Efficient Streaming Language Models with Attention Sinks

• URL: http://arxiv.org/abs/2309.17453v4
• Date: Sun, 7 Apr 2024 00:56:53 GMT
• Title: Efficient Streaming Language Models with Attention Sinks
• Authors: Guangxuan Xiao, Yuandong Tian, Beidi Chen, Song Han, Mike Lewis,
• Abstract summary: StreamingLLM is an efficient framework that enables Large Language Models to generalize to infinite sequence lengths without any fine-tuning. We show that StreamingLLM can enable Llama-2, MPT, Falcon, and Pythia to perform stable and efficient language modeling with up to 4 million tokens and more.
• Score: 72.20260088848987
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Deploying Large Language Models (LLMs) in streaming applications such as multi-round dialogue, where long interactions are expected, is urgently needed but poses two major challenges. Firstly, during the decoding stage, caching previous tokens' Key and Value states (KV) consumes extensive memory. Secondly, popular LLMs cannot generalize to longer texts than the training sequence length. Window attention, where only the most recent KVs are cached, is a natural approach -- but we show that it fails when the text length surpasses the cache size. We observe an interesting phenomenon, namely attention sink, that keeping the KV of initial tokens will largely recover the performance of window attention. In this paper, we first demonstrate that the emergence of attention sink is due to the strong attention scores towards initial tokens as a "sink" even if they are not semantically important. Based on the above analysis, we introduce StreamingLLM, an efficient framework that enables LLMs trained with a finite length attention window to generalize to infinite sequence lengths without any fine-tuning. We show that StreamingLLM can enable Llama-2, MPT, Falcon, and Pythia to perform stable and efficient language modeling with up to 4 million tokens and more. In addition, we discover that adding a placeholder token as a dedicated attention sink during pre-training can further improve streaming deployment. In streaming settings, StreamingLLM outperforms the sliding window recomputation baseline by up to 22.2x speedup. Code and datasets are provided at https://github.com/mit-han-lab/streaming-llm.

Related papers

• LServe: Efficient Long-sequence LLM Serving with Unified Sparse Attention [26.54297116028556]
  LServe is an efficient system that accelerates long-sequence language models. It unifies different hardware-friendly, structured sparsity patterns for both prefilling and decoding attention. On average, LServe accelerates LLM prefilling by up to 2.9x and decoding by 1.3-2.1x over vLLM.
  arXiv  Detail & Related papers  (2025-02-20T18:59:52Z)
• InfiniteHiP: Extending Language Model Context Up to 3 Million Tokens on a Single GPU [48.105361428245736]
  We introduce InfiniteHiP, an inference framework for large language models (LLMs) We dynamically eliminate irrelevant context tokens through a modular hierarchical token pruning algorithm. Our framework achieves an 18.95x speedup in attention decoding for a 1 million token context without requiring additional training.
  arXiv  Detail & Related papers  (2025-02-13T02:52:01Z)
• AttentionPredictor: Temporal Pattern Matters for Efficient LLM Inference [51.1972443343829]
  We propose AttentionPredictor, which is the first learning-based critical token identification approach. AttentionPredictor accurately predicts the attention score while consuming negligible memory. We also propose a cross-token critical cache prefetching framework that hides the token time overhead to accelerate the decoding stage.
  arXiv  Detail & Related papers  (2025-02-06T13:41:46Z)
• Compressing KV Cache for Long-Context LLM Inference with Inter-Layer Attention Similarity [24.118503938098307]
  Existing methods, including selective token retention and window-based attention, improve efficiency but risk discarding important tokens needed for future text generation. We propose an approach that enhances LLM efficiency without token loss by reducing the memory and computational load of less important tokens, rather than discarding them.
  arXiv  Detail & Related papers  (2024-12-03T08:29:27Z)
• Recycled Attention: Efficient inference for long-context language models [54.00118604124301]
  We propose Recycled Attention, an inference-time method which alternates between full context attention and attention over a subset of input tokens. When performing partial attention, we recycle the attention pattern of a previous token that has performed full attention and attend only to the top K most attended tokens. Compared to previously proposed inference-time acceleration method which attends only to local context or tokens with high accumulative attention scores, our approach flexibly chooses tokens that are relevant to the current decoding step.
  arXiv  Detail & Related papers  (2024-11-08T18:57:07Z)
• Training-Free Exponential Context Extension via Cascading KV Cache [49.608367376911694]
  We introduce a novel mechanism that leverages cascading sub-cache buffers to selectively retain the most relevant tokens. Our method reduces prefill stage latency by a factor of 6.8 when compared to flash attention on 1M tokens.
  arXiv  Detail & Related papers  (2024-06-24T03:59:17Z)
• A Training-free Sub-quadratic Cost Transformer Model Serving Framework With Hierarchically Pruned Attention [43.211427581302715]
  We propose Hierarchically Pruned Attention (HiP) to increase context length in large language models. HiP reduces the time complexity of the attention mechanism to $O(T log T)$ and the space complexity to $O(T)$, where $T$ is the sequence length. We show that HiP significantly reduces both prefill and decoding latencies, as well as memory usage, while maintaining high-quality generation with minimal degradation.
  arXiv  Detail & Related papers  (2024-06-14T08:32:45Z)
• An Image is Worth 1/2 Tokens After Layer 2: Plug-and-Play Inference Acceleration for Large Vision-Language Models [65.37846460916042]
  We find out that the attention computation over visual tokens is of extreme inefficiency in the deep layers of popular LVLMs. We introduce FastV, a versatile plug-and-play method designed to optimize computational efficiency.
  arXiv  Detail & Related papers  (2024-03-11T14:35:32Z)
• LM-Infinite: Zero-Shot Extreme Length Generalization for Large Language Models [83.98062659664785]
  Large language models (LLMs) typically train on short text segments (e.g., 4K tokens) due to the quadratic complexity of their Transformer architectures. This work identifies three major factors contributing to this length generalization failure. We propose LM-Infinite, a simple and effective method for enhancing LLMs' capabilities of handling long contexts.
  arXiv  Detail & Related papers  (2023-08-30T16:47:51Z)

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.