v0.1.71

allenai/olmocrv0.1.71May 30, 2025by github-actions[bot]

AI Summary

Performance optimizations and benchmarking tooling improvements focused on model behavior and retry logic.

Key Highlights

Added repetition penalty to model configuration
Increased token limit to support up to 5k tokens
Improved retry performance
Added Pareto plots for benchmarking analysis

New Features

Repetition penalty
Increased context length
Enhanced retry logic
Pareto plotting

Full Release Notes

## What's new



## Commits

23f4a0e Bump version to v0.1.71 for release
8b4f6cd Upping version
24b6822 Pushing beaker images now too
208c29d Not including fallbacks in olmocr_pipeline bench runner so we can measure direct model performance better
5faf570 Format fixes
587b73f Try with more aggressive anchor changing
8f5d5bd Revert "Trying to add repetition penalty"
90f754e Trying to add repetition penalty
9dcdef6 Going to try with up to 5k tokens
8d92620 Merge remote-tracking branch 'origin/main' into retry_improvements
2cb14cc ALlowing more tokens
022be37 Some better info strings in benchmark runner
22ee068 Merge remote-tracking branch 'origin/main' into retry_improvements
fbcd82a Cleanup attempt lookup code a bit
f8fd234 Idea to improve retry performance
61d427e Repo cleanup
7a50ee1 merge
241e5bf Merge branch 'main' of github.com:allenai/olmocr
470394d pareto plot