相关论文: NOSA: Native and Offloadable Sparse Attention

Attention mechanisms are the core of foundation models, but their quadratic complexity remains a critical bottleneck for scaling. This challenge has driven the development of efficient attention mechanisms, with sparsity emerging as the…

Neural Processes (NPs) are a rapidly evolving class of models designed to directly model the posterior predictive distribution of stochastic processes. While early architectures were developed primarily as a scalable alternative to Gaussian…

Sparse attention improves LLM inference efficiency by selecting a subset of key-value entries, but at the cost of potential accuracy degradation. In particular, omitting critical KV entries can induce substantial errors in model outputs.…

Being an effective non-orthogonal multiple access (NOMA) technique, sparse code multiple access (SCMA) is promising for future wireless communication. Compared with orthogonal techniques, SCMA enjoys higher overloading tolerance and lower…

Large Language Models (LLMs) are increasingly deployed in long-context tasks such as reasoning, code generation, and multi-turn dialogue. However, inference over extended contexts is bottlenecked by the Key-Value (KV) cache, whose memory…

Long contexts improve capabilities of large language models but pose serious hardware challenges: compute and memory footprints grow linearly with sequence length. Particularly, the decoding phase continuously accesses massive KV cache,…

KV cache pruning has emerged as a promising technique for reducing memory and computation costs in long-context auto-regressive generation. Existing methods for vision-language models (VLMs) typically rely on self-attention scores from…

Memory consumption of the Key-Value (KV) cache represents a major bottleneck for efficient large language model inference. While attention-score-based KV cache pruning shows promise, it faces critical practical limitations: attention scores…

The last few years have seen gigantic leaps in algorithms and systems to support efficient deep learning inference. Pruning and quantization algorithms can now consistently compress neural networks by an order of magnitude. For a compressed…

Sparse Attention is a technique that approximates standard attention computation with sub-quadratic complexity. This is achieved by selectively ignoring smaller entries in the attention matrix during the softmax function computation.…

Autoregressive decoding becomes bandwidth-limited at long contexts, as generating each token requires reading all $n_k$ key and value vectors from KV cache. We present Stochastic Additive No-mulT Attention (SANTA), a method that sparsifies…

Chunked prefill has become a widely adopted serving strategy for long-context large language models, but efficient attention computation in this regime remains challenging. Existing sparse attention methods are primarily designed for…

Diffusion Transformers (DiTs) achieve strong video generation quality but suffer from high inference cost due to dense 3D attention, motivating sparse attention techniques for improving efficiency. However, existing training-free sparse…

DeepSeek Sparse Attention (DSA) sets the state of the art for fine-grained inference-time sparse attention by introducing a learned token-wise indexer that scores every prefix token and selects the most relevant ones for the main attention.…

As the accuracy of machine learning models increases at a fast rate, so does their demand for energy and compute resources. On a low level, the major part of these resources is consumed by data movement between different memory units.…

As large language models (LLMs) continue to support increasingly longer contexts, the memory demand for key-value (KV) caches during decoding grows rapidly, becoming a critical bottleneck in both GPU memory capacity and PCIe bandwidth.…

While CNNs naturally lend themselves to densely sampled data, and sophisticated implementations are available, they lack the ability to efficiently process sparse data. In this work we introduce a suite of tools that exploit sparsity in…

Efficient inference is critical for long-context language models, where attention computation and KV-cache access dominate the cost. Recent work RAT+, introduces a recurrence-augmented attention backbone that enables flexible dilated…

KV cache quantization reduces the memory cost of long-context LLM inference, but introduces approximation error that is typically validated only empirically. Existing systems rely on average-case robustness, with no mechanism to detect or…

Multi-head-self-attention (MHSA) mechanisms achieve state-of-the-art (SOTA) performance across natural language processing and vision tasks. However, their quadratic dependence on sequence lengths has bottlenecked inference speeds. To…