返回全部动态
Liquid AI 发布 LFM2.5 Q4_0 量化感知蒸馏检查点
原标题：LFM2.5 Q4\_0 Checkpoints from Quantization-Aware Distillation
AI 摘要
Liquid AI 发布了 LFM2.5 系列模型的 Q4_0 量化检查点，采用量化感知蒸馏（QAD）技术，将高精度教师模型蒸馏为量化学生模型，恢复了 BF16 精度损失的 97%。这些检查点在保持 Q4_0 低内存和高吞吐的同时，质量接近或超过 Q5_K_M 和 Q4_K_M，并已在 Hugging Face 上开放下载。
以上摘要由 AI 生成，可能存在误差。事实请以原文为准。
正文节选
- Trained with Quantization-Aware Distillation (QAD): a high-precision teacher model is distilled into a quantized student model - Same memory and speed as native Q4_0: They keep the low memory footprint and high throughput of Q4_0 GGUFs - Recovery: 97% of their BF16 average accuracy lost to quantization is recovered For all four models, we compare their released GGUFs produced with post-training quantization (PTQ) against the trained QAD Q4_0 checkpoints on a benchmark suite spanning reasoning,
发布时间：—
抓取时间：2026-08-21 01:05
来源机构：Hugging Face