按年被提到
20242024 · 1 次2026
大模型与语言模型›量化、解码与分词·arXiv:2404.00456 · cs.LG

QuaRot: Outlier-Free 4-Bit Inference in Rotated LLMs

Saleh Ashkboos, Amirkeivan Mohtashami, Maximilian L. Croci, Bo Li, Pashmina Cameron, Martin Jaggi, Dan Alistarh, Torsten Hoefler, James Hensman · 2024
开始阅读 ↗PDFalphaXiv 讨论
被提及 · 2 人首次 2024-05-11 · 最近 2026-05-27
2不同的人
2活跃月份
2讨论串
115被引(OpenAlex)
arXiv 摘要(候选层不写导读)

We introduce QuaRot, a new Quantization scheme based on Rotations, which is able to quantize LLMs end-to-end, including all weights, activations, and KV cache in 4 bits. QuaRot rotates LLMs in a way that removes outliers from the hidden state without changing the output, making quantization easier. This computational invariance is applied to the hidden state (residual) of the LLM, as well as to the activations of the feed-forward components, aspects of the attention mechanism, and to the KV cache. The result is a quantized model where all matrix multiplications are performed in 4 bits, without any channels identified for retention in higher precision. Our 4-bit quantized LLaMa2-70B model has losses of at most 0.47 WikiText-2 perplexity and retains 99% of the zero-shot performance. We also show that QuaRot can provide lossless 6 and 8 bit LLaMa2 models without any calibration data using round-to-nearest quantization. Code is available at: https://github.com/spcl/QuaRot.

候选层:至少 2 个不同的人、在 2 个不同月份提到过,还没到主书架的 3 人 3 个月。没有做立场标注,所以只写“被提及”。

HN 书评

推荐在前,同类按时间 · 每篇最多保留 20 条,统计用全部提及
评论buildbot · 2026-05

“Not that it super matters, but random hadamards for quantization have been a thing since way before turboquant. https://arxiv.org/abs/2404.00456”

Matrix Multiplications on GPUs Run Faster When Given "Predictable" Data原评论 ↗
被提交为 story · 1 次