| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "global_step": 20000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0002, | |
| "loss": 1.8051, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0002, | |
| "loss": 1.7655, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0002, | |
| "loss": 1.7675, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0002, | |
| "loss": 1.7158, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0002, | |
| "loss": 1.7021, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0002, | |
| "loss": 1.6921, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0002, | |
| "loss": 1.7088, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0002, | |
| "loss": 1.6985, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0002, | |
| "loss": 1.6791, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0002, | |
| "loss": 1.6797, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0002, | |
| "loss": 1.6752, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0002, | |
| "loss": 1.6761, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0002, | |
| "loss": 1.6278, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0002, | |
| "loss": 1.6576, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0002, | |
| "loss": 1.6409, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0002, | |
| "loss": 1.6454, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0002, | |
| "loss": 1.5936, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0002, | |
| "loss": 1.621, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0002, | |
| "loss": 1.6192, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0002, | |
| "loss": 1.5758, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0002, | |
| "loss": 1.6007, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0002, | |
| "loss": 1.5688, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0002, | |
| "loss": 1.5973, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0002, | |
| "loss": 1.5517, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0002, | |
| "loss": 1.5694, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0002, | |
| "loss": 1.5177, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0002, | |
| "loss": 1.5637, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0002, | |
| "loss": 1.5593, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0002, | |
| "loss": 1.5087, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0002, | |
| "loss": 1.5299, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0002, | |
| "loss": 1.5264, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0002, | |
| "loss": 1.5103, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0002, | |
| "loss": 1.5064, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0002, | |
| "loss": 1.5014, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0002, | |
| "loss": 1.5104, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0002, | |
| "loss": 1.4832, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0002, | |
| "loss": 1.5121, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.0002, | |
| "loss": 1.4547, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0002, | |
| "loss": 1.4498, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0002, | |
| "loss": 1.4706, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0002, | |
| "loss": 1.4534, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0002, | |
| "loss": 1.4664, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0002, | |
| "loss": 1.4408, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.0002, | |
| "loss": 1.4809, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.0002, | |
| "loss": 1.4656, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.0002, | |
| "loss": 1.4435, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.0002, | |
| "loss": 1.4093, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.0002, | |
| "loss": 1.4241, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.0002, | |
| "loss": 1.439, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.0002, | |
| "loss": 1.3942, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0002, | |
| "loss": 1.4163, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0002, | |
| "loss": 1.4359, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.0002, | |
| "loss": 1.4112, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.0002, | |
| "loss": 1.4354, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.0002, | |
| "loss": 1.4081, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.0002, | |
| "loss": 1.3576, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.0002, | |
| "loss": 1.4104, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.0002, | |
| "loss": 1.3598, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.0002, | |
| "loss": 1.4263, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0002, | |
| "loss": 1.371, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0002, | |
| "loss": 1.3493, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.0002, | |
| "loss": 1.396, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0002, | |
| "loss": 1.387, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0002, | |
| "loss": 1.3604, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.0002, | |
| "loss": 1.3396, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.0002, | |
| "loss": 1.3333, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.0002, | |
| "loss": 1.3497, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.0002, | |
| "loss": 1.3387, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.0002, | |
| "loss": 1.3431, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.0002, | |
| "loss": 1.3469, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.0002, | |
| "loss": 1.3268, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.0002, | |
| "loss": 1.3166, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.0002, | |
| "loss": 1.3069, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.0002, | |
| "loss": 1.3503, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.0002, | |
