pytorch-labs / float8_experimental

This repository contains the experimental PyTorch native float8 training UX
BSD 3-Clause "New" or "Revised" License
211 stars 20 forks source link

delete Float8DynamicLinear #304

Closed vkuzo closed 4 months ago

vkuzo commented 4 months ago

Stack from ghstack (oldest at bottom):

Summary:

We are standardizing on Float8Linear as the only float8 linear object:

  1. the stack ending with https://github.com/pytorch-labs/float8_experimental/pull/300 moved all of the functionality of Float8DynamicLinear to Float8Linear. The default settings of Float8Linear are to use dynamic scaling.
  2. this PR deletes Float8DynamicLinear from the codebase and patches the relevant callsites in fbsource.

Test Plan:

// all tests pass
./test_everything.sh

// also run all benchmarks and verify correctness

Reviewers:

Subscribers:

Tasks:

Tags:

Differential Revision: D59342767

vkuzo commented 4 months ago

@vkuzo has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator.

facebook-github-bot commented 4 months ago

This pull request has been merged in pytorch-labs/float8_experimental@8e9623ac8173f83b6ab0ee0d86546730c70211bf.