EPISODE · Dec 17, 2025 · 26 MIN
ReFusion: A Diffusion Large Language Model with Parallel Autoregressive Decoding
from Daily Paper Cast · host Jingwen Liang, Gengyu Wang
🤗 Upvotes: 75 | cs.CL, cs.AI, cs.LG Authors: Jia-Nan Li, Jian Guan, Wei Wu, Chongxuan Li Title: ReFusion: A Diffusion Large Language Model with Parallel Autoregressive Decoding Arxiv: http://arxiv.org/abs/2512.13586v1 Abstract: Autoregressive models (ARMs) are hindered by slow sequential inference. While masked diffusion models (MDMs) offer a parallel alternative, they suffer from critical drawbacks: high computational overhead from precluding Key-Value (KV) caching, and incoherent generation arising from learning dependencies over an intractable space of token combinations. To address these limitations, we introduce ReFusion, a novel masked diffusion model that achieves superior performance and efficiency by elevating parallel decoding from the token level to a higher slot level, where each slot is a fixed-length, contiguous sub-sequence. This is achieved through an iterative ``plan-and-infill'' decoding process: a diffusion-based planning step first identifies a set of weakly dependent slots, and an autoregressive infilling step then decodes these selected slots in parallel. The slot-based design simultaneously unlocks full KV cache reuse with a unified causal framework and reduces the learning complexity from the token combination space to a manageable slot-level permutation space. Extensive experiments on seven diverse benchmarks show that ReFusion not only overwhelmingly surpasses prior MDMs with 34% performance gains and an over 18$\times$ speedup on average, but also bridges the performance gap to strong ARMs while maintaining a 2.33$\times$ average speedup.
What this episode covers
🤗 Upvotes: 75 | cs.CL, cs.AI, cs.LG Authors: Jia-Nan Li, Jian Guan, Wei Wu, Chongxuan Li Title: ReFusion: A Diffusion Large Language Model with Parallel Autoregressive Decoding Arxiv: http://arxiv.org/abs/2512.13586v1 Abstract: Autoregressive models (ARMs) are hindered by slow sequential inference. While masked diffusion models (MDMs) offer a parallel alternative, they suffer from critical drawbacks: high computational overhead from precluding Key-Value (KV) caching, and incoherent generation arising from learning dependencies over an intractable space of token combinations. To address these limitations, we introduce ReFusion, a novel masked diffusion model that achieves superior performance and efficiency by elevating parallel decoding from the token level to a higher slot level, where each slot is a fixed-length, contiguous sub-sequence. This is achieved through an iterative ``plan-and-infill'' decoding process: a diffusion-based planning step first identifies a set of weakly dependent slots, and an autoregressive infilling step then decodes these selected slots in parallel. The slot-based design simultaneously unlocks full KV cache reuse with a unified causal framework and reduces the learning complexity from the token combination space to a manageable slot-level permutation space. Extensive experiments on seven diverse benchmarks show that ReFusion not only overwhelmingly surpasses prior MDMs with 34% performance gains and an over 18$\times$ speedup on average, but also bridges the performance gap to strong ARMs while maintaining a 2.33$\times$ average speedup.
NOW PLAYING
ReFusion: A Diffusion Large Language Model with Parallel Autoregressive Decoding
No transcript for this episode yet
Similar Episodes
No similar episodes found.