Challenge: Prior work typically decomposes inference into prefill and decode stages, with the decode stage dominating total latency.
Approach: They propose an algorithm that detects threshold where information loss exceeds information gain during sparse decoding to reduce token consumption by up to 90% and a marginal accuracy degradation of less than 2%.
Outcome: The proposed algorithm reduces token consumption by 90% with a marginal accuracy degradation of less than 2% across reasoning-intensive benchmarks.

Similar Papers

Sparse-to-Dense: A Free Lunch for Lossless Acceleration of Video Understanding in LLMs (2025.acl-short)

Copied to clipboard

Challenge: Recent advances in Video Large Language Models (Video-LLMs) have achieved exceptional performance on tasks like video question answering and captioning.
Approach: They propose a decoding strategy that leverages sparse top-K attention and dense full attention to accelerate Video-LLMs without loss.
Outcome: The proposed approach achieves a 1.94 walltime speedup in video processing.
Evolving Sparsity: Leveraging Token Importance Dynamics for Efficient LLM Decoding with Sparse Attention (2026.acl-long)

Copied to clipboard

Challenge: Efficient long-context inference remains a major challenge for large language models (LLMs), as the cost of attention computation during auto-regressive decoding grows linearly with the context length.
Approach: They propose to model token importance as a dynamic process that evolves over decoding steps and propagates through model layers.
Outcome: The proposed method outperforms baseline sparse attention methods and achieves speedups of up to 5.36 for attention latency and 2.33 for end-to-end decoding.
Less is More: Improving LLM Reasoning with Minimal Test-Time Intervention (2026.acl-long)

Copied to clipboard

Challenge: Recent advances in large language models (LLMs) have focused on test-time scaling to improve reasoning quality but at the cost of efficiency.
Approach: They propose a training-free framework that enhances reasoning accuracy and stability with minimal overhead.
Outcome: The proposed framework yields consistent gains across general, coding, and STEM tasks while remaining highly efficient.
RaaS: Reasoning-Aware Attention Sparsity for Efficient LLM Reasoning (2025.findings-acl)

Copied to clipboard

Challenge: Large Language Models (LLMs) have demonstrated strong capabilities across various domains, but their large-scale deployment faces a major obstacle: the high computational cost of long-sequence inference.
Approach: They propose an algorithm that retains key-value vectors until they are no longer needed to solve reasoning tasks.
Outcome: The proposed algorithm achieves high accuracy with O(L) time but O(N) memory complexities.
SLiM: Speculative Decoding with Hypothesis Reduction (2024.findings-naacl)

Copied to clipboard

Challenge: Speculative decoding has emerged as an alternative to autoregressive decoding for expediting inference in large language models (LLMs). prevailing assumptions focus solely on latency reduction, neglecting the computational expenses.
Approach: They propose a speculative decoding enhancement to reduce the speculation set while validating more effective tokens.
Outcome: The proposed method reduces the speculation set while validating more effective tokens.
CoLA: Compute-Efficient Pre-Training of LLMs via Low-Rank Activation (2025.emnlp-main)

Copied to clipboard

Challenge: Large foundation models have become huge, but they consume computational resources in pretraining.
Approach: They propose to replace full-size layers with compute-efficient auto-encoders that enforce low-rank activations throughout training.
Outcome: The proposed method reduces the computing cost by 2pmbtimes and improves training throughput by 1.86pmtime.
The Sparse Frontier: Sparse Attention Trade-offs in Transformer LLMs (2026.findings-acl)

Copied to clipboard

Challenge: Sparse attention is a promising strategy to extend long-context capabilities in LLMs . but its efficiency–accuracy trade-offs remain unclear due to the lack of comprehensive evaluation .
Approach: They evaluate sparse attention methods across multiple model families and sizes . they find larger sparser models outperform smaller dense ones at equivalent cost .
Outcome: The proposed methods outperform smaller sparse models at equivalent cost and improve the Pareto frontier.
Less is More: A Simple yet Effective Token Reduction Method for Efficient Multi-modal LLMs (2025.coling-main)

Copied to clipboard

Challenge: Recent advances in Multimodal Large Language Models have led to a significant surge in the resource consumption of these models.
Approach: They propose a method to reduce image tokens using visual query data by using CLIP metrics to reduce computational overhead and maintain consistent performance.
Outcome: The proposed method has been extensively tested across 12 datasets and shows a significant reduction in computational overhead while maintaining a consistent level of performance.
Efficient Sparse Attention needs Adaptive Token Release (2024.findings-acl)

Copied to clipboard

Challenge: Large Language Models (LLMs) have demonstrated remarkable capabilities across a wide array of text-centric tasks, however, their ‘large’ scale introduces significant computational and storage challenges, particularly in managing the key-value states of the transformer, which limits their wider applicability.
Approach: They propose to release resources from caches and rebuild key-value states by a lightweight controller module to approximate an ideal top-K sparse attention.
Outcome: The proposed method achieves a significant throughput improvement of 221.8% over full attention and a model with 7 billion tokens.
1+1>2: A Synergistic Sparse and Low-Rank Compression Method for Large Language Models (2025.findings-emnlp)

Copied to clipboard

Challenge: Low-rank approximation compresses the model by retaining its essential structure with minimal information loss.
Approach: They propose a method that leverages the strengths of pruning and low-rank approximation for LLMs.
Outcome: The proposed methods surpass the existing methods on LLaMA and Qwen2.5 models.

What is GenGO?

GenGO is an NLP powered publication search system. It currenctly indexes 30k+ papers from ACL Anthology, and implements multi-aspect summarization, semantic search, and more!

Information

About
Limitations