Abstract:Combining diverse foundation models is promising, but weight-merging is limited by mismatched architectures and closed APIs. Trinity addresses this with a lightweight coordinator that orchestrates collaboration among large language models (LLMs). The coordinator, comprising a compact language model (approximately $0.6$B parameters) and a lightweight head (approximately $10$K parameters), is optimized with an evolutionary strategy for efficient and adaptive delegation. Trinity processes queries over multiple turns, where at each turn the coordinator assigns one of three roles (Thinker, Worker, or Verifier) to a selected LLM, effectively offloading complex skill acquisition from the coordinator itself. Experiments show that Trinity consistently outperforms individual models and existing methods across coding, math, reasoning, and domain knowledge tasks, and generalizes robustly to out-of-distribution tasks. On standard benchmarks, Trinity achieves state-of-the-art results, including a score of 86.2% on LiveCodeBench. Theoretical and empirical analyses identify two main factors behind this performance: (1) the coordinator's hidden-state representations provide rich contextualization of inputs, and (2) under high dimensionality and strict budget constraints, the separable Covariance Matrix Adaptation Evolution Strategy offers advantages over reinforcement learning, imitation learning, and random search by exploiting potential block-epsilon-separability.
| Comments: | To appear at the 14th International Conference on Learning Representation (ICLR 2026) |
| Subjects: | Machine Learning (cs.LG) |
| Cite as: | arXiv:2512.04695 [cs.LG] |
| (or arXiv:2512.04695v3 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2512.04695 arXiv-issued DOI via DataCite |
From: Jinglue Xu [view email]
[v1]
Thu, 4 Dec 2025 11:45:21 UTC (10,646 KB)
[v2]
Mon, 2 Mar 2026 03:04:07 UTC (10,638 KB)
[v3]
Mon, 27 Apr 2026 04:31:24 UTC (10,646 KB)
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。