }, | |
| { | |
| "epoch": 17.92, | |
| "learning_rate": 1.1284840244731476e-05, | |
| "loss": 0.0469, | |
| "step": 57200 | |
| }, | |
| { | |
| "epoch": 17.95, | |
| "learning_rate": 1.1114887831407207e-05, | |
| "loss": 0.0476, | |
| "step": 57300 | |
| }, | |
| { | |
| "epoch": 17.99, | |
| "learning_rate": 1.0944935418082937e-05, | |
| "loss": 0.046, | |
| "step": 57400 | |
| }, | |
| { | |
| "epoch": 18.02, | |
| "learning_rate": 1.0774983004758668e-05, | |
| "loss": 0.045, | |
| "step": 57500 | |
| }, | |
| { | |
| "epoch": 18.05, | |
| "learning_rate": 1.0605030591434399e-05, | |
| "loss": 0.0426, | |
| "step": 57600 | |
| }, | |
| { | |
| "epoch": 18.08, | |
| "learning_rate": 1.043507817811013e-05, | |
| "loss": 0.0432, | |
| "step": 57700 | |
| }, | |
| { | |
| "epoch": 18.11, | |
| "learning_rate": 1.026512576478586e-05, | |
| "loss": 0.0441, | |
| "step": 57800 | |
| }, | |
| { | |
| "epoch": 18.14, | |
| "learning_rate": 1.009517335146159e-05, | |
| "loss": 0.0435, | |
| "step": 57900 | |
| }, | |
| { | |
| "epoch": 18.17, | |
| "learning_rate": 9.925220938137322e-06, | |
| "loss": 0.0451, | |
| "step": 58000 | |
| }, | |
| { | |
| "epoch": 18.2, | |
| "learning_rate": 9.755268524813053e-06, | |
| "loss": 0.0448, | |
| "step": 58100 | |
| }, | |
| { | |
| "epoch": 18.24, | |
| "learning_rate": 9.585316111488784e-06, | |
| "loss": 0.0439, | |
| "step": 58200 | |
| }, | |
| { | |
| "epoch": 18.27, | |
| "learning_rate": 9.415363698164515e-06, | |
| "loss": 0.0447, | |
| "step": 58300 | |
| }, | |
| { | |
| "epoch": 18.3, | |
| "learning_rate": 9.245411284840246e-06, | |
| "loss": 0.042, | |
| "step": 58400 | |
| }, | |
| { | |
| "epoch": 18.33, | |
| "learning_rate": 9.075458871515977e-06, | |
| "loss": 0.0446, | |
| "step": 58500 | |
| }, | |
| { | |
| "epoch": 18.36, | |
| "learning_rate": 8.905506458191706e-06, | |
| "loss": 0.0426, | |
| "step": 58600 | |
| }, | |
| { | |
| "epoch": 18.39, | |
| "learning_rate": 8.735554044867438e-06, | |
| "loss": 0.044, | |
| "step": 58700 | |
| }, | |
| { | |
| "epoch": 18.42, | |
| "learning_rate": 8.565601631543169e-06, | |
| "loss": 0.0427, | |
| "step": 58800 | |
| }, | |
| { | |
| "epoch": 18.46, | |
| "learning_rate": 8.3956492182189e-06, | |
| "loss": 0.0432, | |
| "step": 58900 | |
| }, | |
| { | |
| "epoch": 18.49, | |
| "learning_rate": 8.225696804894629e-06, | |
| "loss": 0.0438, | |
| "step": 59000 | |
| }, | |
| { | |
| "epoch": 18.52, | |
| "learning_rate": 8.05574439157036e-06, | |
| "loss": 0.0417, | |
| "step": 59100 | |
| }, | |
| { | |
| "epoch": 18.55, | |
| "learning_rate": 7.885791978246091e-06, | |
| "loss": 0.0442, | |
| "step": 59200 | |
| }, | |
| { | |
| "epoch": 18.58, | |
| "learning_rate": 7.715839564921822e-06, | |
| "loss": 0.0428, | |
| "step": 59300 | |
| }, | |
| { | |
| "epoch": 18.61, | |
| "learning_rate": 7.545887151597553e-06, | |
| "loss": 0.0426, | |
| "step": 59400 | |
| }, | |
| { | |
| "epoch": 18.64, | |
| "learning_rate": 7.375934738273284e-06, | |
| "loss": 0.042, | |
| "step": 59500 | |
| }, | |
