Train loss and val loss is coming as a seperate entity #3122
Replies: 1 comment 3 replies
-
It is because we perform training and evaluation at different time, it is reasonable |
Beta Was this translation helpful? Give feedback.
3 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
{"current_steps": 4, "total_steps": 44, "loss": 0.4112, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.898732434036244e-05, "epoch": 0.18, "percentage": 9.09, "elapsed_time": "0:00:06", "remaining_time": "0:01:01"}
{"current_steps": 4, "total_steps": 44, "loss": null, "eval_loss": 0.23153726756572723, "predict_loss": null, "reward": null, "learning_rate": null, "epoch": 0.18, "percentage": 9.09, "elapsed_time": "0:00:06", "remaining_time": "0:01:01"}
{"current_steps": 8, "total_steps": 44, "loss": 0.1252, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.6031338320779534e-05, "epoch": 0.36, "percentage": 18.18, "elapsed_time": "0:04:19", "remaining_time": "0:19:28"}
{"current_steps": 8, "total_steps": 44, "loss": null, "eval_loss": 0.1610538214445114, "predict_loss": null, "reward": null, "learning_rate": null, "epoch": 0.36, "percentage": 18.18, "elapsed_time": "0:04:19", "remaining_time": "0:19:28"}
{"current_steps": 12, "total_steps": 44, "loss": 0.05, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.137151834863213e-05, "epoch": 0.55, "percentage": 27.27, "elapsed_time": "0:06:34", "remaining_time": "0:17:33"}
{"current_steps": 12, "total_steps": 44, "loss": null, "eval_loss": 0.1615099310874939, "predict_loss": null, "reward": null, "learning_rate": null, "epoch": 0.55, "percentage": 27.27, "elapsed_time": "0:06:34", "remaining_time": "0:17:33"}
{"current_steps": 16, "total_steps": 44, "loss": 0.3448, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 3.5385375325047166e-05, "epoch": 0.73, "percentage": 36.36, "elapsed_time": "0:10:08", "remaining_time": "0:17:45"}
{"current_steps": 16, "total_steps": 44, "loss": null, "eval_loss": 0.13311704993247986, "predict_loss": null, "reward": null, "learning_rate": null, "epoch": 0.73, "percentage": 36.36, "elapsed_time": "0:10:08", "remaining_time": "0:17:45"}
While finetuning Dupilcate entries are being done One for train_loss and one for eval_loss. Can't we get both on the same line.
Beta Was this translation helpful? Give feedback.
All reactions