Add :type option to load model under specific precision #311
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Allows for
Bumblebee.load_model({:hf, "..."}, type: :bf16)
to set mixed precision policy on the model and cast params on load. The user can also pass the policy struct itself, type is just a shorthand.Note: currently loading a f16 checkpoint returns params in f16, but that's somewhat a bug, specifically if all params are available we skip Axon init, which would otherwise cast the params to f32. Now it's going to always be f32 by default and the user can override it using
:type
.hf/transformers look at
"torch_dtype"
in the config file to determine the type, but I'm not sure if we want to configure the policy automatically based on that. @seanmor5 thoughts?