Knowledge-weighted fine-tuning uses multi-sampled inference to estimate per-query knowledge and adjusts training to make models express uncertainty appropriately on out-of-scope questions.
In Findings of the Association for Computational Lin- guistics: ACL 2025, pages 24085–24100, Vienna, Austria
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
What Models Know, How Well They Know It: Knowledge-Weighted Fine-Tuning for Learning When to Say "I Don't Know"
Knowledge-weighted fine-tuning uses multi-sampled inference to estimate per-query knowledge and adjusts training to make models express uncertainty appropriately on out-of-scope questions.