提出W4A4量化方案，在Wan2.2视频生成模型上实现高效推理，挑战低比特大模型量化极限
arXiv:2606.29337v1 Announce Type: new Abstract: We summarize our submission to Sub-Challenge 1: W4A4 Quantization for Inference (HiF4 / MXFP4) of the …
ICML 2026收录，提出NeUQI量化参数初始化方法，实现低比特LLM接近最优的均匀量化，兼顾压缩与精度。
arXiv:2505.17595v4 Announce Type: replace Abstract: Large language models (LLMs) achieve impressive performance across domains but face significant ch…
LFQ方法通过关注logit感知的最终块量化，显著提升低比特LLM生成质量，已被ICML 2026接收。
arXiv:2605.29756v1 Announce Type: new Abstract: As large language models continue to scale, low-bit weight-only post-training quantization (PTQ) offer…
低比特LLM量化新突破：通过重塑激活分布提升精度，值得AI从业者关注。
arXiv:2605.26175v1 Announce Type: cross Abstract: Low-bit activation quantization remains a major bottleneck in efficient large language model (LLM) d…
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…
国产算力里程碑：基于华为昇腾训练的1.58-bit端侧大模型开源，显存效率提升6倍！
IT之家 5 月 25 日消息，面壁智能联合清华大学、OpenBMB 开源社区，今天正式发布并开源其在低比特大模型训练方向的最新成果 ——BitCPM-CANN。 官方表示，这是中国首个完全基于国产算力平台（华为昇腾）实现端到端训练并开源的三值（1.58-bit）大模型。从量化算子、训练算法到全链路…
LLM极低比特量化新突破：SignRoundV2大幅缩小性能差距，实现高效训练后量化
arXiv:2512.04746v2 Announce Type: replace Abstract: Extremely low-bit quantization is critical for efficiently deploying Large Language Models (LLMs),…
MXFP4量化新突破：两级正交旋转显著提升大模型低比特推理精度，部署友好。
arXiv:2605.19561v1 Announce Type: new Abstract: As Large Language Models (LLMs) advance toward practical deployment, the Microscaling FP4 (MXFP4) form…
揭秘开源LLM激活动态范围新发现，挑战旧有量化认知，影响推理效率优化。
arXiv:2605.15572v1 Announce Type: new Abstract: The dynamic range of activations is a first-order constraint for low-bit quantization, activation scal…