[Bugfix] Lower gemma's unloaded_params exception to warning #7002
Conversation
This file contains hidden or 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
|
👋 Hi! Thank you for contributing to the vLLM project. Once the PR is approved and ready to go, please make sure to run full CI as it is required to merge (or just use auto-merge). To run full CI, you can do one of these:
🚀 |
robertgshaw2-redhat
approved these changes
Aug 1, 2024
Alvant
pushed a commit
to compressa-ai/vllm
that referenced
this pull request
Oct 26, 2024
…ject#7002) Signed-off-by: Alvant <alvasian@yandex.ru>
LeiWang1999
pushed a commit
to LeiWang1999/vllm-bitblas
that referenced
this pull request
Mar 26, 2025
…ject#7002) Signed-off-by: LeiWang1999 <leiwang1999@outlook.com>
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Outdated suggestions cannot be applied.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.
FIX #6957
Sometimes we add extra tensors when constructing quantized models in order to cover all permutations of checkpoints. In the linked issue, Gemma currently fails when loading FP8 quantization because we add k/v_scale parameters to be loaded in the case that calibrated kv cache scales are available. Since Gemma has a strict "no unloaded parameters" check, it fails in this case. To quickly unblock, I propose making this exception a warning. Eventually we can be more strict in allowing or disallowing unloaded parameters for all models, rather than just some.