kohya-ss / kohya-ss/sd-scripts

About sdxl_merge_lora.py using GPU for merging

Open
#1,574 1 comment 0 reactions 0 assignees View on GitHub
Dominant language
Python
Stars
7.2k
Forks
1.2k
Avg merge
11m
Merged PRs (30d)
2

Description

"Can an option to merge GPUs be added? Merging XL LoRA takes at least 10 minutes or more to complete, and sometimes it even fails. The CPU is having a hard time.

My notebook configuration:
![image](https://github.com/user-attachments/assets/2f5ab20d-d31d-467e-b766-1f39d7a6c818)

Contributor guide

No contributing guide indexed for this repository

Research direction

Start with sdxl_merge_lora.py and inspect how XL LoRA merging currently uses the CPU and handles failures. Check the repository for related merge scripts or tests before deciding where an option for GPU-based merging belongs. Done means XL LoRA merging can use GPUs and completes more reliably than the reported CPU path.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
machine-learning, performance
Issue type
Feature
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Mostly clear
Newbie friendliness
35/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.