Context Compression for Auto-regressive Transformers with Sentinel Tokens

• URL: http://arxiv.org/abs/2310.08152v2
• Date: Sun, 15 Oct 2023 09:15:02 GMT
• Title: Context Compression for Auto-regressive Transformers with Sentinel Tokens
• Authors: Siyu Ren, Qi Jia, Kenny Q. Zhu
• Abstract summary: We propose a plug-and-play approach that is able to incrementally compress the intermediate activation of a specified span of tokens into compact ones. Experiments on both in-domain language modeling and zero-shot open-ended document generation demonstrate the advantage of our approach.
• Score: 37.07722536907739
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: The quadratic complexity of the attention module makes it gradually become the bulk of compute in Transformer-based LLMs during generation. Moreover, the excessive key-value cache that arises when dealing with long inputs also brings severe issues on memory footprint and inference latency. In this work, we propose a plug-and-play approach that is able to incrementally compress the intermediate activation of a specified span of tokens into compact ones, thereby reducing both memory and computational cost when processing subsequent context. Experiments on both in-domain language modeling and zero-shot open-ended document generation demonstrate the advantage of our approach over sparse attention baselines in terms of fluency, n-gram matching, and semantic similarity. At last, we comprehensively profile the benefit of context compression on improving the system throughout. Code is available at https://github.com/DRSY/KV_Compression.

