Skip to content

Mistral-format support for compressed-tensors #16803

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.

Sign up for GitHub

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

Merged
merged 5 commits into from
Apr 23, 2025
Merged

Mistral-format support for compressed-tensors #16803

Changes from all commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
c9b3451
Support dynamic activations for mistral fp8
mgoin Apr 16, 2025
d4967e2
Add compressed-tensors support
mgoin Apr 17, 2025
7e351c6
Extra cruft
mgoin Apr 17, 2025
91e8113
Fix type
mgoin Apr 17, 2025
7c44db3
Merge branch 'main' into dynamic-act-mistral-fp8
mgoin Apr 22, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
6 changes: 6 additions & 0 deletions vllm/transformers_utils/config.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -690,6 +690,9 @@ def recurse_elems(elem: Any):
"quant_method": "fp8",
"activation_scheme": "static"
}
elif quantization.get("quant_method") == "compressed-tensors":
# Pass through the quantization config to compressed-tensors
quantization_config = quantization
else:
raise ValueError(
f"Found unknown quantization='{quantization}' in config")
Expand All @@ -707,13 +710,16 @@ def recurse_elems(elem: Any):

if config_type == "multimodal":
multimodal_config = config_dict.pop("vision_encoder")
quantization_config = config_dict.get("quantization_config", {})

config_dict = {
"text_config": config_dict,
"vision_config": multimodal_config
}
config_dict["architectures"] = ["PixtralForConditionalGeneration"]
config_dict["model_type"] = "pixtral"
if quantization_config:
config_dict["quantization_config"] = quantization_config

config_dict.update(kwargs)

Expand Down