huggingface / huggingface/diffusers

Z-Image's modular TE block only produces negative_prompt_embeds when the pipeline's guider is enabled (CFG active).

Open
#13,505 4 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

bug modular-pipelines needs-code-example
Dominant language
Python
Stars
34.5k
Forks
7.3k
Avg merge
3d 3h
Merged PRs (30d)
91

Description

Z-Image's modular TE block only produces negative_prompt_embeds when the pipeline's guider is enabled (CFG active).

Even if i give a negative prompt, the block does not produce a negative_prompt_embeds.

The modular approach is too much dependent on the pipeline, making it no more modular to my point of view. That is the case also with this init_pipeline() method and the dependency on the pipeline to find for the execution device. Should be more modular to my point of view.

@yiyixuxu

Contributor guide

Open the contributing guide

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

The issue names the Z-Image modular TE block and init_pipeline(); start by locating those entry points and reproducing the negative-prompt case with CFG disabled. Done means the block produces negative_prompt_embeds from a negative prompt without requiring the pipeline guider, and its execution-device setup is less dependent on the pipeline.

Written by the indexing model from the issue text.

Assessment

Tech stack
python, pytorch
Domain
machine-learning
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Quiet
Clarity
Mostly clear
Newbie friendliness
48/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.