8 papers
AlignPred: Error-Controlled KV Cache Compression via Alignment and Predictive Coding
Xu Yang, Jiapeng Zhang, Zhiwei Zuo +5
Long-running agents retain KV caches across turns and sessions for hours or days. Compressing these persistent caches requires reducing storage while controlling reconstruction err…
Self-Indexing Attention for Compression-Compatible Sparse Long-Context LLM Inference
Xu Yang, Jiapeng Zhang, Zhangke +6
Sparse long-context inference requires efficient token retrieval in both prefill and decode. Existing methods often use different retrieval strategies for the two stages, preventin…
Rethinking Local Learning: A Cheaper and Faster Recipe for LLM Post-Training
Hengyu Shi, Tianyang Han, Peizhe Wang +3
LLM post-training typically propagates task gradients through the full depth of the model. Although this end-to-end structure is simple and general, it couples task adaptation to f…
Chain-based Distillation for Effective Initialization of Variable-Sized Small Language Models
Boyu Shi, YiCheng Jiang, Chang Liu +3
Large language models (LLMs) achieve strong performance but remain costly to deploy in resource-constrained settings. Training small language models (SLMs) from scratch is computat…
Hierarchical Dual-Subspace Decoupling for Continual Learning in Vision-Language Models
Mengxin Qin, Xiang Zhang, Kun Wei +2
Class-incremental learning aims to continuously acquire new knowledge while preserving previously learned information, thereby mitigating catastrophic forgetting. Existing methods…
DIMoE-Adapters: Dynamic Expert Evolution for Continual Learning in Vision-Language Models
Mengxin Qin, Xiang Zhang, Xi Wang +3
Continual learning enables vision-language models to accumulate knowledge and adapt to evolving tasks without retraining from scratch. However, in multi-domain task-incremental lea…
Understanding Performance Collapse in Layer-Pruned Large Language Models via Decision Representation Transitions
Boyu Shi, Chang Liu, ChuanBao Gao +2
Layer pruning efficiently reduces Large Language Model (LLM) computational costs but often triggers sudden performance collapse. Existing representation-based analyses struggle to…
Self-Indexing KVCache: Predicting Sparse Attention from Compressed Keys
Xu Yang, Jiapeng Zhang, Dongyang Zhao +2
The KV cache in self-attention has emerged as a major bottleneck in long-context and large-batch inference for LLMs. Existing approaches often treat sparsity prediction and compres…