0
0

Delete article

Deleted articles cannot be recovered.

Draft of this article would be also deleted.

Are you sure you want to delete this article?

More than 1 year has passed since last update.

ICML2024の量子化・枝刈り論文

Last updated at Posted at 2024-08-13

概要

この記事では、ICML2024の量子化・枝刈り論文を紹介します。1

Outlier-aware Slicing for Post-Training Quantization in Vision Transformer

  • https://openreview.net/forum?id=Uh5XN9d2J4

  • 概要:PTQの改善

  • 新規性:"Reconstraction granularity"の導入

  • キモ:量子化誤差最適化の粒度は、Transformer-likeは粗く、それ以外は細かくする。
    image.png

ERQ: Error Reduction for Post-Training Quantization of Vision Transformers

LayerMerge: Neural Network Depth Compression through Layer Pruning and Merging

  • https://openreview.net/forum?id=uDoy7AGvEC

  • 概要:DNNのレイヤをマージして軽量化する

  • 新規性:Fig.2 のようなマージを最適に求める

  • キモ:(4)式と動的計画法で効率的に最適化問題を解く
    image.png

Pruner-Zero: Evolving Symbolic Pruning Metric From Scratch for Large Language Models

  • https://openreview.net/forum?id=1tRLxQzdep

  • 概要:枝刈りの指標

  • 新規性:演算の組み合わせの最適化で指標を探索する点

  • キモ:Table 10を組み合わせた指標から、進化的アルゴで最適な組み合わせを探索する。スコアはLlama-2から実際に刈って確かめた
    image.png

  1. 画像や数式は論文から引用しています。

0
0
0

Register as a new user and use Qiita more conveniently

  1. You get articles that match your needs
  2. You can efficiently read back useful information
  3. You can use dark theme
What you can do with signing up
0
0

Delete article

Deleted articles cannot be recovered.

Draft of this article would be also deleted.

Are you sure you want to delete this article?