Tag
KV4-RL提出了结合无校准INT4 KV缓存量化与注意力沉降和近期上下文的INT8选择性保护的方法，用于加速可验证奖励强化学习（RLVR）的异步rollout。在Qwen3-8B-Base GRPO训练中实现了1.80倍累积rollout加速，最终六基准平均准确率与BF16训练相差不到0.5个百分点。
The paper introduces HeadGuard, a composable method that protects a small subset of KV heads in high precision while applying low-bit quantization to the rest of a VLM's image KV cache, substantially recovering accuracy lost by aggressive 2-bit quantization across multiple models and benchmarks.
QuantWM 提出一种免训练的 2-bit KV cache 量化框架，通过量化敏感性感知聚类（QSAC）与主子空间注意力补偿（PSAC）来稳定视频世界模型的注意力 logits，缓解时间闪烁与视觉退化，并在多个基准上实现最高 6.20 倍 KV cache 内存压缩。
D-Quant is a flexible KV cache quantization framework using driftable entropy coding to convert variable-length representations into fixed-size bitstreams, enabling efficient parallel dequantization in attention kernels and achieving up to 7x memory reduction and 3.5x throughput improvement while preserving near BF16 performance.
The Gemma 4 QAT 31B model demonstrates improved behavior with KV cache quantization, suggesting enhanced inference efficiency.
OSCAR is an offline spectral covariance-aware rotation method for 2-bit KV cache quantization that aligns quantization with attention covariance structures, achieving high accuracy and efficiency for long-context LLM serving.
This paper analyzes KV cache quantization schemes inspired by TurboQuant, using statistical inference and a new 6D error framework to evaluate quality measures like KL divergence and geometric error.