| { | |
| "epoch": 18.67, | |
| "learning_rate": 7.205982324949015e-06, | |
| "loss": 0.0414, | |
| "step": 59600 | |
| }, | |
| { | |
| "epoch": 18.71, | |
| "learning_rate": 7.036029911624746e-06, | |
| "loss": 0.0415, | |
| "step": 59700 | |
| }, | |
| { | |
| "epoch": 18.74, | |
| "learning_rate": 6.866077498300476e-06, | |
| "loss": 0.0434, | |
| "step": 59800 | |
| }, | |
| { | |
| "epoch": 18.77, | |
| "learning_rate": 6.696125084976207e-06, | |
| "loss": 0.0425, | |
| "step": 59900 | |
| }, | |
| { | |
| "epoch": 18.8, | |
| "learning_rate": 6.526172671651938e-06, | |
| "loss": 0.0411, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 18.8, | |
| "eval_loss": 0.15011541545391083, | |
| "eval_runtime": 151.7868, | |
| "eval_samples_per_second": 105.022, | |
| "eval_steps_per_second": 1.647, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 18.83, | |
| "learning_rate": 6.3562202583276694e-06, | |
| "loss": 0.0414, | |
| "step": 60100 | |
| }, | |
| { | |
| "epoch": 18.86, | |
| "learning_rate": 6.1862678450034e-06, | |
| "loss": 0.0427, | |
| "step": 60200 | |
| }, | |
| { | |
| "epoch": 18.89, | |
| "learning_rate": 6.01631543167913e-06, | |
| "loss": 0.0417, | |
| "step": 60300 | |
| }, | |
| { | |
| "epoch": 18.93, | |
| "learning_rate": 5.846363018354861e-06, | |
| "loss": 0.0414, | |
| "step": 60400 | |
| }, | |
| { | |
| "epoch": 18.96, | |
| "learning_rate": 5.676410605030591e-06, | |
| "loss": 0.0414, | |
| "step": 60500 | |
| }, | |
| { | |
| "epoch": 18.99, | |
| "learning_rate": 5.506458191706323e-06, | |
| "loss": 0.0417, | |
| "step": 60600 | |
| }, | |
| { | |
| "epoch": 19.02, | |
| "learning_rate": 5.3365057783820534e-06, | |
| "loss": 0.0405, | |
| "step": 60700 | |
| }, | |
| { | |
| "epoch": 19.05, | |
| "learning_rate": 5.1665533650577845e-06, | |
| "loss": 0.0398, | |
| "step": 60800 | |
| }, | |
| { | |
| "epoch": 19.08, | |
| "learning_rate": 4.996600951733515e-06, | |
| "loss": 0.0393, | |
| "step": 60900 | |
| }, | |
| { | |
| "epoch": 19.11, | |
| "learning_rate": 4.826648538409246e-06, | |
| "loss": 0.0395, | |
| "step": 61000 | |
| }, | |
| { | |
| "epoch": 19.14, | |
| "learning_rate": 4.656696125084976e-06, | |
| "loss": 0.0401, | |
| "step": 61100 | |
| }, | |
| { | |
| "epoch": 19.18, | |
| "learning_rate": 4.486743711760707e-06, | |
| "loss": 0.0412, | |
| "step": 61200 | |
| }, | |
| { | |
| "epoch": 19.21, | |
| "learning_rate": 4.316791298436438e-06, | |
| "loss": 0.0379, | |
| "step": 61300 | |
| }, | |
| { | |
| "epoch": 19.24, | |
| "learning_rate": 4.146838885112169e-06, | |
| "loss": 0.0374, | |
| "step": 61400 | |
| }, | |
| { | |
| "epoch": 19.27, | |
| "learning_rate": 3.9768864717879e-06, | |
| "loss": 0.0419, | |
| "step": 61500 | |
| }, | |
| { | |
| "epoch": 19.3, | |
| "learning_rate": 3.8069340584636307e-06, | |
| "loss": 0.0386, | |
| "step": 61600 | |
| }, | |
| { | |
| "epoch": 19.33, | |
| "learning_rate": 3.636981645139361e-06, | |
| "loss": 0.0394, | |
| "step": 61700 | |
| }, | |
| { | |
| "epoch": 19.36, | |
