跳到正文
实时
Hugging Face Daily Papers·· 23 天前65

TTR-Compressor 重塑序列压缩技术

Rethinking Sequence Compression with Test-time Regression

要点速览

TTR-Compressor 作为标准池化与投影压缩的替代方案,在高效大模型块状稀疏注意力压缩和 MLLM 视听域投影(合并器)方面均显示显著实证优势。该技术可拓展至视频/音频时序压缩、原生高效 ViT 训练及深度感知压缩等新应用领域。

正文

Hi everyone! We introduce TTR-Compressor, a superior and fully practical alternative to standard pooling and projection-based compression.

It shows strong empirical gains both as a compressor for block-wise sparse attention in efficient LLMs and as a VL projector (merger) in MLLMs.

Beyond that, it opens up many exciting applications to explore: temporal compression for video/audio, training natively efficient ViTs from scratch, depth-wise compression for better depth scaling...

来源: Hugging Face Daily Papers · huggingface.co