{"ID":5438770,"CreatedAt":"2026-07-01T01:17:58.482524686Z","UpdatedAt":"2026-07-03T09:10:46.706950747Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2606.31413","arxiv_id":"2606.31413","title":"Learning to Select, Not Relearn: Hard-Routed Mixtures of Reasoning LoRAs","abstract":"Composing independently trained LoRA adapters into a single large language model is useful for multi-domain adaptation, especially when the original training data cannot be shared. A common approach is to use MoE-style routing over LoRA experts, but for frozen pretrained adapters, soft weighted combinations can change the unit-scale additive update under which each LoRA module was originally trained. We propose \\textbf{Hard-Routed MoR-LoRA}, a two-stage framework for composing frozen reasoning LoRA experts through unit-scale hard selection. First, domain-specific LoRA adapters are trained independently using reinforcement learning from verifiable feedback to obtain reasoning experts. Then, all experts are frozen, reasoning traces are distilled from them, and only a lightweight shared router together with a small attention LoRA is trained for integration. The router selects exactly one expert per token using hard top-1 routing, while a straight-through estimator enables gradient-based training. Experiments across five benchmarks, multiple model scales, and additional model families show that Hard-Routed MoR-LoRA preserves expert behavior while requiring substantially fewer trainable parameters than soft-routing mixture baselines. Our analysis further shows that normalized soft mixtures often concentrate most routing mass on a single expert, suggesting that hard unit-scale routing provides a simple and efficient abstraction for frozen LoRA expert composition.","short_abstract":"Composing independently trained LoRA adapters into a single large language model is useful for multi-domain adaptation, especially when the original training data cannot be shared. A common approach is to use MoE-style routing over LoRA experts, but for frozen pretrained adapters, soft weighted combinations can change...","url_abs":"https://arxiv.org/abs/2606.31413","url_pdf":"https://arxiv.org/pdf/2606.31413v1","authors":"[\"Seyed Alireza Molavi\",\"Zhan Su\",\"Yan Hu\",\"Peyman Sheikholharam Mashhadi\",\"Stefan Byttner\",\"Prayag Tiwari\"]","published":"2026-06-30T09:40:38Z","proceeding":"cs.AI","tasks":"[\"cs.AI\",\"cs.LG\"]","methods":"[\"Reinforcement Learning\",\"Language Model\",\"LoRA\"]","has_code":false}