arXiv AI
Sigmoid Attention as a Better Substrate for Learned KV Cache Eviction
Read the original on arXiv AI →The Flow has not summarised this story yet — read it at arXiv AI.
Related stories
Hugging Face Trending Papers
Aug 24
Sigmoid Attention as a Better Substrate for Learned KV Cache Eviction
Learned KV-cache eviction often faces a soft-to-hard mismatch: during training, differentiable gates typically attenuate token contributions, whereas inference saves memory only when KV entries are ph...
arXiv Machine Learning
2d ago
Routing Absorption in Sparse Attention: Why Random Gates Are Hard to Beat
The paper examines why learned gates in sparse attention models offer little advantage over random gates when jointly trained with the transformer. Through experiments on a 31M-parameter transformer, the authors attribute this to routing absorption, where the model’s representations adapt to the imposed mask, diminishing the benefit of learned routing. They also explore hard masking, stochastic mask training, and the impact of trainable attention layers on gate performance, concluding that freezing the model stabilizes routing targets for effective post‑hoc sparsification.
arXiv AI
Jun 9
FlashMemory-DeepSeek-V4: Lightning Index Ultra-Long Context via Lookahead Sparse Attention
arXiv:2606. 09079v1 Announce Type: cross Abstract: Conventional LLMs keep the full KV cache loaded during decoding, causing a severe GPU memory bottleneck for ultra-long context serving.
arXiv AI
Jun 17
AnchorKV: Safety-Aware KV Cache Compression via Soft Penalty with a Refusal Anchor
arXiv:2606. 17872v1 Announce Type: cross Abstract: Large language models (LLMs) outperform earlier architectures on generative inference and long-context tasks, but their large size introduces significant challenges in memory usage, energy cost, and on-device deployment.
arXiv Computation and Language
Sep 1
Attention Amnesia in Hybrid LLMs: When CoT Fine-Tuning Breaks Long-Range Recall, and How to Fix It
The paper reports that chain‑of‑thought (CoT) supervised fine‑tuning (SFT) improves reasoning but systematically harms long‑context recall in hybrid linear‑attention models such as HypeNet and Jet‑Nemotron. Retrieval performance on the Needle‑In‑A‑Haystack benchmark drops dramatically after CoT‑SFT, especially with harder settings and longer contexts. The authors introduce QK‑Restore, a training‑free method that reinstates the query‑key projection matrices from the pre‑SFT checkpoint, which recovers long‑range recall while preserving reasoning gains.
arXiv AI
Sep 3
Language Models Can Control Their Own Attention
The paper introduces Declarative Attention (DA), a protocol that lets language models explicitly declare which parts of their context to focus on during generation. By partitioning decoding into full-context, region-specific, and recent-output-only modes, the inference engine can skip large portions of the KV cache, dramatically reducing attended tokens. Experiments on 15 long-context tasks with off-the-shelf models show significant savings (52.0% and 31.1% reductions) with only modest accuracy drops that diminish as model size increases.