Winning without optimization / SFT, by modeling (outcome, moves) pairs #1
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
pHaeusler
reviewed
Sep 25, 2023
generate_data.py
Outdated
| if __name__ == "__main__": | ||
| # Use typer instead | ||
| import typer | ||
| app = typer.Typer() |
There was a problem hiding this comment.
is app needed here?
also, let's import at the top of the file
pHaeusler
reviewed
Sep 25, 2023
requirements.txt
Outdated
| torch | ||
| typer | ||
| wandb | ||
| chardet |
There was a problem hiding this comment.
This was likely just an issue with my conda env; removing that
pHaeusler
reviewed
Sep 25, 2023
generate_data.py
Outdated
| @@ -53,29 +53,52 @@ def seq_to_board(seq): | |||
| return board | |||
|
|
|||
|
|
|||
| def save_data(trajectories): | |||
| w_map = {-1: 0, 1: 1, None: 2} | |||
There was a problem hiding this comment.
Should probably use new tokens for these. (declare in tokens.py)
pHaeusler
approved these changes
Sep 29, 2023
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Add the
incl_winnerflag which adds the outcome (player 1 win, player 2 win, draw) in front of the list of moves. Training on the outcome enables conditioning on it at inference time; conditioning on (player 1 win) increases the win rate from under 50% to around 63%. This is somewhat inspired by Decision Transformers.