1
英伟达 DLSS 5 混合精度 Mod 测试：RTX 50 系列显卡性能仅提升 1% 至 2%
IT之家 9 月 12 日消息，在英伟达 DLSS 5 神经渲染 DLL 文件流出后，各路民间大神纷纷下场进行调试和优化，还有 Mod 开发者尝试通过降低模型推理成本以提升游戏性能。 根据 X 用户 SwurvGaming 分享的情报，有 Mod 开发者将 FP8 与 NVFP4 混合精度应用于 D…
2
MixFrag: Fragility-Guided Mixed-Precision Post-Training Quantization for Vision Transformers
视觉Transformer部署提速新思路，用脆弱性引导混合精度量化，兼顾压缩率与精度的双赢方案。
arXiv:2607.28589v1 Announce Type: cross Abstract: Post-training quantization (PTQ) has emerged as an effective solution for deploying Vision Transform…
3
PM-KVQ: Progressive Mixed-precision KV Cache Quantization for Long-CoT LLMs
针对长思维链大语言模型的KV缓存量化新突破，渐进混合精度方案，已被ICLR 2026接收，代码已开源。
arXiv:2505.18610v2 Announce Type: replace Abstract: Recently, significant progress has been made in developing reasoning-capable Large Language Models…
4
TileFuse: A Fused Mixed-Precision Kernel Library for Efficient Quantized LLM Inference on AMD NPUs
面向AMD NPU的融合混合精度内核库，显著提升量化大语言模型推理效率。
arXiv:2606.11357v1 Announce Type: cross Abstract: With the growing demand for on-device LLM inference, edge SoCs increasingly integrate NPUs to improv…
5
Joint Structural Pruning and Mixed-Precision Quantization for LLM Compression
联合结构化剪枝与混合精度量化，为LLM压缩提供高效新方案，兼顾精度与速度。
arXiv:2606.07819v1 Announce Type: cross Abstract: Recently, the efficiency of Large Language Models (LLMs) deployment has become a critical concern in…
6
Channel-Wise Mixed-Precision Quantization for Large Language Models
大语言模型量化新突破：通道级混合精度策略如何在保持性能同时压缩模型？
arXiv:2410.13056v4 Announce Type: replace-cross Abstract: Large Language Models (LLMs) have demonstrated remarkable success across a wide range of lan…
7
KForge: LLM-Driven Cross-Platform Kernel Generation for AI Accelerators
用大模型自动生成跨平台AI加速器内核，通过算子融合与混合精度提升算力效率
arXiv:2606.02963v1 Announce Type: new Abstract: Production inference increasingly targets a heterogeneous mix of accelerators. Agentic pipelines inter…
8
WINDQuant: Weight-Informed Neural Decision-Making for Global Mixed-Precision LLM Quantization
权重指导神经决策，实现大模型全局混合精度量化，性能与效率的新平衡。
arXiv:2605.26660v1 Announce Type: new Abstract: Quantization is an effective approach to reduce the memory footprint and inference cost of large langu…
9
MoBiQuant: Mixture-of-Bits Quantization for Token-Adaptive Any-Precision LLM
MoBiQuant通过混合位数量化实现token自适应任意精度，大幅降低大模型内存与计算开销，推理更高效。
arXiv:2602.20191v2 Announce Type: replace-cross Abstract: Dynamic runtime latency and memory constraints necessitate flexible large language model (LL…
10
CONF-KV: Confidence-Aware KV Cache Eviction with Mixed-Precision Storage for Long-Horizon LLM
利用解码步骤的置信度信号，提出混合精度存储的KV缓存逐出策略，有效缓解长序列推理中的GPU内存瓶颈。
arXiv:2605.24786v1 Announce Type: cross Abstract: Long-horizon LLM inference turns the key--value (KV) cache into the dominant GPU memory consumer and…
11
GAMMA: Global Bit Allocation for Mixed-Precision Models under Arbitrary Budgets
提出GAMMA方法，实现任意预算下大语言模型混合精度量化的全局比特自动分配，无需量化感知训练。
arXiv:2605.18475v1 Announce Type: new Abstract: Mixed-precision quantization improves the budget--accuracy trade-off for large language models (LLMs) …
12
Dynamic Mixed-Precision Routing for Efficient Multi-step LLM Interaction
提出动态混合精度路由方法，在多步LLM交互中实现高效推理，在精度与效率间取得平衡。
arXiv:2602.02711v2 Announce Type: replace Abstract: Large language models (LLMs) achieve strong performance in long-horizon decision-making tasks thro…