Skip to content

Some Problems on HiSVD + GPTQ implementation #1

Description

@ttxshang

Thank you for releasing the paper and code. I have three questions about the HiSVD + GPTQ-4bit result in Table 5:

  1. What HiSVD compression ratio or rank budget was used before quantization to achieve 2.8 GB memory?

  2. In compress.py, svd_u and svd_v are multiplied and written back as a dense weight. This does not reduce storage or inference memory. Is this operation only used for evaluation, while actual deployment stores the two low-rank factors separately?

  3. Did GPTQ quantize svd_u and svd_v separately, or quantize the reconstructed dense matrix svd_u @ svd_v? Is there a significant accuracy difference between these two approaches?

Could you provide the quantization code or configuration used for Table 5?

Activity

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

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions