Updates to run faster

app.py

@@ -3,18 +3,20 @@ import spaces
 
 import gradio as gr
 
-MODEL_ID = "davidr99/qwen2-7b-instruct-blackjack"
+MODEL_ID = "Qwen/Qwen2-VL-7B-Instruct"
+MODEL_FINETUNE_ID = "davidr99/qwen2-7b-instruct-blackjack"
 
 
 from transformers import Qwen2VLForConditionalGeneration, AutoTokenizer, AutoProcessor
 from qwen_vl_utils import process_vision_info
 
+model = Qwen2VLForConditionalGeneration.from_pretrained(MODEL_ID, torch_dtype="auto").to('cuda')
+model.load_adapter(MODEL_FINETUNE_ID)
+processor = AutoProcessor.from_pretrained(MODEL_FINETUNE_ID)
+
 @spaces.GPU(duration=30)
 def blackjack_ai(image):
 
-    model = Qwen2VLForConditionalGeneration.from_pretrained(MODEL_ID, torch_dtype="auto", device="auto")
-    processor = AutoProcessor.from_pretrained(MODEL_ID)
-
     instruction = "extract json from this image."
     
     messages = [

requirements.txt

@@ -4,4 +4,5 @@ pillow
 qwen-vl-utils
 torchvision
 torch
-spaces
+spaces
+peft