| "loss": 1.3374, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.0002, | |
| "loss": 1.3348, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.0002, | |
| "loss": 1.3306, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.0002, | |
| "loss": 1.3174, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.0002, | |
| "loss": 1.2756, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.0002, | |
| "loss": 1.298, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.0002, | |
| "loss": 1.2558, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.0002, | |
| "loss": 1.2841, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.0002, | |
| "loss": 1.2717, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.0002, | |
| "loss": 1.3095, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.0002, | |
| "loss": 1.2651, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.0002, | |
| "loss": 1.2514, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.0002, | |
| "loss": 1.251, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.0002, | |
| "loss": 1.2891, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.0002, | |
| "loss": 1.2787, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.0002, | |
| "loss": 1.2872, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.0002, | |
| "loss": 1.2401, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.0002, | |
| "loss": 1.2464, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.0002, | |
| "loss": 1.2558, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.0002, | |
| "loss": 1.2351, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.0002, | |
| "loss": 1.2528, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.0002, | |
| "loss": 1.2324, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.0002, | |
| "loss": 1.2067, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.0002, | |
| "loss": 1.223, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.0002, | |
| "loss": 1.2198, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.0002, | |
| "loss": 1.2219, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.0002, | |
| "loss": 1.1942, | |
| "step": 10100 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.0002, | |
| "loss": 1.2175, | |
| "step": 10200 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 0.0002, | |
| "loss": 1.1902, | |
| "step": 10300 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 0.0002, | |
| "loss": 1.1868, | |
| "step": 10400 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.0002, | |
| "loss": 1.2066, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.0002, | |
| "loss": 1.183, | |
| "step": 10600 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 0.0002, | |
| "loss": 1.1652, | |
| "step": 10700 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 0.0002, | |
| "loss": 1.1789, | |
| "step": 10800 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.0002, | |
| "loss": 1.1821, | |
| "step": 10900 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.0002, | |
| "loss": 1.1542, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.0002, | |
| "loss": 1.1606, | |
| "step": 11100 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.0002, | |
| "loss": 1.2044, | |
| "step": 11200 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.0002, | |
| "loss": 1.1583, | |
| "step": 11300 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 0.0002, | |
| "loss": 1.2191, | |
| "step": 11400 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 0.0002, | |
| "loss": 1.1519, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.0002, | |
| "loss": 1.1523, | |
| "step": 11600 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.0002, | |
| "loss": 1.1948, | |
| "step": 11700 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.0002, | |
| "loss": 1.1603, | |
| "step": 11800 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.0002, | |
| "loss": 1.1601, | |
| "step": 11900 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.0002, | |
| "loss": 1.1245, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.0002, | |
| "loss": 1.1801, | |
| "step": 12100 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.0002, | |
| "loss": 1.1747, | |
| "step": 12200 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.0002, | |
| "loss": 1.1694, | |
| "step": 12300 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 0.0002, | |
| "loss": 1.1405, | |
| "step": 12400 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 0.0002, | |
| "loss": 1.1568, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 0.0002, | |
| "loss": 1.1642, | |
| "step": 12600 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 0.0002, | |
| "loss": 1.1422, | |
| "step": 12700 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 0.0002, | |
| "loss": 1.0952, | |
| "step": 12800 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 0.0002, | |
| "loss": 1.1208, | |
| "step": 12900 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 0.0002, | |
| "loss": 1.1284, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 0.0002, | |
| "loss": 1.1416, | |
| "step": 13100 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 0.0002, | |
| "loss": 1.1485, | |
| "step": 13200 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 0.0002, | |
| "loss": 1.1013, | |
| "step": 13300 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 0.0002, | |
| "loss": 1.1042, | |
| "step": 13400 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 0.0002, | |
| "loss": 1.0657, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 0.0002, | |
| "loss": 1.0916, | |
| "step": 13600 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 0.0002, | |
| "loss": 1.102, | |
| "step": 13700 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 0.0002, | |
| "loss": 1.1345, | |
| "step": 13800 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 0.0002, | |
| "loss": 1.1131, | |
| "step": 13900 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 0.0002, | |
| "loss": 1.115, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 0.0002, | |
| "loss": 1.0934, | |
