Fix issue in loading mixed precision vocab pruned models during torchtune generation for evaluation

[ifed-ucsd]

Summary: Fix issue in loading mixed precision vocab pruned models during torchtune generation for evaluation

Reviewed By: RylanC24

Differential Revision: D66029040

[pytorch-bot[bot]]

:link: Helpful Links

:test_tube: See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/torchtune/2043

• :page_facing_up: Preview Python docs built from this PR

Note: Links to docs will display an error until the docs builds have been completed.

:white_check_mark: No Failures

As of commit 89dea365afcdc41ecb6cbef4e7fc7c7edc890f51 with merge base a9aadf5d24696fa8790c52739f8edd5ebd298881 (): :green_heart: Looks good so far! There are no failures yet. :green_heart:

This comment was automatically generated by Dr. CI and updates every 15 minutes.

[facebook-github-bot]

This pull request was exported from Phabricator. Differential Revision: D66029040

[facebook-github-bot]

This pull request was exported from Phabricator. Differential Revision: D66029040

[facebook-github-bot]

This pull request was exported from Phabricator. Differential Revision: D66029040

[facebook-github-bot]

This pull request was exported from Phabricator. Differential Revision: D66029040

[facebook-github-bot]

This pull request was exported from Phabricator. Differential Revision: D66029040

[facebook-github-bot]

This pull request was exported from Phabricator. Differential Revision: D66029040

[facebook-github-bot]

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