[ORPO] add the log odds part to the loss
What does this PR do?
Adds the missing log-odds part to the loss
The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.
This issue has been automatically marked as stale because it has not had recent activity. If you think this still needs to be addressed please comment on this thread.
Please note that issues that do not follow the contributing guidelines are likely to be ignored.
@sayakpaul Shouldn't the log_odds be -(model_losses_w - model_losses_l)?
Referring the pdf of gaussian distribution, log_prob is akin to -mse not mse.
This issue has been automatically marked as stale because it has not had recent activity. If you think this still needs to be addressed please comment on this thread.
Please note that issues that do not follow the contributing guidelines are likely to be ignored.