Related papers

• Cognitive Chunking for Soft Prompts: Accelerating Compressor Learning via Block-wise Causal Masking [28.492055407384495]
  Long contexts increase inference latency, as the computational cost of self-attention grows quadratically with sequence length.<n>Existing methods typically compress the entire context indiscriminately into a set of memory tokens.<n>We propose Parallelized Iterative Compression (PIC), which restricts the receptive field of memory tokens to sequential local chunks.
  arXiv  Detail & Related papers  (2026-02-15T03:58:13Z)
• Context Compression via Explicit Information Transmission [25.078241611630585]
  Long-context inference with Large Language Models (LLMs) is costly due to quadratic attention and growing key-value caches.<n>We propose ComprExIT, a lightweight framework that formulates soft compression into a new paradigm.
  arXiv  Detail & Related papers  (2026-02-03T17:44:12Z)
• Gradual Forgetting: Logarithmic Compression for Extending Transformer Context Windows [3.437656066916039]
  We introduce an approach that modifies the input representation itself, rather than the transformer architecture.<n>Inspired by cognitive models of human memory, our method applies a scale-invariant logarithmic compression to the input tokens.
  arXiv  Detail & Related papers  (2025-10-25T01:29:37Z)
• UniGist: Towards General and Hardware-aligned Sequence-level Long Context Compression [86.33995240043936]
  UniGist is a sequence-level long-context compression framework for large language models.<n>It efficiently preserves context information by replacing raw tokens with special compression tokens (gists) in a fine-grained manner.<n>Our scheme also supports flexible inference by allowing the actual removal of compressed tokens, resulting in real-time memory savings.
  arXiv  Detail & Related papers  (2025-09-19T08:47:37Z)
• SCOUT: Toward Sub-Quadratic Attention via Segment Compression for Optimized Utility in Transformers [15.142822497807236]
  We propose SCOUT, a hybrid architecture that compresses tokens locally within fixed-size segments and applies attention only over these compressed representations.<n>SCOUT retains much of the expressivity of full attention while substantially reducing the computational and memory cost.<n>We analyze SCOUT's computational and memory efficiency and evaluate it empirically on long-context language modeling and reasoning tasks.
  arXiv  Detail & Related papers  (2025-08-31T17:08:33Z)
• Lag-Relative Sparse Attention In Long Context Training [8.365610885641276]
  We propose Lag-Relative Sparse Attention(LRSA) anchored by the LagKV compression method for long context post-training.<n>Our method performs chunk-by-chunk prefilling, which selects the top K most relevant key-value pairs in a fixed-size lagging window.
  arXiv  Detail & Related papers  (2025-06-13T06:49:53Z)
• zip2zip: Inference-Time Adaptive Tokenization via Online Compression [27.16551923444618]
  zip2zip is a novel method for achieving context-adaptive tokenization in large language models.<n>We show that an existing LLM can be uptrained for zip2zip in 10 GPU-hours via parameter-efficient finetuning.<n>The resulting LLM performs test-time adaptation, learning to use hypertokens in unseen contexts and reducing input and output tokens by 15-40%.
  arXiv  Detail & Related papers  (2025-06-01T17:03:02Z)
• Efficient Token Compression for Vision Transformer with Spatial Information Preserved [59.79302182800274]
  Token compression is essential for reducing the computational and memory requirements of transformer models. We propose an efficient and hardware-compatible token compression method called Prune and Merge.
  arXiv  Detail & Related papers  (2025-03-30T14:23:18Z)
• ChunkKV: Semantic-Preserving KV Cache Compression for Efficient Long-Context LLM Inference [61.412894960600205]
  Large Language Models (LLMs) require significant GPU memory when processing long texts.<n>ChunkKV reimagines KV cache compression by treating semantic chunks as basic compression units.<n>Result: ChunkKV outperforms state-of-the-art methods by up to 8.7% in precision.
  arXiv  Detail & Related papers  (2025-02-01T03:49:47Z)
• RefreshKV: Updating Small KV Cache During Long-form Generation [54.00118604124301]
  We propose a new inference method, RefreshKV, that flexibly alternates between full context attention and attention over a subset of input tokens during generation. Applying our method to off-the-shelf LLMs achieves comparable speedup to eviction-based methods while improving performance for various long-form generation tasks.
  arXiv  Detail & Related papers  (2024-11-08T18:57:07Z)
• LoRC: Low-Rank Compression for LLMs KV Cache with a Progressive Compression Strategy [59.1298692559785]
  Key-Value ( KV) cache is crucial component in serving transformer-based autoregressive large language models (LLMs) Existing approaches to mitigate this issue include: (1) efficient attention variants integrated in upcycling stages; (2) KV cache compression at test time; and (3) KV cache compression at test time. We propose a low-rank approximation of KV weight matrices, allowing plug-in integration with existing transformer-based LLMs without model retraining. Our method is designed to function without model tuning in upcycling stages or task-specific profiling in test stages.
  arXiv  Detail & Related papers  (2024-10-04T03:10:53Z)
• 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)
• In-Context Former: Lightning-fast Compressing Context for Large Language Model [48.831304302467004]
  In this paper, we propose a new approach to compress the long input contexts of Transformer-based large language models (LLMs) We use the cross-attention mechanism and a small number of learnable digest tokens to condense information from the contextual word embeddings. Experimental results indicate that our method requires only 1/32 of the floating-point operations of the baseline during compression and improves processing speed by 68 to 112 times.
  arXiv  Detail & Related papers  (2024-06-19T15:14:55Z)
• SubGen: Token Generation in Sublinear Time and Memory [48.35076900702408]
  Large language models (LLMs) have extensive memory requirements for token generation. In this work, we focus on developing an efficient compression technique for the KV cache. We have devised a novel caching method with sublinear complexity, employing online clustering on key tokens and online $ell$ sampling on values. Not only does this algorithm ensure a sublinear memory footprint and sublinear time complexity, but we also establish a tight error bound for our approach.
  arXiv  Detail & Related papers  (2024-02-08T22:17:40Z)
• Long Context Compression with Activation Beacon [22.054232261437186]
  Activation Beacon is a plug-in module for transformer-based LLMs. It targets effective, efficient, and flexible compression of long contexts. It achieves a 2x acceleration in inference time and an 8x reduction of memory costs for KV cache.
  arXiv  Detail & Related papers  (2024-01-07T11:57:40Z)
• Compressed Context Memory For Online Language Model Interaction [39.72054168889216]
  This paper presents a context key/value compression method for Transformer language models in online scenarios. As the context lengthens, the attention process demands increasing memory and computations, which in turn reduces the throughput of the language model. We propose a compressed context memory system that continually compresses the accumulating attention key/value pairs into a compact memory space.
  arXiv  Detail & Related papers  (2023-12-06T10:50:43Z)
• Cluster-Former: Clustering-based Sparse Transformer for Long-Range Dependency Encoding [90.77031668988661]
  Cluster-Former is a novel clustering-based sparse Transformer to perform attention across chunked sequences. The proposed framework is pivoted on two unique types of Transformer layer: Sliding-Window Layer and Cluster-Former Layer. Experiments show that Cluster-Former achieves state-of-the-art performance on several major QA benchmarks.
  arXiv  Detail & Related papers  (2020-09-13T22:09:30Z)

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.