Event ·Multi-source

BitDecoding-FP4:FP4 KV 缓存解码的算法-系统协同设计

BitDecoding-FP4 针对长上下文 LLM 推理中 KV 缓存占用内存和带宽的问题,提出原生 FP4 KV 解码方案。算法层面用 Dual-Residual Query Quantization 将 query logit 误差从 60% 降到 1%,并用 Underflow-Aware Probability Scaling 和 Outlier-Aware Key Quantization 抑制另外两类误差。系统层面通过转置 softmax 融合概率量化消除 softmax 瓶颈。在 NVIDIA B300 上对 Qwen3-4B 和 Qwen3-30B-A3B 的测试中,精度保持在 BF16 的 0.5 点(NVFP4)或 1 点(MXFP4)以内,峰值服务吞吐提升 2.4-3.0 倍。代码已在 GitHub 开源。

Current read

KV 缓存量化到 FP4 还不掉精度,B300 上吞吐直接翻倍多,长上下文部署的可以看看这份开源实现。

11 sources56° AI heatLast updated 2026/11/6 18:13:54

Evidence

11 sources
Related source 2rohanpaul_ai
View source
Related source 6Stanford AI Lab
View source
Related source 9Artificial Analysis
View source
Related source 10marktechpost
View source

Conflict check

Dedup data only shows that these sources discuss the same event, not that they agree. There is no structured supporting or refuting evidence yet, so no conflict is inferred automatically.