{"version":1,"type":"story","url":"https://digestai.news/story/researchers-propose-joint-subspace-sparse-attention-for-faster-long-co","json":"https://digestai.news/story/researchers-propose-joint-subspace-sparse-attention-for-faster-long-co.json","markdown":"https://digestai.news/story/researchers-propose-joint-subspace-sparse-attention-for-faster-long-co.md","slug":"researchers-propose-joint-subspace-sparse-attention-for-faster-long-co","headline":"Researchers propose Joint Subspace Sparse Attention for faster long-context decoding","summary":"A new paper on arXiv introduces **Joint Subspace Sparse Attention (JSSA)**, a method to speed up long-context decoding in large language models. The technique reduces **KV-cache traffic**—a bottleneck for long prompts—by selecting tokens at a fine granularity while preserving accuracy. Unlike prior methods, JSSA avoids **latency trade-offs** by learning a shared low-dimensional subspace for query-key alignment and using a **gather-free indexed-attention kernel** that works within standard attention architectures like GQA and MLA.\n\nOn an **NVIDIA B200 GPU**, JSSA achieves **6.43× faster decoding** than full-KV attention at **128K context length** while maintaining reasoning accuracy comparable to model-native DSA. It also scales efficiently to **700B-scale models**, retaining **96–99% of full-KV throughput** even at high context lengths. The authors claim the method avoids **generation-length inflation** and architecture-dependent failures seen in competing sparse attention approaches.","keyPoints":["JSSA learns a shared query-key subspace to improve token selection accuracy without retraining models","Method achieves **6.43× faster decoding** than full-KV attention at **128K context** on NVIDIA B200 GPUs","Works across GQA, hybrid sliding-window, and MLA architectures with **96–99% throughput retention** at 700B scale"],"whyItMatters":"Faster long-context decoding could reduce costs for AI inference at scale, enabling cheaper high-context applications like document analysis or codebase querying. If adopted, JSSA could lower barriers for models requiring deep context without sacrificing accuracy.","category":{"slug":"research","name":"Research","url":"https://digestai.news/category/research"},"entities":{"companies":[],"models":[],"people":[]},"firstPublishedAt":"2026-10-15T04:00:00Z","updatedAt":"2026-10-15T04:00:00Z","sourceCount":1,"hasPrimarySource":true,"sources":[{"outlet":"arXiv cs.CL","title":"Joint Subspace Sparse Attention for Efficient Long-Context Decoding","url":"https://arxiv.org/abs/2610.19026","publishedAt":"2026-10-15T04:00:00Z","type":"primary","primary":true,"lead":true}],"sourceNotes":null,"discussions":[],"thread":null,"cite":{"text":"Digest AI, \"Researchers propose Joint Subspace Sparse Attention for faster long-context decoding\", 15 October 2026, https://digestai.news/story/researchers-propose-joint-subspace-sparse-attention-for-faster-long-co","publisher":"Digest AI","title":"Researchers propose Joint Subspace Sparse Attention for faster long-context decoding","datePublished":"2026-10-15T04:00:00Z","url":"https://digestai.news/story/researchers-propose-joint-subspace-sparse-attention-for-faster-long-co"},"generatedBy":"Written by Digest AI's editorial model from the linked sources; the sources are the record.","license":"Headlines, digests and key points are written by Digest AI and may be quoted with a link to the story page. Linked articles belong to their publishers. Terms: https://digestai.news/terms#reuse"}