trainer_log.jsonl

{"current_steps": 10, "total_steps": 250, "loss": 2.124, "learning_rate": 4.000000000000001e-06, "epoch": 0.08, "percentage": 4.0, "elapsed_time": "0:00:21", "remaining_time": "0:08:24", "throughput": "0.00", "total_tokens": 0}
{"current_steps": 20, "total_steps": 250, "loss": 1.971, "learning_rate": 8.000000000000001e-06, "epoch": 0.16, "percentage": 8.0, "elapsed_time": "0:00:38", "remaining_time": "0:07:25", "throughput": "0.00", "total_tokens": 0}
{"current_steps": 30, "total_steps": 250, "loss": 1.8813, "learning_rate": 9.987820251299121e-06, "epoch": 0.24, "percentage": 12.0, "elapsed_time": "0:00:56", "remaining_time": "0:06:56", "throughput": "0.00", "total_tokens": 0}
{"current_steps": 40, "total_steps": 250, "loss": 1.8264, "learning_rate": 9.890738003669029e-06, "epoch": 0.32, "percentage": 16.0, "elapsed_time": "0:01:14", "remaining_time": "0:06:30", "throughput": "0.00", "total_tokens": 0}
{"current_steps": 50, "total_steps": 250, "loss": 1.7815, "learning_rate": 9.698463103929542e-06, "epoch": 0.4, "percentage": 20.0, "elapsed_time": "0:01:31", "remaining_time": "0:06:04", "throughput": "0.00", "total_tokens": 0}
{"current_steps": 60, "total_steps": 250, "loss": 1.7795, "learning_rate": 9.414737964294636e-06, "epoch": 0.48, "percentage": 24.0, "elapsed_time": "0:01:48", "remaining_time": "0:05:44", "throughput": "0.00", "total_tokens": 0}
{"current_steps": 70, "total_steps": 250, "loss": 1.7885, "learning_rate": 9.045084971874738e-06, "epoch": 0.56, "percentage": 28.0, "elapsed_time": "0:02:05", "remaining_time": "0:05:23", "throughput": "0.00", "total_tokens": 0}
{"current_steps": 80, "total_steps": 250, "loss": 1.771, "learning_rate": 8.596699001693257e-06, "epoch": 0.64, "percentage": 32.0, "elapsed_time": "0:02:24", "remaining_time": "0:05:06", "throughput": "0.00", "total_tokens": 0}
{"current_steps": 90, "total_steps": 250, "loss": 1.7706, "learning_rate": 8.078307376628292e-06, "epoch": 0.72, "percentage": 36.0, "elapsed_time": "0:02:41", "remaining_time": "0:04:47", "throughput": "0.00", "total_tokens": 0}
{"current_steps": 100, "total_steps": 250, "loss": 1.7634, "learning_rate": 7.500000000000001e-06, "epoch": 0.8, "percentage": 40.0, "elapsed_time": "0:02:59", "remaining_time": "0:04:29", "throughput": "0.00", "total_tokens": 0}
{"current_steps": 110, "total_steps": 250, "loss": 1.7241, "learning_rate": 6.873032967079562e-06, "epoch": 0.88, "percentage": 44.0, "elapsed_time": "0:03:16", "remaining_time": "0:04:10", "throughput": "0.00", "total_tokens": 0}
{"current_steps": 120, "total_steps": 250, "loss": 1.7287, "learning_rate": 6.209609477998339e-06, "epoch": 0.96, "percentage": 48.0, "elapsed_time": "0:03:33", "remaining_time": "0:03:51", "throughput": "0.00", "total_tokens": 0}
{"current_steps": 130, "total_steps": 250, "loss": 1.6224, "learning_rate": 5.522642316338268e-06, "epoch": 1.04, "percentage": 52.0, "elapsed_time": "0:03:52", "remaining_time": "0:03:34", "throughput": "0.00", "total_tokens": 0}
{"current_steps": 140, "total_steps": 250, "loss": 1.4283, "learning_rate": 4.825502516487497e-06, "epoch": 1.12, "percentage": 56.0, "elapsed_time": "0:04:09", "remaining_time": "0:03:16", "throughput": "0.00", "total_tokens": 0}
{"current_steps": 150, "total_steps": 250, "loss": 1.4763, "learning_rate": 4.131759111665349e-06, "epoch": 1.2, "percentage": 60.0, "elapsed_time": "0:04:28", "remaining_time": "0:02:58", "throughput": "0.00", "total_tokens": 0}
{"current_steps": 160, "total_steps": 250, "loss": 1.4036, "learning_rate": 3.4549150281252635e-06, "epoch": 1.28, "percentage": 64.0, "elapsed_time": "0:04:45", "remaining_time": "0:02:40", "throughput": "0.00", "total_tokens": 0}
{"current_steps": 170, "total_steps": 250, "loss": 1.4223, "learning_rate": 2.8081442660546126e-06, "epoch": 1.3599999999999999, "percentage": 68.0, "elapsed_time": "0:05:03", "remaining_time": "0:02:22", "throughput": "0.00", "total_tokens": 0}
{"current_steps": 180, "total_steps": 250, "loss": 1.416, "learning_rate": 2.204035482646267e-06, "epoch": 1.44, "percentage": 72.0, "elapsed_time": "0:05:21", "remaining_time": "0:02:04", "throughput": "0.00", "total_tokens": 0}
{"current_steps": 190, "total_steps": 250, "loss": 1.4118, "learning_rate": 1.6543469682057105e-06, "epoch": 1.52, "percentage": 76.0, "elapsed_time": "0:05:38", "remaining_time": "0:01:46", "throughput": "0.00", "total_tokens": 0}
{"current_steps": 200, "total_steps": 250, "loss": 1.4386, "learning_rate": 1.1697777844051105e-06, "epoch": 1.6, "percentage": 80.0, "elapsed_time": "0:05:56", "remaining_time": "0:01:29", "throughput": "0.00", "total_tokens": 0}
{"current_steps": 210, "total_steps": 250, "loss": 1.4189, "learning_rate": 7.597595192178702e-07, "epoch": 1.6800000000000002, "percentage": 84.0, "elapsed_time": "0:06:14", "remaining_time": "0:01:11", "throughput": "0.00", "total_tokens": 0}
{"current_steps": 220, "total_steps": 250, "loss": 1.4129, "learning_rate": 4.322727117869951e-07, "epoch": 1.76, "percentage": 88.0, "elapsed_time": "0:06:32", "remaining_time": "0:00:53", "throughput": "0.00", "total_tokens": 0}
{"current_steps": 230, "total_steps": 250, "loss": 1.388, "learning_rate": 1.9369152030840553e-07, "epoch": 1.8399999999999999, "percentage": 92.0, "elapsed_time": "0:06:50", "remaining_time": "0:00:35", "throughput": "0.00", "total_tokens": 0}
{"current_steps": 240, "total_steps": 250, "loss": 1.3457, "learning_rate": 4.865965629214819e-08, "epoch": 1.92, "percentage": 96.0, "elapsed_time": "0:07:06", "remaining_time": "0:00:17", "throughput": "0.00", "total_tokens": 0}
{"current_steps": 250, "total_steps": 250, "loss": 1.4088, "learning_rate": 0.0, "epoch": 2.0, "percentage": 100.0, "elapsed_time": "0:07:24", "remaining_time": "0:00:00", "throughput": "0.00", "total_tokens": 0}
{"current_steps": 250, "total_steps": 250, "epoch": 2.0, "percentage": 100.0, "elapsed_time": "0:08:23", "remaining_time": "0:00:00", "throughput": "0.00", "total_tokens": 0}