6 papers · 1 filter
Depth-Attention: Cross-Layer Value Mixing for Language Models
Boyi Zeng, Yiqin Hao, Zitong Wang +7
Self-attention selects information freely across the sequence, but across depth, Transformers merely add each layer's output to the residual stream, so later layers cannot selectiv…
AdaPonderLM: Gated Pondering Language Models with Token-Wise Adaptive Depth
Shixiang Song, He Li, Zitong Wang +6
Test-time scaling via recurrent/iterative Transformers enables large language models to spend more computation at inference, but most pretrained recurrent LMs run a fixed number of…
PonderLM-3: Adaptive Token-Wise Pondering with Differentiable Masking
He Li, Feichen Song, Boyi Zeng +4
Test-time scaling has shown that allocating more additional computation at inference can improve generation quality, motivating a natural follow-up question: where should this comp…
FreqKV: Key-Value Compression in Frequency Domain for Context Window Extension
Jushi Kai, Yixuan Wang, Boyi Zeng +4
Existing key-value (KV) cache compression methods for large language models (LLMs) often rely on token eviction, which risks losing critical local information in both long prefilli…
TreeKV: Smooth Key-Value Cache Compression with Tree Structures
Ziwei He, Jian Yuan, Haoli Bai +2
Efficient key-value (KV) cache compression is critical for scaling transformer-based Large Language Models (LLMs) in long sequences and resource-limited settings. Existing methods…
WeightedKV: Attention Scores Weighted Key-Value Cache Merging for Large Language Models
Jian Yuan, Ziwei He, Haoli Bai +2
Large Language Models (LLMs) use key-value (KV) cache to reduce redundant computation in autoregressive generation. However, the KV cache size increases linearly during generation,…