Sajib-006/sevra-recovery-outcomes

SEVRA Recovery Outcomes Intervention outcomes used to train and evaluate SEVRA, the recoverability-aware controller from Think Again or Think Longer? Selective Verification for Budget-Aware Reasoning. Each row pairs a frozen Qwen3-4B base attempt with one post-generation action and records whether that action repaired an incorrect answer (helpful_fix), damaged a correct answer (harmful_flip), and how many realized tokens it consumed. Gold answers were used to compute offline… See the full description on the dataset page: https://huggingface.co/datasets/Sajib-006/sevra-recovery-outcomes.

<iframe src="https://silytics.si/embed/item/ccc2e06b-8abf-4617-9788-6ae0b9a292a0" width="420" height="190" style="border:0" loading="lazy" title="Silytics entry"></iframe>

Silytics shows public metadata from Hugging Face and links to the original. Information may be incomplete or out of date. Last refreshed 18 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.