Update app.py

app.py

@@ -2,13 +2,13 @@ import os
 from threading import Thread
 from typing import Iterator, List, Tuple
 import json
+import requests
 
 import gradio as gr
 import spaces
 import torch
 import transformers
 from transformers import AutoModelForCausalLM, AutoTokenizer, TextIteratorStreamer
-from flask import Flask, request, jsonify
 
 DESCRIPTION = """\
 # Zero GPU Model Comparison Arena
@@ -43,34 +43,23 @@ for model_id in MODEL_OPTIONS:
     if tokenizers[model_id].pad_token_id is None:
         tokenizers[model_id].pad_token_id = tokenizers[model_id].eos_token_id
 
-# Initialize Flask app
-app = Flask(__name__)
-
-@app.route('/log', methods=['POST'])
-def log_results():
-    data = request.json
-    # Here you can implement any additional processing or storage logic
-    print("Logged:", json.dumps(data, indent=2))
-    return jsonify({"status": "success"}), 200
-
-def prepare_input(model_id: str, message: str, chat_history: List[Tuple[str, str]]):
-    if "OpenHathi" in model_id:
-        # OpenHathi model doesn't use a specific chat template
-        full_prompt = message
-        for history_message in chat_history:
-            full_prompt = f"{history_message[0]}\n{history_message[1]}\n{full_prompt}"
-        return tokenizers[model_id](full_prompt, return_tensors="pt")
-    elif "Navarna" in model_id:
-        # Navarna model uses a chat template
-        conversation = []
-        for user, assistant in chat_history:
-            conversation.extend([
-                {"role": "user", "content": user},
-                {"role": "assistant", "content": assistant},
-            ])
-        conversation.append({"role": "user", "content": message})
-        prompt = tokenizers[model_id].apply_chat_template(conversation, tokenize=False, add_generation_prompt=True)
-        return tokenizers[model_id](prompt, return_tensors="pt")
+def log_comparison(model1_name: str, model2_name: str, question: str, answer1: str, answer2: str, winner: str = None):
+    log_data = {
+        "question": question,
+        "model1": {"name": model1_name, "answer": answer1},
+        "model2": {"name": model2_name, "answer": answer2},
+        "winner": winner
+    }
+    
+    # Send log data to remote server
+    try:
+        response = requests.post('http://144.24.151.32:5000/log', json=log_data, timeout=5)
+        if response.status_code == 200:
+            print("Successfully logged to server")
+        else:
+            print(f"Failed to log to server. Status code: {response.status_code}")
+    except requests.RequestException as e:
+        print(f"Error sending log to server: {e}")
 
 @spaces.GPU(duration=90)
 def generate(
@@ -135,24 +124,7 @@ def compare_models(
 
     return chat_history1, chat_history2, chat_history1, chat_history2
 
-def log_comparison(model1_name: str, model2_name: str, question: str, answer1: str, answer2: str, winner: str = None):
-    log_data = {
-        "question": question,
-        "model1": {"name": model1_name, "answer": answer1},
-        "model2": {"name": model2_name, "answer": answer2},
-        "winner": winner
-    }
-    
-    # Send log data to Flask server
-    import requests
-    try:
-        response = requests.post('http://144.24.151.32:5000/log', json=log_data)
-        if response.status_code == 200:
-            print("Successfully logged to server")
-        else:
-            print(f"Failed to log to server. Status code: {response.status_code}")
-    except requests.RequestException as e:
-        print(f"Error sending log to server: {e}")
+
 
 def vote_better(model1_name, model2_name, question, answer1, answer2, choice):
     winner = model1_name if choice == "Model 1" else model2_name
@@ -205,8 +177,7 @@ with gr.Blocks(css="style.css") as demo:
 
 if __name__ == "__main__":
     # Start Flask server in a separate thread
-    flask_thread = Thread(target=app.run, kwargs={"host": "0.0.0.0", "port": 5000})
-    flask_thread.start()
+
     
     # Start Gradio app with public link
-    demo.queue(max_size=10).launch(share=True)
+    demo.queue(max_size=3).launch(share=True)