Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Add FP8 Support #4

Merged
merged 2 commits into from
Jun 25, 2024
Merged

Add FP8 Support #4

merged 2 commits into from
Jun 25, 2024

Conversation

Satrat
Copy link
Contributor

@Satrat Satrat commented Jun 24, 2024

Migrating this PR from sparseml: neuralmagic/sparseml#2306

  • add code for inferring compression format from model
  • e2e fp8 example
  • e2e tests

@@ -0,0 +1,19 @@
quant_stage:
quant_modifiers:
GPTQModifier:
Copy link
Collaborator

@robertgshaw2-redhat robertgshaw2-redhat Jun 24, 2024

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

nice that gptq works out of the box

but note that we shouldnt need this for fp8

so in our examples its should just be quantization modifier

perhaps we should have a test that matches the expected user flow for fp8 as well?

Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Can do as follow up

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Gotcha, I just updated the PR with this change (both in the example and the integration test). However the nice scheme/target UX is only for GPTQModifier so it made the example a bit messier. I'll move the scheme parsing to QM in a follow up PR

@Satrat Satrat merged commit dcca2be into main Jun 25, 2024
8 of 12 checks passed
@Satrat Satrat deleted the fp8_support branch June 25, 2024 18:14
markmc pushed a commit to markmc/llm-compressor that referenced this pull request Nov 13, 2024
* Apply quantization config implementation

* add TODO

* integrate full lifecycle support, QuantizationStatus updates, add tinyllama test

* fix comment
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants