PTQ-Level Efficiency, QAT-Level Accuracy | 训练后量化的效率,量化感知训练的精度
raise vlm loading error (#863)
Wenhua Chengab94a4b3d47f
diff --git a/auto_round/utils.py b/auto_round/utils.pyindex 11eb4c343..38ec68de7 100644--- a/auto_round/utils.py+++ b/auto_round/utils.py@@ -1585,6 +1585,8 @@ def mllm_load_model()torch.cuda.get_device_capability = orig_funclogger.warning("the support for fp8 model as input is experimental, please use with caution.")+ else:+ raiseif "Mistral-Small-3.2" in pretrained_model_name_or_path:from mistral_common.tokens.tokenizers.mistral import MistralTokenizer # pylint: disable=E0401