Spaces:

---

prarthana-rameshraj

/

FixOS

like 0

Running

Modified inference.py to implement clipping logic on all reported scores.
Ensured task_score, final_score, and the overall summary average are strictly greater than 0.0 and strictly less than 1.0 by applying a buffer ([0.0001, 0.9999]).
This prevents the validator from rejecting episodes with absolute success (1.0) or absolute failure (0.0).