Update app.py

app.py

@@ -1,87 +1,42 @@
-import gradio as gr
-from huggingface_hub import InferenceClient
-
-# Initialize the inference client with the model
-client = InferenceClient("HuggingFaceH4/zephyr-7b-beta")
-
-def classify_text(
-    text,
-    max_tokens,
-    temperature,
-    top_p
-):
-    # Define system message with instructions for the model
-    system_message = (
-        "You are tasked with labeling text data based on both emotion temperature and text type categories. "
-        "The final output must be a 13-character code that consists of the following structure:\n\n"
-        "   0 index : Emotion Temperature code\n"
-        "   1 index : Informative\n"
-        "   2 index : Hate\n"
-        "   3 index : Toxic\n"
-        "   4 index : Appreciation\n"
-        "   5 index : Constructive Criticism\n"
-        "   6 index : Genuine Questions\n"
-        "   7 index : Advice/Suggestions\n"
-        "   8 index : Requests\n"
-        "   9 index : Spam\n"
-        "   10 index : Off-Topic\n"
-        "   11 index : Engagement Boosters\n"
-        "   12 index : Neutral/General\n\n"
-        "Every index should have a number between 0-9. 0 means not applicable, 4 means normal, 9 means high. "
-        "Choose appropriate numbers to showcase how much each category is related to the text input.\n\n"
-        "Output Format:\n"
-        "Always return a 13-character code following this structure."
-    )
-
-    # Prepare the messages, starting with the system message
-    messages = [{"role": "system", "content": system_message}]
-
-    # Add the user message
-    messages.append({"role": "user", "content": text})
+# app.py
 
-    response = ""
-
-    # Stream the response from the model
-    for msg in client.chat_completion(
-        messages,
-        max_tokens=max_tokens,
-        stream=True,
-        temperature=temperature,
-        top_p=top_p,
-    ):
-        token = msg.choices[0].delta.content
-        response += token
-
-    # Validate the response format
-    if len(response) == 13 and all(c.isdigit() for c in response):
-        return response
-    else:
-        return "Invalid response format"
-
-# Create the Gradio interface
-demo = gr.Interface(
-    fn=classify_text,
-    inputs=[
-        gr.Textbox(
-            lines=5,
-            label="Input Comment",
-            placeholder="Enter a comment to classify",
-        ),
-        gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
-        gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
-        gr.Slider(
-            minimum=0.1,
-            maximum=1.0,
-            value=0.95,
-            step=0.05,
-            label="Top-p (nucleus sampling)",
-        ),
-    ],
-    outputs=gr.Textbox(
-        label="13-Character Code",
-        placeholder="The generated 13-character code will appear here",
-    ),
+import gradio as gr
+from transformers import AutoTokenizer, AutoModelForCausalLM
+import torch
+import numpy as np
+
+# Load the model and tokenizer
+model_name = "huggingface/zephyr-7b-beta"  # Replace this with the actual model path
+tokenizer = AutoTokenizer.from_pretrained(model_name)
+model = AutoModelForCausalLM.from_pretrained(model_name)
+
+def get_category_scores(text):
+    # Tokenize input text
+    inputs = tokenizer(text, return_tensors="pt")
+    
+    # Generate model outputs
+    with torch.no_grad():
+        outputs = model(**inputs)
+    
+    # Process model output to extract labels
+    # For demonstration, we'll mock the extraction with random values
+    # Replace this part with actual logic based on model output
+    
+    # Example random values for demonstration (these should be extracted from model)
+    scores = np.random.randint(0, 10, size=13)  # Replace with actual logic
+    
+    # Convert scores to 13-digit code
+    result_code = ''.join(map(str, scores))
+    return result_code
+
+# Define the Gradio interface
+iface = gr.Interface(
+    fn=get_category_scores,
+    inputs=gr.Textbox(lines=2, placeholder="Enter text here..."),
+    outputs="text",
+    title="Text Labeling",
+    description="Label text data based on emotion temperature and categories. Returns a 13-digit code."
 )
 
-if __name__ == "__main__":
-    demo.launch()
+# Launch the interface
+iface.launch()