lllyasviel/ControlNet

Training still OOM on 8GB gpu.

Open

#21 opened on Feb 12, 2023

View on GitHub
 (2 comments) (3 reactions) (0 assignees)Python (27,526 stars) (2,500 forks)batch import
help wanted

Description

It seems that I already used many tricks but training still OOM for 8GB gpu. But inference is good now.

This is strange because I know some textural inversion or dreambooth can be trained on 8GB.

What is the secrect of Automatic1111's optimization? Although xformers may help a bit, the currect sliced attention should require even smaller mem than xformers.

Does it make sence to move text encoder and vae outside gpu when training?

Contributor guide

Tech stack
pythonpytorch
Domain
machine learning
Issue type
bug
DifficultyEstimated implementation difficulty for a new contributor, from 1 for very small changes to 5 for expert-level work.
3
Estimated timeA rough time range for an experienced contributor to investigate, implement, test, and prepare a pull request.
1-2 days
Activity statusHow available the issue appears right now: fresh, active, stale, blocked, or waiting on maintainer input.
stale
ClarityHow clearly the issue explains the expected change, acceptance criteria, and next step.
needs investigation
Prerequisites
PyTorch basicsGPU memory managementdiffusion modelsattention mechanisms
Newbie friendlinessA 1-100 score estimating how approachable this issue is for first-time contributors.
20
Research direction
Investigate the memory usage during training in ControlNet. Focus on the attention mechanism and compare with xformers. Profile memory using PyTorch's memory tools. Consider moving text encoder and VAE to CPU as suggested. Check if there are any known optimizations from Automatic1111's repository.