SLA2: Sparse-Linear Attention with Learnable Routing and QAT

• URL: http://arxiv.org/abs/2602.12675v1
• Date: Fri, 13 Feb 2026 07:16:02 GMT
• Title: SLA2: Sparse-Linear Attention with Learnable Routing and QAT
• Authors: Jintao Zhang, Haoxu Wang, Kai Jiang, Kaiwen Zheng, Youhe Jiang, Ion Stoica, Jianfei Chen, Jun Zhu, Joseph E. Gonzalez,
• Abstract summary: We show that SLA2 can achieve 97% attention sparsity and deliver an 18.6x attention speedup while preserving generation quality.<n>Experiments show that SLA2 can achieve 97% attention sparsity and deliver an 18.6x attention speedup while preserving generation quality.
• Score: 86.22100800353991
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Sparse-Linear Attention (SLA) combines sparse and linear attention to accelerate diffusion models and has shown strong performance in video generation. However, (i) SLA relies on a heuristic split that assigns computations to the sparse or linear branch based on attention-weight magnitude, which can be suboptimal. Additionally, (ii) after formally analyzing the attention error in SLA, we identify a mismatch between SLA and a direct decomposition into sparse and linear attention. We propose SLA2, which introduces (I) a learnable router that dynamically selects whether each attention computation should use sparse or linear attention, (II) a more faithful and direct sparse-linear attention formulation that uses a learnable ratio to combine the sparse and linear attention branches, and (III) a sparse + low-bit attention design, where low-bit attention is introduced via quantization-aware fine-tuning to reduce quantization error. Experiments show that on video diffusion models, SLA2 can achieve 97% attention sparsity and deliver an 18.6x attention speedup while preserving generation quality.

Related papers

• Higher-order Linear Attention [59.92962330635185]
  quadratic cost of scaled dot-product attention is a central obstacle to scaling autoregressive language models to long contexts.<n>We introduce Higher-order Linear Attention (HLA), a causal, streaming mechanism that realizes higher interactions via compact prefix sufficient statistics.
  arXiv  Detail & Related papers  (2025-10-31T07:54:37Z)
• GraphTARIF: Linear Graph Transformer with Augmented Rank and Improved Focus [32.63390871016499]
  We propose a novel framework that enhances both the rank and focus of attention.<n>Specifically, we enhance linear attention by attaching a gated local graph network branch to the value matrix.<n>We also introduce a learnable log-power function into the attention scores to reduce entropy and sharpen focus.
  arXiv  Detail & Related papers  (2025-10-12T14:22:32Z)
• SLA: Beyond Sparsity in Diffusion Transformers via Fine-Tunable Sparse-Linear Attention [88.47701139980636]
  In Diffusion Transformer (DiT) models, particularly for video generation, attention latency is a major bottleneck.<n>We find that attention weights can be separated into two parts: a small fraction of large weights with high rank and the remaining weights with very low rank.<n>We propose SLA, a trainable attention method that fuses sparse and linear attention to accelerate diffusion models.
  arXiv  Detail & Related papers  (2025-09-28T17:58:59Z)
• Bridging the Divide: Reconsidering Softmax and Linear Attention [116.34723260730405]
  We present two key perspectives to understand and alleviate the limitations of linear attention.<n>We prove that linear attention is not injective, which is prone to assign identical attention weights to different query vectors.<n> Secondly, we confirm that effective local modeling is essential for the success of Softmax attention, in which linear attention falls short.
  arXiv  Detail & Related papers  (2024-12-09T15:44:22Z)
• Breaking the Low-Rank Dilemma of Linear Attention [61.55583836370135]
  Linear attention provides a far more efficient solution by reducing the complexity to linear levels.<n>Our experiments indicate that this performance drop is due to the low-rank nature of linear attention's feature map.<n>We introduce Rank-Augmented Linear Attention (RALA), which rivals the performance of Softmax attention while maintaining linear complexity and high efficiency.
  arXiv  Detail & Related papers  (2024-11-12T08:30:59Z)
• Short-Long Convolutions Help Hardware-Efficient Linear Attention to Focus on Long Sequences [60.489682735061415]
  We propose CHELA, which replaces state space models with short-long convolutions and implements linear attention in a divide-and-conquer manner. Our experiments on the Long Range Arena benchmark and language modeling tasks demonstrate the effectiveness of the proposed method.
  arXiv  Detail & Related papers  (2024-06-12T12:12:38Z)
• Efficient Learnable Collaborative Attention for Single Image Super-Resolution [18.955369476815136]
  Non-Local Attention (NLA) is a powerful technique for capturing long-range feature correlations in deep single image super-resolution (SR) We propose a novel Learnable Collaborative Attention (LCoA) that introduces inductive bias into non-local modeling. Our LCoA can reduce the non-local modeling time by about 83% in the inference stage.
  arXiv  Detail & Related papers  (2024-04-07T11:25:04Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.