#DeepSeek-R1-Distill-Llama-8B

3 stories · 0 in 7 days · 1 in 30 days

KV缓存 2 Gemini 1.5 Flash 1 GPT-4o Mini 1 Qwen3-8B 1 SVD 1 vLLM 1 多租户 1 推理优化 1

Stories, products and papers tagged "DeepSeek-R1-Distill-Llama-8B", newest first.

Oct 5

Models Builders19:13

Chinese contentiS-KV：基于块增量 SVD 的在线 KV 缓存压缩方法

长链推理时 KV 缓存会爆炸，这篇提出用增量 SVD 压到 4-5 倍还能保住准确率，比删 token 的老办法稳，做推理部署的可以看看。

#KV缓存 #SVD #DeepSeek-R1-Distill-Llama-8B #Qwen3-8B

arXiv: DeepSeek@Yiren Zhao et al. (6)Source

Sep 15

Papers BuildersSelected13:05

LLMs identify JavaScript code vulnerabilities

For developers, this research shows that LLMs can be a practical tool for finding common JavaScript security issues, though they should be used alongside existing security tools.

#LLM #vulnerability detection #JavaScript #Gemini

arXiv: DeepSeek@Manit Kaushik et al. (5)Source

Sep 9

Papers BuildersSelected10:32

KV-Cache Timing Side Channels in LLM Serving

The study reveals how multi-tenant contention affects KV-cache timing attacks, showing measurements on quiet systems can overestimate operational attack reliability.

#DeepSeek-R1-Distill-Llama-8B #vLLM #KV-cache #timing side channels

arXiv: DeepSeek@Rana Abu BakarSource