Use quantization_config in hf config (#1695)
This commit is contained in:
@@ -66,6 +66,7 @@ def get_model(model_config: ModelConfig) -> nn.Module:
|
||||
if model_config.quantization is not None:
|
||||
quant_config = get_quant_config(model_config.quantization,
|
||||
model_config.model,
|
||||
model_config.hf_config,
|
||||
model_config.download_dir)
|
||||
capability = torch.cuda.get_device_capability()
|
||||
capability = capability[0] * 10 + capability[1]
|
||||
|
||||
Reference in New Issue
Block a user