Skip to content

Update README with RL-ready datasets information #78

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
kenjiro-mk merged 2 commits into from
Mar 4, 2026
Merged

Update README with RL-ready datasets information #78

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
06e6124
Update README with RL-ready datasets information
akshatgarg06 Feb 27, 2026
c36ea25
Update README with SAR-to-RL dataset conversion details
akshatgarg06 Mar 4, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
11 changes: 11 additions & 0 deletions preprocessing/sports/SAR_data/soccer/README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -25,3 +25,14 @@ Here are some examples of how to download and preprocess data:
- **StatsBomb and SkillCorner Data:**
- [Read the Docs Example](https://openstarlab.readthedocs.io/en/latest/Pre_Processing/Sports/SAR_data/Example/Soccer/Example_2/contents.html)
- [Example Config File](https://github.com/open-starlab/PreProcessing/blob/master/example/config/statsbomb_skillcorner/preprocessing_statsbomb_skillcorner2024.json)

## SAR-to-RL Dataset Conversion (DQN / QMIX)
This section describes a SAR-to-RL dataset conversion step that formats SAR outputs (`events.jsonl`) into tensors used by
DQN and QMIX training. This is a preprocessing/data-format step, not a training algorithm.
The conversion script is `soccer_sar_to_rl_dataset.py`.

This produces a single shared multi-agent dataset with:
- `observation`: `(B, T, N, O)` (N=10 attackers)
- `action`: `(B, T, N)` (discrete action ids; default vocab size 16 with `PAD=15`)
- `reward`, `done`, `mask`: `(B, T)`
- `onball_mask`: `(B, T, N)` (for masking unavailable actions)