AC/DC coevolves LLMs via merging and tasks via synthetic generation to produce compact expert model archives with broader benchmark coverage than larger baselines.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.AI 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Discovering Novel LLM Experts via Task-Capability Coevolution
AC/DC coevolves LLMs via merging and tasks via synthetic generation to produce compact expert model archives with broader benchmark coverage than larger baselines.