| "step": 14100 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 0.0002, | |
| "loss": 1.0929, | |
| "step": 14200 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 0.0002, | |
| "loss": 1.1246, | |
| "step": 14300 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 0.0002, | |
| "loss": 1.1284, | |
| "step": 14400 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 0.0002, | |
| "loss": 1.0908, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 0.0002, | |
| "loss": 1.104, | |
| "step": 14600 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 0.0002, | |
| "loss": 1.0935, | |
| "step": 14700 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 0.0002, | |
| "loss": 1.125, | |
| "step": 14800 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 0.0002, | |
| "loss": 1.093, | |
| "step": 14900 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 0.0002, | |
| "loss": 1.0783, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 0.0002, | |
| "loss": 1.0933, | |
| "step": 15100 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 0.0002, | |
| "loss": 1.0803, | |
| "step": 15200 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 0.0002, | |
| "loss": 1.1047, | |
| "step": 15300 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 0.0002, | |
| "loss": 1.0661, | |
| "step": 15400 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 0.0002, | |
| "loss": 1.0404, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 0.0002, | |
| "loss": 1.0883, | |
| "step": 15600 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 0.0002, | |
| "loss": 1.0613, | |
| "step": 15700 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 0.0002, | |
| "loss": 1.0591, | |
| "step": 15800 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 0.0002, | |
| "loss": 1.0712, | |
| "step": 15900 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 0.0002, | |
| "loss": 1.0551, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 0.0002, | |
| "loss": 1.042, | |
| "step": 16100 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 0.0002, | |
| "loss": 1.0873, | |
| "step": 16200 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 0.0002, | |
| "loss": 1.0212, | |
| "step": 16300 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 0.0002, | |
| "loss": 1.0706, | |
| "step": 16400 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 0.0002, | |
| "loss": 1.0268, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 0.0002, | |
| "loss": 1.0624, | |
| "step": 16600 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 0.0002, | |
| "loss": 1.0172, | |
| "step": 16700 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 0.0002, | |
| "loss": 1.0391, | |
| "step": 16800 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 0.0002, | |
| "loss": 1.0386, | |
| "step": 16900 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 0.0002, | |
| "loss": 1.0418, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 0.0002, | |
| "loss": 1.0722, | |
| "step": 17100 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 0.0002, | |
| "loss": 1.0623, | |
| "step": 17200 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 0.0002, | |
| "loss": 1.0313, | |
| "step": 17300 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 0.0002, | |
| "loss": 1.0032, | |
| "step": 17400 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 0.0002, | |
| "loss": 1.0141, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 0.0002, | |
| "loss": 1.0268, | |
| "step": 17600 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 0.0002, | |
| "loss": 1.0475, | |
| "step": 17700 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 0.0002, | |
| "loss": 1.0664, | |
| "step": 17800 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 0.0002, | |
| "loss": 1.0096, | |
| "step": 17900 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 0.0002, | |
| "loss": 1.0158, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 0.0002, | |
| "loss": 1.0472, | |
| "step": 18100 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 0.0002, | |
| "loss": 1.0588, | |
| "step": 18200 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 0.0002, | |
| "loss": 1.0205, | |
| "step": 18300 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 0.0002, | |
| "loss": 1.0078, | |
| "step": 18400 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 0.0002, | |
| "loss": 1.0085, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 0.0002, | |
| "loss": 1.0189, | |
| "step": 18600 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 0.0002, | |
| "loss": 1.0223, | |
| "step": 18700 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 0.0002, | |
| "loss": 0.9967, | |
| "step": 18800 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 0.0002, | |
| "loss": 0.9659, | |
| "step": 18900 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 0.0002, | |
| "loss": 1.0014, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 0.0002, | |
| "loss": 1.0076, | |
| "step": 19100 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 0.0002, | |
| "loss": 1.0119, | |
| "step": 19200 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 0.0002, | |
| "loss": 1.0287, | |
| "step": 19300 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 0.0002, | |
| "loss": 0.99, | |
| "step": 19400 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 0.0002, | |
| "loss": 1.0108, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 0.0002, | |
| "loss": 0.9991, | |
| "step": 19600 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 0.0002, | |
| "loss": 0.9656, | |
| "step": 19700 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 0.0002, | |
| "loss": 0.9977, | |
| "step": 19800 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 0.0002, | |
| "loss": 0.9915, | |
| "step": 19900 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 0.0002, | |
| "loss": 0.9808, | |
| "step": 20000 | |
| } | |
| ], | |
| "max_steps": 20000, | |
| "num_train_epochs": 1, | |
| "total_flos": 3.5798846029020365e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |