Enable Llama 3.1 405B in FP8 (#124) #1745
Open
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
* add changes for fix * add keepmoduleonhosst, modfy quant json * remove buffer check * add llama 405 checks * remove hardcoded path, reuse module on host check * fix: undefined variable * remove unused import --------- Co-authored-by: Your Name <[email protected]>
libinta
added
run-test
jiminha
approved these changes
Feb 6, 2025
regisss
reviewed
Feb 7, 2025
| @@ -429,8 +429,13 @@ def setup_distributed_model(args, model_dtype, model_kwargs, logger): | |||
|
|
|||
| logger.info("DeepSpeed is enabled.") | |||
| deepspeed.init_distributed(dist_backend="hccl") | |||
| config = AutoConfig.from_pretrained(args.model_name_or_path, torch_dtype=model_dtype, **model_kwargs) | |||
There was a problem hiding this comment.
We cannot remove torch_dtype=model_dtype here and then use config.torch_dtype afterwards. Because we won't be able to specify anymore if we want to run the model in fp32 or in bf16.
Why do you need to do that?
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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Outdated suggestions cannot be applied.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.
add changes for fix
add keep_moduleon_host, modify quant json
remove buffer check
add llama 405 checks
remove hardcoded path, reuse module on host check
fix: undefined variable
remove unused import
What does this PR do?
Fixes # (issue)
Before submitting