initial commit

app.py

@@ -1,10 +1,12 @@
 import gradio as gr
-from transformers import Blip2ForVisualQuestionAnswering, Blip2Config, AutoTokenizer
+from transformers import AutoProcessor, AutoModelForVisualQuestionAnswering, AutoModelForCausalLM, AutoTokenizer
 from PIL import Image
 import torch
 
+model = AutoModelForCausalLM.from_pretrained("microsoft/git-base-vqav2")
 model_path = "microsoft/git-base-vqav2"
 dataset_name = "Multimodal-Fatima/OK-VQA_train"
+tokenizer = AutoTokenizer.from_pretrained(model_path)
 
 questions = ["What can happen the objects shown are thrown on the ground?",
              "What was the machine beside the bowl used for?",
@@ -17,7 +19,8 @@ questions = ["What can happen the objects shown are thrown on the ground?",
              "Where can that toilet seat be bought?",
              "What do you call the kind of pants that the man on the right is wearing?"]
 
-model = Blip2ForVisualQuestionAnswering.from_pretrained(model_path)
+processor = AutoProcessor.from_pretrained(model_path)
+model = AutoModelForVisualQuestionAnswering.from_pretrained(model_path)
 
 
 def main(select_exemple_num):