Update app.py: remove model.cpython-38.pyc binary diff, add generate\_response function from model.py, update model paths, and modify chat route to use generate\_response

* Remove binary diff of model

Author: End3r6

__pycache__/model.cpython-38.pyc

@@ -2,24 +2,29 @@
 from transformers import AutoModelForCausalLM, AutoTokenizer
 from transformers import Conversation as Conversation
 from flask_cors import CORS
+from model import generate_response
 
 app = Flask(__name__)
 CORS(app)
 
 model_paths = {
-    'meena': 'google/meena-chatbot',
-    'dia_small': 'microsoft/DialoGPT-small',
     'dia_medium': 'microsoft/DialoGPT-medium',
-    'dia_large': 'microsoft/DialoGPT-large',
-    'blender_400m': 'facebook/blenderbot-400M',
-    'blender_90m': 'facebook/blenderbot-90M',
+    'dia_small': 'microsoft/DialoGPT-small',
+    'blender_90m': 'facebook/blenderbot-90M'
+
+    # 'dia_large': 'microsoft/DialoGPT-large',
+    # 'blender_400m': 'facebook/blenderbot-400M',
+    # 'octo': 'NexaAIDev/Octopus-v2',
+    # 'dia_base': 'microsoft/DialoGPT-base',
 }
 
 chat_model = None
 chat_tokenizer = None
 conversations = []
 model_name = None
 
+step = 0
+
 @app.route('/model', methods=['GET', 'POST'])
 def post_model():
     global chat_model, chat_tokenizer, model_name
@@ -33,6 +38,7 @@ def post_model():
 
     chat_model = AutoModelForCausalLM.from_pretrained(path)
     chat_tokenizer = AutoTokenizer.from_pretrained(path)
+    chat_tokenizer.pad_token = chat_tokenizer.eos_token
 
     # Add CORS headers to the response
     response_headers = {
@@ -45,31 +51,31 @@ def post_model():
 
 @app.route('/chat', methods=['GET', 'POST'])
 def chat():
-    global conversations, chat_model, chat_tokenizer
+    global step, tokenizer, last_output, conversations
+    
     data = request.get_json()
     message = data['message']
-    conversation_id = data['conversation_id']
-
-    if len(conversations) <= conversation_id:
-        conversations.append(Conversation())
-
-    conversations[conversation_id].add_user_input({'role': 'user', 'content': message})
-    inputs = chat_tokenizer(conversations[conversation_id].messages, return_tensors='pt')
-    reply = chat_model.generate(inputs.input_ids, max_length=50)
-
-    conversations[conversation_id].add_system_message({'role': 'system', 'content': reply})
-
+    # conversation_id = data['conversation_id']
+        
+    reply = generate_response(chat_model, chat_tokenizer, message, step, max_length=250)
+    
     # Add CORS headers to the response
     response_headers = {
         'Access-Control-Allow-Origin': '*',  # Change the '*' to the appropriate origin if needed
         'Access-Control-Allow-Headers': 'Content-Type',
         'Access-Control-Allow-Methods': 'POST'
     }
 
+    step += 1
+    
     return jsonify({'reply': reply}), 200, response_headers
 
 if __name__ == '__main__':
-    test = Conversation()
+    # Add CORS headers to the response
+    response_headers = {
+        'Access-Control-Allow-Origin': '*',  # Change the '*' to the appropriate origin if needed
+        'Access-Control-Allow-Headers': 'Content-Type',
+        'Access-Control-Allow-Methods': 'POST'
+    }
 
-    test.add_message({'':''})
-    app.run(port=5000)
+    app.run(port=8000)