modelscope / modelscope/DiffSynth-Studio

num_inference_steps of Wan2.1

Open
#767 1 comment 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
Python
Stars
13.1k
Forks
1.3k
Avg merge
13h 12m
Merged PRs (30d)
45

Description

Hi.
For the Wan2.1 model, num_inference_steps is set to 1000 during training and 50 during inference. How are these two values (1000 and 50) determined? Is it possible to use other values? For example, if I want to speed up inference, can I reduce num_inference_steps during both training and inference? If so, how should I determine the appropriate values to use for training and inference?
Thanks!

Contributor guide

No contributing guide indexed for this repository

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

No file, test, or entry point is named. Start by locating the Wan2.1 training and inference configuration, then trace how num_inference_steps is consumed in each path. Done means documenting how 1000 and 50 are chosen, whether other values are supported, and how changing them affects training and inference speed or quality.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
machine-learning
Issue type
Documentation
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
25/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.