Spaces:
Sleeping
Sleeping
Commit History
-
fixes done in env
121dc28
-
Implement RL recovery plan for compliance training.
Align environment curriculum, graders, dataset generation, and training scripts with GRPO/SFT eval flow; extend tests and Colab notebook publish guidance. Co-authored-by: Cursor <cursoragent@cursor.com>e54ca85
-
Colab-ready Unsloth training, curriculum env fixes, and eval tooling.
Add split-aware dataset generation, component graders, in-process GRPO rollouts, local eval, training smoke tests, and updated docs. Stop tracking generated training artifacts and remove stale validation script from data/. Co-authored-by: Cursor <cursoragent@cursor.com>6111cca
-
fix: add step-aware prompting with urgency to prevent LLM loops
0503d9c -
Deploy to HuggingFace Spaces: Dockerfile, tests, and production-ready environment
4e4c595