Skip to content
This repository was archived by the owner on Jun 3, 2025. It is now read-only.

Conversation

rahul-tuli
Copy link
Member

This PR incorporates changes from @abhinavnmagic's PR #2222 into new modifier UX

We introduce a new argument preserve_sparsity_mask in SparseGPTModifier which can be used to extend or ignore the base weight mask during SparseGPT application

Test Recipe:

test_stage:
  obcq_modifiers:
    SparseGPTModifier:
      sparsity: 0.5
      block_size: 128
      percdamp: 0.01
      mask_structure: "2:4"
      targets: ["re:model.layers.\\d*$"]
      preserve_sparsity_mask: true

@rahul-tuli rahul-tuli changed the base branch from main to preserve-sparsity-gptq May 13, 2024 14:03
Base automatically changed from preserve-sparsity-gptq to gptq-ux-config-groups May 17, 2024 16:15
@rahul-tuli rahul-tuli merged commit 446555f into gptq-ux-config-groups May 17, 2024
@rahul-tuli rahul-tuli deleted the preserve-sparsity-sparsegpt branch May 17, 2024 16:15
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants