Instructions to use yiting/PerspectiveVision-LLaVA-LoRA with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- PEFT
How to use yiting/PerspectiveVision-LLaVA-LoRA with PEFT:
from peft import PeftModel from transformers import AutoModelForCausalLM base_model = AutoModelForCausalLM.from_pretrained("liuhaotian/llava-v1.5-7b") model = PeftModel.from_pretrained(base_model, "yiting/PerspectiveVision-LLaVA-LoRA") - Notebooks
- Google Colab
- Kaggle
Prompt template for inference?
#1
by vtu81 - opened
Hi, I'm wondering if you can add more instruction (e.g., prompt template) on how to inference with the model? I've tried the prompt template in the paper (Figure 9), but it seems not working well (the model consistently outputs Safety Status: Safe, Unsafe Category: None, even for the UnsafeBench train set images).