-
Notifications
You must be signed in to change notification settings - Fork 295
Load Compressed Tensors Models with run_compressed=False #923
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Conversation
| "create_fake_dataloader", | ||
| "POSSIBLE_TOKENIZER_FILES", | ||
| "download_repo_from_huggingface_hub", | ||
| "detect_last_checkpoint", |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
alphabetical order
|
👋 Hi! Thank you for contributing to llm-compressor. Please add the ready label when the PR is ready for review. |
kylesayrs
left a comment
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Beautiful
| config.quantization_config["run_compressed"] = False | ||
|
|
||
| return AutoModelForCausalLM.from_pretrained( | ||
| pretrained_model_name_or_path, config=config, **kwargs |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Wont we get the error you were seeing unless we set the status to Frozen in HFQuant? In which case, the test below will error out?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Yes good point. The module set to frozen will be triggered from the HFQuantizer side
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
If this will error until then, we should wait until your change lands in HF and keep this in draft form.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
yes wait until its merged
Co-authored-by: Kyle Sayers <[email protected]>
dsikka
left a comment
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Can you set this to draft since blocked on hf
|
@dsikka |
|
Dont need this no more, closing this pr |
SUMMARY:
Load Compressed Tensors Models with run_compressed=False
TEST PLAN:
Made tests to check that run_compressed=False (new pathway) do not have any Compressed Linear
Made tests to check that run_compressed=True (AutoModelForCausalLM pathway) has Compressed Linear.