Skip to content
This repository has been archived by the owner on Aug 7, 2024. It is now read-only.

Commit

Permalink
Update on "[bc-breaking] rename top level UX to `convert_to_float8_tr…
Browse files Browse the repository at this point in the history
…aining`"

Summary:

Old name: `swap_linear_with_float8_linear`
New name: `convert_to_float8_training`

Choosing a more generic name, with the following improvements from the
old name:
1. doesn't mention module swaps, which is an implementation detail
2. doesn't mention `Float8Linear`, which is an implementation detail
3. clarifies that this is for training, not to be confused with
   inference APIs
4. doesn't mention `linear`, which gives more freedom to add other
   modules later

```
find . -name '*.py' -print0 | xargs -0 sed -i 's/swap_linear_with_float8_linear/convert_to_float8_training/g'
```

Test Plan:

```
./test/test_everything.sh
```

Reviewers:

Subscribers:

Tasks:

Tags:

[ghstack-poisoned]
  • Loading branch information
vkuzo committed Jul 24, 2024
1 parent 65b14a0 commit d47fc95
Showing 1 changed file with 0 additions and 1 deletion.
1 change: 0 additions & 1 deletion float8_experimental/float8_linear_utils.py
Original file line number Diff line number Diff line change
Expand Up @@ -122,7 +122,6 @@ def post_order_traversal(
return root_module


# def convert_to_float8_training(
def convert_to_float8_training(
module: nn.Module,
*,
Expand Down

0 comments on commit d47fc95

Please sign in to comment.