From b9da974eb739f6b8efa69df9f8e33983fae24a76 Mon Sep 17 00:00:00 2001 From: somebody Date: Mon, 14 Aug 2023 00:56:40 -0500 Subject: [PATCH] GenericHFTorch: Change use_4_bit to quantization in __init__ --- modeling/inference_models/generic_hf_torch/class.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/modeling/inference_models/generic_hf_torch/class.py b/modeling/inference_models/generic_hf_torch/class.py index a059ebb0..a7734e7d 100644 --- a/modeling/inference_models/generic_hf_torch/class.py +++ b/modeling/inference_models/generic_hf_torch/class.py @@ -29,7 +29,7 @@ model_backend_type = "Huggingface" #This should be a generic name in case multip class model_backend(HFTorchInferenceModel): def __init__(self) -> None: super().__init__() - self.use_4_bit = False + self.quantization = False def is_valid(self, model_name, model_path, menu_path): base_is_valid = super().is_valid(model_name, model_path, menu_path) @@ -72,7 +72,7 @@ class model_backend(HFTorchInferenceModel): else: logger.warning("Bitsandbytes is not installed, you can not use Quantization for Huggingface models") return requested_parameters - + def set_input_parameters(self, parameters): super().set_input_parameters(parameters) self.quantization = parameters['quantization'] if 'quantization' in parameters else False