Skip to content

[enhancement]: T5-XXL and CLIP text encoders as single files #9736

Description

@Pfannkuchensack

Is there an existing issue for this?

  • I have searched the existing issues

What should this feature add?

Install T5-XXL and CLIP-L/CLIP-G as single safetensors files. Today T5 loads only as a Diffusers-style folder, GGUF,
SDNQ or bitsandbytes int8, and CLIP only as a Diffusers folder; the single files most users already have from
ComfyUI (t5xxl_fp16.safetensors, t5xxl_fp8_e4m3fn.safetensors, clip_l.safetensors, clip_g.safetensors)
do not install as encoders.

Alternatives

Install the Diffusers folders or the T5 GGUF.

Additional Content

Affects FLUX.1 and SD 3, and SDXL/SD 1.x where CLIP is chosen separately.

The full matrix of what loads today is in #9726 (Model Format Support page).

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions