Update app.py

app.py

@@ -168,14 +168,16 @@ pipe = load_and_prepare_model()
 #checkpoint = "microsoft/Phi-3.5-mini-instruct"
 checkpoint = "ford442/Phi-3.5-mini-instruct-bf16"
 #captioner = pipeline(model="ydshieh/vit-gpt2-coco-en",device='cuda:0', task="image-to-text")
-captioner_2 = pipeline(model="Salesforce/blip-image-captioning-base",device='cuda', task="image-to-text")
-#captioner_3 = pipeline(model="ford442/blip-image-to-text-large-bf16",device='cuda', task="image-to-text")
-model5 = Blip2ForConditionalGeneration.from_pretrained("ford442/blip2-image-to-text-bf16").to('cuda')
-processor5 = Blip2Processor.from_pretrained("ford442/blip2-image-to-text-bf16", device_map='cuda')
+#captioner_2 = pipeline(model="Salesforce/blip-image-captioning-base",device='cuda', task="image-to-text")
+captioner_2 = pipeline(model="ford442/blip-image-to-text-large-bf16",device='cuda', task="image-to-text")
+#model5 = Blip2ForConditionalGeneration.from_pretrained("ford442/blip2-image-to-text-bf16").to('cuda')
+#processor5 = Blip2Processor.from_pretrained("ford442/blip2-image-to-text-bf16", device_map='cuda')
 #txt_tokenizer = AutoTokenizer.from_pretrained(checkpoint, device_map='cuda', add_prefix_space=False)
 #txt_tokenizer.tokenizer_legacy=False
 #model = Phi3ForCausalLM.from_pretrained(checkpoint).to('cuda:0')
 #model = AutoModelForCausalLM.from_pretrained(checkpoint, device_map='cuda') #.to('cuda')
+model5 = InstructBlipForConditionalGeneration.from_pretrained("Salesforce/instructblip-vicuna-7b").to('cuda')
+processor5 = InstructBlipProcessor.from_pretrained("Salesforce/instructblip-vicuna-7b")
 
 ip_model = IPAdapterXL(pipe, local_folder, ip_ckpt, device)
 text_encoder_1=CLIPTextModel.from_pretrained('ford442/RealVisXL_V5.0_BF16', subfolder='text_encoder',token=True) #.to(device=device, dtype=torch.bfloat16)