Thanks to visit codestin.com
Credit goes to github.com

Skip to content

> context length? #48

@hlchen23

Description

@hlchen23

in the case of paper(cook egg), the model generate interleaved image-text which contains 5 images (~ 5K tokens), but chameleon is trained with 4K context. So I think it can not generate such good outputs, and I do experiments and find it cannot generate 5 images.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions