remove argument 'save_dir' in method from_quantized

auto_gptq/modeling/_base.py

@@ -676,8 +676,7 @@ class BaseGPTQForCausalLM(nn.Module, PushToHubMixin):
     @classmethod
     def from_quantized(
         cls,
-        model_name_or_path: Optional[str] = None,
-        save_dir: Optional[str] = None,
+        model_name_or_path: Optional[str],
         device_map: Optional[Union[str, Dict[str, Union[int, str]]]] = None,
         max_memory: Optional[dict] = None,
         device: Optional[Union[str, int]] = None,
@@ -726,14 +725,6 @@ class BaseGPTQForCausalLM(nn.Module, PushToHubMixin):
             use_triton = False
 
         # == step1: prepare configs and file names == #
-        if model_name_or_path and save_dir:
-            logger.warning("save_dir will be ignored because model_name_or_path is explicit specified.")
-        if not model_name_or_path and save_dir:
-            model_name_or_path = save_dir
-            warnings.warn("save_dir is deprecated and will be removed in version 0.3.0", PendingDeprecationWarning, stacklevel=2)
-        if not model_name_or_path and not save_dir:
-            raise ValueError("at least one of model_name_or_path or save_dir should be specified.")
-        
         config = AutoConfig.from_pretrained(model_name_or_path, trust_remote_code=trust_remote_code, **cached_file_kwargs)
 
         if config.model_type not in SUPPORTED_MODELS:

auto_gptq/modeling/auto.py

@@ -64,8 +64,7 @@ class AutoGPTQForCausalLM:
     @classmethod
     def from_quantized(
         cls,
-        model_name_or_path: Optional[str] = None,
-        save_dir: Optional[str] = None,
+        model_name_or_path: Optional[str],
         device_map: Optional[Union[str, Dict[str, Union[str, int]]]] = None,
         max_memory: Optional[dict] = None,
         device: Optional[Union[str, int]] = None,
@@ -82,9 +81,7 @@ class AutoGPTQForCausalLM:
         trainable: bool = False,
         **kwargs
     ) -> BaseGPTQForCausalLM:
-        model_type = check_and_get_model_type(
-            save_dir or model_name_or_path, trust_remote_code
-        )
+        model_type = check_and_get_model_type(model_name_or_path, trust_remote_code)
         quant_func = GPTQ_CAUSAL_LM_MODEL_MAP[model_type].from_quantized
         # A static list of kwargs needed for huggingface_hub
         huggingface_kwargs = [
@@ -107,7 +104,6 @@ class AutoGPTQForCausalLM:
         }
         return quant_func(
             model_name_or_path=model_name_or_path,
-            save_dir=save_dir,
             device_map=device_map,
             max_memory=max_memory,
             device=device,

examples/benchmark/generation_speed.py

@@ -165,7 +165,7 @@ def load_model_tokenizer(
         )
     else:
         model = AutoGPTQForCausalLM.from_quantized(
-            save_dir=model_name_or_path,
+            model_name_or_path,
             max_memory=max_memory,
             low_cpu_mem_usage=True,
             use_triton=use_triton,