Use the open api to do the score

app.py

@@ -4,7 +4,13 @@ from threading import Thread
 import gradio as gr
 import spaces
 from PIL import Image
+from huggingface_hub import InferenceClient
+import os
 
+client = InferenceClient(
+    "openai/gpt-4o-mini",  # or other models like "openai/gpt-4"
+    token=os.environ["OPENAI_API_KEY"]  # Your OpenAI API key
+)
 
 # from zipnn import zipnn_hf
 
@@ -22,10 +28,10 @@ processor = AutoProcessor.from_pretrained(model_id)
 
 
 @spaces.GPU
-def score_it(input_img):
+def ocr(input_img):
     # Convert numpy array to PIL Image
     pil_image = Image.fromarray(input_img.astype('uint8'))
-    
+
     # Now use the PIL Image as before
     image = pil_image.convert("RGB").resize((224, 224))
 
@@ -34,13 +40,43 @@ def score_it(input_img):
 
     output = model.generate(**inputs, max_new_tokens=200)
     return processor.decode(output[0])
-    
+
+
+prompt = '''Review the following essay and score it. Output the score and the reasoning.The output format is:
+Score: *<score>*
+Reasoning: 1... 2... 3... 4... 5...
+'''
+
+def score_with_gpt(text):
+    messages = [
+        {"role": "system", "content":
+            "You are a high school English teacher,"
+            + "and you are grading a student's English essay. You are given a student's essay, and you need to score it based on the following criteria:"
+            + "1. The essay is well-organized and flows logically."
+            + "2. The essay is well-written and uses correct grammar and punctuation."
+            + "3. The words are spelled correctly."
+            + "4. The words and phrases are appropriate for the context."
+            + "5. Scale the score from 0 to 100."
+            },
+        {"role": "user", "content": text}
+    ]
+
+    return client.chat_completion(
+        messages=messages,
+        max_tokens=500
+    )
+
+
+def score_it(input_img):
+    text = ocr(input_img)
+    return score_with_gpt(text)
 
 
 demo = gr.Interface(fn=score_it, title="Upload your English script and get the score",
-                        inputs=[gr.Image()],
-                        outputs=['text'],
-                        stop_btn="Stop Generation",
-                        )
+                    inputs=[gr.Image()],
+                    outputs=[gr.Textbox(label="Text", lines=10), gr.Textbox(
+                        label="Score", lines=10)],
+                    stop_btn="Stop Generation",
+                    )
 
 demo.launch(debug=True)

requirements.txt

@@ -1,4 +1,5 @@
 torch==2.4.0
 transformers==4.46.2
 accelerate>=0.26.0
-zipnn==0.3.4
+zipnn==0.3.4
+huggingface_hub