simcop2387/AutoGPTQ
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages 1 Wiki Activity

remove argument 'save_dir' in method from_quantized

Browse source
This commit is contained in:
PanQiWei 2023-07-26 17:58:04 +08:00
parent 722a621aaa
commit ff1f100ded
3 changed files with 4 additions and 17 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

11
auto_gptq/modeling/_base.py
View file

@ -676,8 +676,7 @@ class BaseGPTQForCausalLM(nn.Module, PushToHubMixin):
@classmethod @classmethod
def from_quantized( def from_quantized(
cls, cls,
model_name_or_path: Optional[str] = None, model_name_or_path: Optional[str],
save_dir: Optional[str] = None,
device_map: Optional[Union[str, Dict[str, Union[int, str]]]] = None, device_map: Optional[Union[str, Dict[str, Union[int, str]]]] = None,
max_memory: Optional[dict] = None, max_memory: Optional[dict] = None,
device: Optional[Union[str, int]] = None, device: Optional[Union[str, int]] = None,
@ -726,14 +725,6 @@ class BaseGPTQForCausalLM(nn.Module, PushToHubMixin):
use_triton = False use_triton = False
# == step1: prepare configs and file names == # # == 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) 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: if config.model_type not in SUPPORTED_MODELS:

8
auto_gptq/modeling/auto.py
View file

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

2
examples/benchmark/generation_speed.py
View file

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