NousResearch/atropos
1
0
Fork 0
mirror of https://github.com/NousResearch/atropos.git synced 2026-05-01 17:45:16 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Add BLEUBERI environment for reference-based RL

Browse source
This commit is contained in:
Allan Niemerg 2025-06-08 18:02:33 -05:00
parent 3f6015e622
commit 5bb5bd2c3d
7 changed files with 948 additions and 0 deletions
Download patch file Download diff file Expand all files Collapse all files

12
environments/bleuberi/__init__.py Normal file
View file

@ -0,0 +1,12 @@
"""
BLEUBERI: BLEU-based environment for instruction following.
This environment uses BLEU scores as a reward function for training
models to follow instructions. Based on the paper:
"BLEUBERI: BLEU is a surprisingly effective reward for instruction following"
https://arxiv.org/abs/2505.11080
"""
__all__ = ["BLEUBERIEnv"]
from .bleuberi_env import BLEUBERIEnv # noqa