Compress, Gather, and Recompute: REFORMing Long-Context Processing in Transformers

• URL: http://arxiv.org/abs/2506.01215v1
• Date: Sun, 01 Jun 2025 23:49:14 GMT
• Title: Compress, Gather, and Recompute: REFORMing Long-Context Processing in Transformers
• Authors: Woomin Song, Sai Muralidhar Jayanthi, Srikanth Ronanki, Kanthashree Mysore Sathyendra, Jinwoo Shin, Aram Galstyan, Shubham Katiyar, Sravan Babu Bodapati,
• Abstract summary: REFORM is a novel inference framework that efficiently handles long contexts through a two-phase approach.<n>It achieves over 50% and 27% performance gains on RULER and BABILong respectively at 1M context length.<n>It also outperforms baselines on Infinite-Bench and MM-NIAH, demonstrating flexibility across diverse tasks and domains.
• Score: 58.98923344096319
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: As large language models increasingly gain popularity in real-world applications, processing extremely long contexts, often exceeding the model's pre-trained context limits, has emerged as a critical challenge. While existing approaches to efficient long-context processing show promise, recurrent compression-based methods struggle with information preservation, whereas random access approaches require substantial memory resources. We introduce REFORM, a novel inference framework that efficiently handles long contexts through a two-phase approach. First, it incrementally processes input chunks while maintaining a compressed KV cache, constructs cross-layer context embeddings, and utilizes early exit strategy for improved efficiency. Second, it identifies and gathers essential tokens via similarity matching and selectively recomputes the KV cache. Compared to baselines, REFORM achieves over 50% and 27% performance gains on RULER and BABILong respectively at 1M context length. It also outperforms baselines on Infinite-Bench and MM-NIAH, demonstrating flexibility across diverse tasks and domains. Additionally, REFORM reduces inference time by 30% and peak memory usage by 5%, achieving both efficiency and superior performance.

Related papers

• Scaling Linear Attention with Sparse State Expansion [58.161410995744596]
  Transformer architecture struggles with long-context scenarios due to quadratic computation and linear memory growth.<n>We introduce a row-sparse update formulation for linear attention by conceptualizing state updating as information classification.<n>Second, we present Sparse State Expansion (SSE) within the sparse framework, which expands the contextual state into multiple partitions.
  arXiv  Detail & Related papers  (2025-07-22T13:27:31Z)
• MEM1: Learning to Synergize Memory and Reasoning for Efficient Long-Horizon Agents [84.62985963113245]
  We introduce MEM1, an end-to-end reinforcement learning framework that enables agents to operate with constant memory across long multi-turn tasks.<n>At each turn, MEM1 updates a compact shared internal state that jointly supports memory consolidation and reasoning.<n>We show that MEM1-7B improves performance by 3.5x while reducing memory usage by 3.7x compared to Qwen2.5-14B-Instruct on a 16-objective multi-hop QA task.
  arXiv  Detail & Related papers  (2025-06-18T19:44:46Z)
• Overflow Prevention Enhances Long-Context Recurrent LLMs [41.8230324612529]
  A recent trend in LLMs is developing recurrent sub-quadratic models that improve long-context processing efficiency.<n>We investigate leading large long-context models, focusing on how their fixed-size recurrent memory affects their performance.<n>Our experiments reveal that, even when these models are trained for extended contexts, their use of long contexts remains underutilized.
  arXiv  Detail & Related papers  (2025-05-12T17:45:05Z)
• ZSMerge: Zero-Shot KV Cache Compression for Memory-Efficient Long-Context LLMs [7.958429361868486]
  We propose ZSMerge, a dynamic KV cache compression framework for efficient cache management.<n>ZSMerge significantly enhances memory efficiency and inference speed with negligible performance degradation.
  arXiv  Detail & Related papers  (2025-03-13T03:36:03Z)
• Does RAG Really Perform Bad For Long-Context Processing? [15.889864680212147]
  RetroLM is a novel framework for long-context processing.<n>Unlike traditional methods, RetroLM employs KV-level retrieval augmentation.<n>Building on this framework, we develop a specialized retriever for precise retrieval of critical pages.
  arXiv  Detail & Related papers  (2025-02-17T05:02:25Z)
• SHERL: Synthesizing High Accuracy and Efficient Memory for Resource-Limited Transfer Learning [63.93193829913252]
  We propose an innovative METL strategy called SHERL for resource-limited scenarios. In the early route, intermediate outputs are consolidated via an anti-redundancy operation. In the late route, utilizing minimal late pre-trained layers could alleviate the peak demand on memory overhead.
  arXiv  Detail & Related papers  (2024-07-10T10:22:35Z)
• UIO-LLMs: Unbiased Incremental Optimization for Long-Context LLMs [111.12010207132204]
  UIO-LLMs is an incremental optimization approach for memory-enhanced transformers under long-context settings. We refine the training process using the Truncated Backpropagation Through Time (TBPTT) algorithm. UIO-LLMs successfully handle long context, such as extending the context window of Llama2-7b-chat from 4K to 100K tokens with minimal 2% additional parameters.
  arXiv  Detail & Related papers  (2024-06-26T08:44:36Z)
• 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.<n>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)
• Scaling Transformer to 1M tokens and beyond with RMT [5.60052250541419]
  A major limitation for the broader scope of problems solvable by transformers is the quadratic scaling of computational complexity with input size. In this study, we investigate the recurrent memory augmentation of pre-trained transformer models to extend input context length while linearly scaling compute. Our approach demonstrates the capability to store information in memory for sequences of up to an unprecedented two million tokens while maintaining high retrieval accuracy.
  arXiv  Detail & Related papers  (2023-04-19T16:18:54Z)

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.