Update app.py

app.py

@@ -1,47 +1,34 @@
 import gradio as gr
-import torch
-import torch.nn as nn
-import torch.nn.functional as F
 import os
-from transformers import DistilBertModel, DistilBertTokenizer
+import requests
 
-# Set Hugging Face API Key (Ensure secure storage of credentials)
-os.environ["HF_HOME"] = "HF_API_KEY"
+# Set your Hugging Face API key
+HF_API_KEY = "HF_API_KEY"
+API_URL = "https://api-inference.huggingface.co/models/mistralai/Mistral-7B-Instruct"
 
-# Check device availability
-device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
+headers = {"Authorization": f"Bearer {HF_API_KEY}"}
 
-# AI Tutor Model
-class SelfAttentionAI(nn.Module):
-    def __init__(self):
-        super(SelfAttentionAI, self).__init__()
-        self.bert = DistilBertModel.from_pretrained("distilbert-base-uncased")
-        self.fc = nn.Linear(768, 1)
-
-    def forward(self, input_ids, attention_mask):
-        outputs = self.bert(input_ids, attention_mask=attention_mask)
-        return self.fc(outputs.last_hidden_state[:, 0, :])
-
-# Load tokenizer and model
-tokenizer = DistilBertTokenizer.from_pretrained("distilbert-base-uncased")
-model = SelfAttentionAI().to(device)  # Move model to appropriate device
+def query_huggingface(payload):
+    response = requests.post(API_URL, headers=headers, json=payload)
+    return response.json()
 
 def ai_tutor(question):
-    inputs = tokenizer(question, return_tensors="pt", padding=True, truncation=True, max_length=512)
-    inputs = {key: val.to(device) for key, val in inputs.items()}  # Move inputs to device
-
-    with torch.no_grad():  # Disable gradient calculation for inference
-        response = model(inputs["input_ids"], inputs["attention_mask"])
-
-    # Convert tensor response into a readable format
-    confidence_score = torch.sigmoid(response).item()
-    response_text = "This is a high-confidence answer." if confidence_score > 0.5 else "This is a low-confidence answer."
+    payload = {
+        "inputs": question,
+        "parameters": {"max_new_tokens": 100, "temperature": 0.7}
+    }
     
-    return response_text
+    result = query_huggingface(payload)
+    
+    # Extract and format the response
+    if "error" in result:
+        return "Sorry, I couldn't process the request at the moment."
+    
+    return result[0]["generated_text"]
 
 # Gradio Chatbot UI
 with gr.Blocks() as demo:
-    gr.Markdown("## AI-Powered Mechanical Engineering Learning Chatbot")
+    gr.Markdown("## 🤖 AI-Powered Mechanical Engineering Learning Chatbot")
     chatbot = gr.Chatbot()
     input_text = gr.Textbox(label="Ask a Question")
     submit_btn = gr.Button("Submit")
@@ -49,12 +36,6 @@ with gr.Blocks() as demo:
     def respond(message, history):
         response = ai_tutor(message)
         history.append((message, response))
-
-        # Limit chat history to avoid excessive memory usage
-        max_history_length = 10
-        if len(history) > max_history_length:
-            history = history[-max_history_length:]
-
         return history, ""
 
     submit_btn.click(respond, inputs=[input_text, chatbot], outputs=[chatbot, input_text])