Skip to content

TensorRT int8 - UFM #19

@ayushzenith

Description

@ayushzenith

Great work on UFM!

Curious if you have been able to successfully trt UFM to int8 precision while still getting meaningful performance and better inference speed than fp16. I haven't been able to trt or optimize UFM beyond fp16 and have it give me meaningful results. I suspect this could also just be a dinov2 problem but was curious if anyone has managed to do this!

Thank you

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type
    No fields configured for issues without a type.

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions