Large language models (LLMs) utilize key-value (KV) cache to store historical information during sequence processing. The size of KV cache grows linearly as the length of the sequence extends, which seriously affects memory usage and decoding efficiency. Current methods for KV cache eviction typically utilize the last window from the pre-filling phase as queries to compute the KV importance scores for eviction. Although this scheme is simple to implement, it tends to overly focus on local information, potentially leading to the neglect or omission of crucial global information. To mitigate this issue, we propose **Judge Q**, a novel training method which incorporates a soft token list. This method only tunes the model’s embedding layer at a low training cost. By concatenating the soft token list at the end of the input sequence, we train these tokens' attention map to the original input sequence to align with that of the actual decoded tokens. In this way, the queries corresponding to the soft tokens can effectively capture global information and better evaluate the importance of the keys and values within the KV cache, thus maintaining decoding quality when KV cache is evicted. Under the same eviction budget, our method exhibits less performance degradation compared to existing eviction approaches. We validate our approach through experiments conducted on models such as Llama-3.1-8B-Instruct and Mistral-7B-Instruct-v0.3, using benchmarks including LongBench, RULER, and Needle-in-a-Haystack. Results indicate an improvement of approximately 1 point on the LongBench and over 3 points on RULER. This proposed methodology can be seamlessly integrated into existing open-source models with minimal training overhead, thereby enhancing performance in KV cache eviction scenarios.

Large Language Models (LLMs) with Mixture-of-Experts (MoE) architectures are distinguished by their strong performance scaling with increasing parameters across a wide range of tasks, yet they also suffer from substantial computational and storage overheads. Notably, the performance gains of MoE models do not scale proportionally with the growth in expert parameters. While prior works attempt to reduce parameters via expert-level pruning, merging, or decomposition, they still suffer from challenges in both performance and computational efficiency. In this paper, we address these challenges by introducing micro-expert as a finer-grained compression unit that spans across matrices. We first establish a more fundamental perspective, viewing MoE layers as mixtures of micro-experts, and present CAMERA, a lightweight and training-free framework for identifying micro-expert redundancy. Our analysis uncovers significant variance in micro-expert contributions during decoding. Based on this insight, we further propose CAMERA-P, a structured micro-expert pruning framework, and CAMERA-Q, a mixed-precision quantization idea designed for micro-experts. Extensive experiments on nine downstream tasks show that CAMERA-P consistently outperforms strong baselines under pruning ratios ranging from 20% to 60%. Furthermore, CAMERA-Q achieves superior results under aggressive 2-bit quantization, surpassing existing matrix- and channel-level ideas. Notably, our method enables complete micro-expert analysis of Qwen2-57B-A14B in less than 5 minutes on a single NVIDIA A100-40GB GPU.

Ziyue Wang, Chi Chen, Fuwen Luo, Yurui Dong, Yuanchi Zhang, Yuzhuang Xu, Xiaolong Wang, Peng Li, Yang Liu. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2025.

Large language models (LLMs) rely on keyvalue cache (KV cache) to accelerate decoding by reducing redundant computations.However, the KV cache memory usage grows substantially with longer text sequences, posing challenges for efficient deployment.Existing KV cache eviction methods prune tokens using prefilling-stage attention scores, causing inconsistency with actual inference queries, especially under tight memory budgets.In this paper, we propose Lookahead Q-Cache (LAQ), a novel eviction framework that generates lowcost pseudo lookahead queries to better approximate the true decoding-stage queries.By using these lookahead queries as the observation window for importance estimation, LAQ achieves more consistent and accurate KV cache eviction aligned with real inference scenarios.Experimental results on LongBench and Needlein-a-Haystack benchmarks show that LAQ outperforms existing methods across various budget levels, achieving a 1 ∼ 4 point improvement on LongBench under limited cache budget.Moreover, LAQ is complementary to existing approaches and can be flexibly combined to yield further improvements.

Abstract Powerful large language models (LLMs) are increasingly expected to be deployed with lower computational costs, enabling their capabilities on resource-constrained devices. Post-training quantization (PTQ) has emerged as a star approach to achieve this ambition, with best methods compressing weights to less than 2 bit on average. In this paper, we propose Channel-Relaxed Vector Quantization (CRVQ), a novel technique that significantly improves the performance of PTQ baselines at the cost of only minimal additional bits. This state-of-the-art extreme compression method achieves its results through two key innovations: (1) carefully selecting and reordering a very small subset of critical weight channels, and (2) leveraging extended codebooks to relax the constraint of critical channels. With our method, we demonstrate a 38.9% improvement over the current strongest sub-2-bit PTQ baseline, enabling nearer lossless 1-bit compression. Furthermore, our approach offers flexible customization of quantization bit-width and performance, providing a wider range of deployment options for diverse hardware platforms. Code and checkpoints are available at https://github.com/xuyuzhuang11/CRVQ.

Haoyu Wang, Shuo Wang, Yukun Yan, Xujia Wang, Zhiyu Yang, Yuzhuang Xu, Zhenghao Liu, Liner Yang, Ning Ding, Xu Han, Zhiyuan Liu, Maosong Sun. Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2024.

Although neural machine translation (NMT) models perform well in the general domain, it remains rather challenging to control their generation behavior to satisfy the requirement of different users. Given the expensive training cost and the data scarcity challenge of learning a new model from scratch for each user requirement, we propose a memory-augmented adapter to steer pretrained NMT models in a pluggable manner. Specifically, we construct a multi-granular memory based on the user-provided text samples and propose a new adapter architecture to combine the model representations and the retrieved results. We also propose a training strategy using memory dropout to reduce spurious dependencies between the NMT model and the memory. We validate our approach on both style- and domain-specific experiments and the results indicate that our method can outperform several representative pluggable baselines.

Fine-tuning is a crucial process for adapting large language models (LLMs) to diverse applications. In certain scenarios, such as multi-tenant serving, deploying multiple LLMs becomes necessary to meet complex demands. Recent studies suggest decomposing a fine-tuned LLM into a base model and corresponding delta weights, which are then compressed using low-rank or low-bit approaches to reduce costs. In this work, we observe that existing low-rank and low-bit compression methods can significantly harm the model performance for task-specific fine-tuned LLMs (e.g., WizardMath for math problems). Motivated by the long-tail distribution of singular values in the delta weights, we propose a delta quantization approach using mixed-precision. This method employs higher-bit representation for singular vectors corresponding to larger singular values. We evaluate our approach on various fine-tuned LLMs, including math LLMs, code LLMs, chat LLMs, and even VLMs. Experimental results demonstrate that our approach performs comparably to full fine-tuned LLMs, surpassing both low-rank and low-bit baselines by a considerable margin. Additionally, we show that our method is compatible with various backbone LLMs, such as Llama-2, Llama-3, and Mistral, highlighting its generalizability.

Model quantification uses low bit-width values to represent the weight matrices of existing models to be quantized, which is a promising approach to reduce both storage and computational overheads of deploying highly anticipated LLMs. However, current quantization methods suffer severe performance degradation when the bit-width is extremely reduced, and thus focus on utilizing 4-bit or 8-bit values to quantize models. This paper boldly quantizes the weight matrices of LLMs to 1-bit, paving the way for the extremely low bit-width deployment of LLMs. For this target, we introduce a 1-bit model compressing framework named OneBit, including a novel 1-bit parameter representation method to better quantize LLMs as well as an effective parameter initialization method based on matrix decomposition to improve the convergence speed of the quantization framework. Sufficient experimental results indicate that OneBit achieves good performance (at least 81% of the non-quantized performance on LLaMA models) with robust training processes when only using 1-bit weight matrices.

The cover image is based on the Original Article Verification Algebra for Multi-Tenant Applications in VaaS Architecture by Kan Luo et al., https://doi.org/10.1002/stvr.1763.

Summary This paper proposes an algebraic system, verification algebra (VA), for reducing the number of component combinations to be verified in multi‐tenant architecture (MTA). MTA is a design architecture used in SaaS (Software‐as‐a‐Service) where a tenant can customize its applications by integrating services already stored in the SaaS databases or newly supplied services. Similar to SaaS, VaaS (Verification‐as‐a‐Service) is a verification service in a cloud that leverages the computing power offered by a cloud environment with automated provisioning, scalability and service composition. In VaaS architecture, however, there is a challenging problem called ‘combinatorial explosion’ that it is difficult to verify a large number of compositions constructed by both quantities of components and various combination structures even with computing resources in cloud. This paper proposes rules to emerge combinations status for future verification, on the basis of the existing results. Both composition patterns and properties are considered and analysed in VA rules.

The published online cover has been updated.