[Quant] Consolidate GPTQ: rename gptq_marlin.py to auto_gptq.py (#38288)

Signed-off-by: Chengyi Nie <cnie@roblox.com>
Co-authored-by: Chengyi Nie <cnie@roblox.com>
Co-authored-by: Cursor <cursoragent@cursor.com>
This commit is contained in:
Chengyi Nie
2026-05-15 08:25:52 +08:00
committed by GitHub
co-authored by Chengyi Nie Cursor
parent 3b6a204789
commit fa2a33b893
21 changed files with 170 additions and 601 deletions
@@ -60,10 +60,9 @@ class MoeWNA16Config(QuantizationConfig):
# Avoid circular import
from vllm.model_executor.layers.quantization.awq import AWQConfig
from vllm.model_executor.layers.quantization.awq_marlin import AWQMarlinConfig
from vllm.model_executor.layers.quantization.gptq_marlin import GPTQMarlinConfig
if self.linear_quant_method == "gptq":
self.use_marlin = GPTQMarlinConfig.is_gptq_marlin_compatible(full_config)
pass
elif self.linear_quant_method in ("awq", "awq_marlin"):
capability_tuple = current_platform.get_device_capability()
device_capability = (
@@ -172,24 +171,18 @@ class MoeWNA16Config(QuantizationConfig):
return UnquantizedLinearMethod()
elif isinstance(layer, LinearBase):
# Avoid circular import
from vllm.model_executor.layers.quantization.auto_gptq import (
AutoGPTQConfig,
)
from vllm.model_executor.layers.quantization.awq import AWQConfig
from vllm.model_executor.layers.quantization.awq_marlin import (
AWQMarlinConfig,
)
from vllm.model_executor.layers.quantization.gptq import GPTQConfig
from vllm.model_executor.layers.quantization.gptq_marlin import (
GPTQMarlinConfig,
)
if self.linear_quant_method == "gptq":
if self.use_marlin:
return GPTQMarlinConfig.from_config(
self.full_config
).get_quant_method(layer, prefix)
else:
return GPTQConfig.from_config(self.full_config).get_quant_method(
layer, prefix
)
return AutoGPTQConfig.from_config(self.full_config).get_quant_method(
layer, prefix
)
elif self.linear_quant_method in ("awq", "awq_marlin"):
if self.use_marlin and check_marlin_supports_layer(
layer, self.group_size