[Refactor] Remove dead quantization code and tests (#45454)
Signed-off-by: yewentao256 <zhyanwentao@126.com>
This commit is contained in:
@@ -156,7 +156,6 @@ def get_quantization_config(quantization: str) -> type[QuantizationConfig]:
|
||||
"quark": QuarkConfig,
|
||||
"moe_wna16": MoeWNA16Config,
|
||||
"torchao": TorchAOConfig,
|
||||
"auto-round": INCConfig,
|
||||
"inc": INCConfig,
|
||||
"mxfp4": Mxfp4Config,
|
||||
"gpt_oss_mxfp4": GptOssMxfp4Config,
|
||||
|
||||
@@ -486,9 +486,6 @@ class Fp8LinearMethod(LinearMethodBase):
|
||||
weight_bf16 = weight_fp8 * weight_scale
|
||||
return torch.nn.functional.linear(x, weight_bf16.t(), bias)
|
||||
|
||||
if self.use_marlin:
|
||||
return self.fp8_linear.apply_weights(layer, x, bias)
|
||||
|
||||
return self.fp8_linear.apply_weights(layer, x, bias)
|
||||
|
||||
|
||||
|
||||
Reference in New Issue
Block a user