| "learning_rate": 3.467029231815092e-06, | |
| "loss": 0.0391, | |
| "step": 61800 | |
| }, | |
| { | |
| "epoch": 19.4, | |
| "learning_rate": 3.2970768184908227e-06, | |
| "loss": 0.0403, | |
| "step": 61900 | |
| }, | |
| { | |
| "epoch": 19.43, | |
| "learning_rate": 3.1271244051665538e-06, | |
| "loss": 0.0382, | |
| "step": 62000 | |
| }, | |
| { | |
| "epoch": 19.46, | |
| "learning_rate": 2.9571719918422845e-06, | |
| "loss": 0.0394, | |
| "step": 62100 | |
| }, | |
| { | |
| "epoch": 19.49, | |
| "learning_rate": 2.787219578518015e-06, | |
| "loss": 0.0385, | |
| "step": 62200 | |
| }, | |
| { | |
| "epoch": 19.52, | |
| "learning_rate": 2.6172671651937458e-06, | |
| "loss": 0.0388, | |
| "step": 62300 | |
| }, | |
| { | |
| "epoch": 19.55, | |
| "learning_rate": 2.4473147518694765e-06, | |
| "loss": 0.0404, | |
| "step": 62400 | |
| }, | |
| { | |
| "epoch": 19.58, | |
| "learning_rate": 2.2773623385452075e-06, | |
| "loss": 0.0387, | |
| "step": 62500 | |
| }, | |
| { | |
| "epoch": 19.61, | |
| "learning_rate": 2.107409925220938e-06, | |
| "loss": 0.0383, | |
| "step": 62600 | |
| }, | |
| { | |
| "epoch": 19.65, | |
| "learning_rate": 1.937457511896669e-06, | |
| "loss": 0.038, | |
| "step": 62700 | |
| }, | |
| { | |
| "epoch": 19.68, | |
| "learning_rate": 1.7675050985723998e-06, | |
| "loss": 0.0393, | |
| "step": 62800 | |
| }, | |
| { | |
| "epoch": 19.71, | |
| "learning_rate": 1.5975526852481306e-06, | |
| "loss": 0.0392, | |
| "step": 62900 | |
| }, | |
| { | |
| "epoch": 19.74, | |
| "learning_rate": 1.4276002719238613e-06, | |
| "loss": 0.0392, | |
| "step": 63000 | |
| }, | |
| { | |
| "epoch": 19.77, | |
| "learning_rate": 1.2576478585995922e-06, | |
| "loss": 0.0388, | |
| "step": 63100 | |
| }, | |
| { | |
| "epoch": 19.8, | |
| "learning_rate": 1.0876954452753228e-06, | |
| "loss": 0.0388, | |
| "step": 63200 | |
| }, | |
| { | |
| "epoch": 19.83, | |
| "learning_rate": 9.177430319510537e-07, | |
| "loss": 0.0375, | |
| "step": 63300 | |
| }, | |
| { | |
| "epoch": 19.87, | |
| "learning_rate": 7.477906186267846e-07, | |
| "loss": 0.0392, | |
| "step": 63400 | |
| }, | |
| { | |
| "epoch": 19.9, | |
| "learning_rate": 5.778382053025154e-07, | |
| "loss": 0.0383, | |
| "step": 63500 | |
| }, | |
| { | |
| "epoch": 19.93, | |
| "learning_rate": 4.0788579197824615e-07, | |
| "loss": 0.0377, | |
| "step": 63600 | |
| }, | |
| { | |
| "epoch": 19.96, | |
| "learning_rate": 2.3793337865397692e-07, | |
| "loss": 0.0378, | |
| "step": 63700 | |
| }, | |
| { | |
| "epoch": 19.99, | |
| "learning_rate": 6.798096532970768e-08, | |
| "loss": 0.0385, | |
| "step": 63800 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "step": 63840, | |
| "total_flos": 1.325011889553408e+20, | |
| "train_loss": 0.13789003552836285, | |
| "train_runtime": 208335.1784, | |
| "train_samples_per_second": 78.446, | |
| "train_steps_per_second": 0.306 | |
| } | |
| ], | |
| "max_steps": 63840, | |
| "num_train_epochs": 21, | |
| "total_flos": 1.325011889553408e+20, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |