AI Paper Atlas

Research map / Architectures and efficiency / Pruning and efficient tuning

DiffFit: Unlocking Transferability of Large Diffusion Models via Simple Parameter-efficient Fine-Tuning

Enze Xie, Lewei Yao, Han Shi, Zhili Liu, Daquan Zhou, Zhaoqiang Liu, Jiawei Li, Zhenguo Li

ICCV 2023 · Architectures and efficiency › Pruning and efficient tuning · approach: Efficient ViT adaptation · 97 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add learnable scale factors to BitFit (ablation)BitFit on DiT-XL/2 · 8 downstream datasetsaverage FID: 16.82 -> 15.39Improved
DiffFit instead of full fine-tuningDiT-XL/2 · 8 downstream datasetsaverage FID: 16.59 -> 15.39Improved
use LoRA instead of DiffFitDiT-XL/2 · 8 downstream datasetsaverage FID: 15.39 -> 81.25Hurt
add scale factor to QKV-linear layers on BitFit (ablation)BitFit · Food101FID: 9.17 -> 7.37Improved
add scale factors in patch embedding or time embedding (ablation)BitFit · Food101FID: 9.17 -> 9.05No clear effect
restrict scale factors to the first 14 blocks instead of all 28 (ablation)DiffFit · Food101FID: 8.19 -> 7.61Improved
raise learning rate to 10x pre-training for parameter-efficient tuning (ablation)DiffFit · Food101FID: 15.68 -> 8.19Improved
raise learning rate from 10x to 20x (ablation)DiffFit · Food101FID: 8.19 -> 8.30Hurt
add positional encoding interpolation trick for 512 resolution (ablation)DiffFit · Food-101 512x512FID: 19.50 -> 19.10Improved
adapt low-resolution DiT to 512x512 with DiffFitDiT-XL/2 · ImageNet 512x512FID and training cost: 3.04 -> 3.02 FID; 1733 -> 51 GPU daysImproved

Later papers that compared with this method

1 later accepted paper reports DiffFit as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy