Update main.py

main.py

@@ -10,8 +10,19 @@ app = Flask(__name__)
 
 # Load the DeepSeek-V3 model and tokenizer
 model_name = "deepseek-ai/DeepSeek-V3"
-tokenizer = AutoTokenizer.from_pretrained(model_name, trust_remote_code=True)
-model = AutoModelForCausalLM.from_pretrained(model_name, trust_remote_code=True)
+revision = "main"  # Pin to a specific revision if needed
+
+try:
+    tokenizer = AutoTokenizer.from_pretrained(model_name, trust_remote_code=True, revision=revision)
+    model = AutoModelForCausalLM.from_pretrained(
+        model_name,
+        trust_remote_code=True,
+        revision=revision,
+        quantization_config=None  # Disable quantization
+    )
+except Exception as e:
+    print(f"Error loading model: {e}")
+    raise
 
 # Supported languages for translation
 SUPPORTED_LANGUAGES = {
@@ -30,37 +41,24 @@ os.makedirs(AUDIO_DIR, exist_ok=True)
 
 # Handle user queries using DeepSeek-V3
 def handle_query(query):
-    # Tokenize the input query
     inputs = tokenizer(query, return_tensors="pt")
-
-    # Generate a response using the DeepSeek-V3 model
     outputs = model.generate(
         inputs.input_ids,
-        max_length=50,  # Adjust the max length as needed
+        max_length=50,
         num_return_sequences=1,
-        temperature=0.7,  # Adjust temperature for creativity
-        pad_token_id=tokenizer.eos_token_id  # Ensure proper padding
+        temperature=0.7,
+        pad_token_id=tokenizer.eos_token_id
     )
-
-    # Decode the generated response
     response = tokenizer.decode(outputs[0], skip_special_tokens=True)
     return response.strip()
 
 # Generate response and translate if needed
 def generate_response(query, language):
     try:
-        # Get response from DeepSeek-V3
         bot_response = handle_query(query)
-
-        # Translate response if the language is not English
         target_lang = SUPPORTED_LANGUAGES.get(language.lower(), "en")
         if target_lang != "en":
-            try:
-                bot_response = GoogleTranslator(source='en', target=target_lang).translate(bot_response)
-            except Exception as e:
-                print(f"Translation error: {e}")
-                bot_response = "Sorry, I couldn't translate the response."
-
+            bot_response = GoogleTranslator(source='en', target=target_lang).translate(bot_response)
         return bot_response
     except Exception as e:
         return f"Error fetching the response: {str(e)}"
@@ -68,14 +66,10 @@ def generate_response(query, language):
 # Convert text to speech using gTTS
 def text_to_speech(text, lang="en"):
     try:
-        # Generate a unique filename for the audio file
         audio_filename = f"{uuid.uuid4()}.mp3"
         audio_path = os.path.join(AUDIO_DIR, audio_filename)
-
-        # Create gTTS object and save the audio file
         tts = gTTS(text=text, lang=lang)
         tts.save(audio_path)
-
         return audio_path
     except Exception as e:
         print(f"Error generating speech: {e}")
@@ -96,10 +90,7 @@ def chat():
         if not user_message:
             return jsonify({"response": "Please say something!", "audio_url": None})
 
-        # Generate response
         bot_response = generate_response(user_message, language)
-
-        # Convert response to speech
         target_lang = SUPPORTED_LANGUAGES.get(language, "en")
         audio_path = text_to_speech(bot_response, lang=target_lang)