The authors developed HACE, a framework with two main components: a specialized foundation model (the specialist) and a heterogeneous LLM agent.
Coevolving specialized models and LLM agents for autonomous self-improvement
HACE jointly trains a specialist foundation model and a heterogeneous LLM agent through reciprocal learning, outperforming existing methods in recommendation, protein function prediction, and crystal stability.
Big Tech
Wenqian Ye · Ziwei Guan · Zi Yang · Shivani Modi · Cici Cong Wu · Yunlong He · +11 more
Meta AI · University of Virginia
Research Digest··3 min read
Thread:Agent Self-Improvement
Ye et al.
Why this paper
From Meta AI and University of Virginia · Part of Agent Self-Improvement, now 52 papers
In one line
Heterogeneous Agentic Coevolution jointly evolves specialized foundation models and LLM agents through reciprocal learning to improve domain-specific performance.
What we could check
- ·No code link found
- ·No weights link found
- ·No dataset link found
- ·No compute details found
- ·No stated limitations found
- ·No benchmark numbers found
Observed from the paper text and links we have. Absence here means we did not find it, not that it does not exist.
§