[Quant][Kernel] Remove GPTQ Group/Dynamic Activation Ordering (#54809)
Signed-off-by: Roderick-Wu <roderickwu2003@gmail.com> Signed-off-by: Roderick Wu <roderickwu2003@gmail.com> Signed-off-by: mgoin <mgoin64@gmail.com> Co-authored-by: mgoin <mgoin64@gmail.com>
This commit is contained in:
@@ -238,6 +238,13 @@ class QuantizationConfig(ABC):
|
||||
}
|
||||
return WeightsMapper(orig_to_new_regex=orig_to_new_regex)
|
||||
|
||||
@staticmethod
|
||||
def get_checkpoint_weight_mapper() -> "WeightsMapper":
|
||||
"""Discard activation-order metadata unused by supported kernels."""
|
||||
from vllm.model_executor.models.utils import WeightsMapper
|
||||
|
||||
return WeightsMapper(orig_to_new_suffix={".g_idx": None})
|
||||
|
||||
def apply_vllm_mapper( # noqa: B027
|
||||
self, hf_to_vllm_mapper: "WeightsMapper"
|
||||
):
|
||||
|
||||
Reference in New Issue
Block a user