J-ICL is proposed, a post-training framework that incorporates context validation into the training objective and improves ICLEval and reality accuracy by an average of 5.84 percentage points and reality accuracy by 9.20 points over the corresponding base models.
Pei-Lin Li, Qing-Le Liu, Jun-Yang Feng et al.· 0 citations
Signed Rescue Routing is introduced, a budgeted routing method that predicts these two events separately and ranks requests by their difference and shows that predicting incremental value, rather than model uncertainty, is a simple and effective objective for efficient LLM cascades.
Zhe-Yuan Wang, Si-Yu Li, Pei-Qiao Song et al.· 1 citation
This work introduces Coarse-to-fine Error-aware Dynamic Attention Routing (CEDAR), a coarse-to-fine method that keeps the language model frozen while preserving global coverage and derives an output-error bound governed by within-chunk key/value dispersion and uses it to allocate a variable refinement budget.
Si-Yu Li, Dong Wang, Jie Zhou et al.· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.