paper-with-me

Papers

MoLoRA: Composable Specialization via Per-Token Adapter Routing

2026-03-16 · Shrey Shah, Justin Wagle arxiv

Multi-adapter serving systems route entire sequences to a single adapter, forcing a choice when requests span multiple domains. This assumption fails in two important settings: (1) multimodal generation, where text and image tokens require different adapters within the same sequence, and (2) mixed-capability requests like "write code to solve this equation," which need expertise from multiple specialized adapters. We introduce per-token routing, which routes individual tokens to adapters based on either vocabulary structure (for multimodal models) or learned gating (for semantic specialization). Per-token routing is provably optimal, achieving work N for N tokens versus K \cdot N for per-sequence routing with K adapter types. Our key contribution is MoLoRA (Mixture of LoRA), which enables composable specialization: load multiple domain-specific adapters and let a learned router select the appropriate adapter per-token. We demonstrate that specialization dramatically beats scale: MoLoRA enables Qwen3-1.7B to exceed Qwen3-8B across four reasoning benchmarks while being 4.7x smaller. This enables modular expertise at inference time: train focused LoRAs independently, combine them without retraining, and add new capabilities by simply loading new adapters.

📄 PDF Abstract BibTeX arXiv:2603.15965

Code (0)

등록된 구현이 없습니다.

Tasks

multimodal generation

Similar Papers 제목 키워드 기반

Self-Routed Tensor Adapters for Parameter-Efficient Universal Visual Adaptation

2026-08-17 · Suraj Yadav arxiv

Universal visual representations require adaptation mechanisms that adapt across heterogeneous domains without fragmenting knowledge into domain-specific modules. Parameter-efficient fine-tuning adapts frozen visual foun…

parameter-efficient fine-tuning

Monkey Jump : MoE-Style PEFT for Efficient Multi-Task Learning

2026-01-09 · Nusrat Jahan Prottasha, Md Kowsher, Chun-Nam Yu, Chen Chen 외 arxiv

Mixture-of-experts variants of parameter-efficient fine-tuning enable per-token specialization, but they introduce additional trainable routers and expert parameters, increasing memory usage and training cost. This under…

parameter-efficient fine-tuningMulti-Task Learning

Effective LoRA Adapter Routing using Task Representations

2026-01-29 · Akash Dhasade, Anne-Marie Kermarrec, Igor Pavlovic, Diana Petrescu 외 arxiv

Low-rank adaptation (LoRA) enables parameter efficient specialization of large language models (LLMs) through modular adapters, resulting in rapidly growing public adapter pools spanning diverse tasks. Effectively using …

SE-MoLoRA: Shared-Expert LoRA Adapters for Domain-Specific Photographic Assessment

2026-08-18 · Bishwash Khanal, Anlan Zhang, Sasu Tarkoma, Tommi Mikkonen 외 arxiv

Vision-language models can describe images fluently, but they often fail to provide actionable photographic critique because semantic content and aesthetic judgment remain entangled. We propose SE-MoLoRA, a modular param…

Towards an empirical understanding of MoE design choices

2024-02-20 · Dongyang Fan, Bettina Messmer, Martin Jaggi

In this study, we systematically evaluate the impact of common design choices in Mixture of Experts (MoEs) on validation performance, uncovering distinct influences at token and sequence levels. We also present empirical…

Mixture-of-Experts