Upload app.py

app.py

@@ -8,7 +8,6 @@ import requests
 import logging
 import subprocess
 from llama_cpp import Llama  # Import Llama from llama_cpp
-import spacy
 
 # Configure logging
 logging.basicConfig(level=logging.INFO, format='%(asctime)s - %(levelname)s - %(message)s')
@@ -21,21 +20,9 @@ def install_requirements():
     except subprocess.CalledProcessError as e:
         logging.error(f"Failed to install requirements: {e}")
 
-# Function to download the spaCy model
-def download_spacy_model(model_name):
-    try:
-        subprocess.check_call([os.sys.executable, '-m', 'spacy', 'download', model_name])
-        logging.info(f"SpaCy model {model_name} downloaded successfully.")
-    except subprocess.CalledProcessError as e:
-        logging.error(f"Failed to download SpaCy model {model_name}: {e}")
-
 # Install requirements
 install_requirements()
 
-# Download the spaCy model if it doesn't exist
-if not spacy.util.is_package('en_core_web_lg'):
-    download_spacy_model('en_core_web_lg')
-
 # Create the directory if it doesn't exist
 local_dir = "models"
 os.makedirs(local_dir, exist_ok=True)
@@ -60,13 +47,7 @@ def download_model(repo_id, filename, save_path):
 
 # Download the model if it doesn't exist
 if not os.path.exists(model_path):
-    download_model("adeptusnull/llama3.2-1b-wizardml-vicuna-uncensored-finetune-test", filename, model_path)
-
-# Function to truncate context to fit within the model's context window
-def truncate_context(context, max_tokens):
-    words = context.split()
-    truncated_context = ' '.join(words[-max_tokens:])
-    return truncated_context
+    download_model("PurpleAILAB/Llama3.2-3B-uncensored-SQLi-Q4_K_M-GGUF", filename, model_path)
 
 def respond(
     message,
@@ -86,24 +67,14 @@ def respond(
     # Load the model with the maximum context length and control the maximum tokens in the response
     llm = Llama(
         model_path=model_path,
-        n_ctx=500,  # Set the maximum context length
-        max_tokens=500  # Control the maximum number of tokens generated in the response
+        n_ctx=5000,  # Set the maximum context length
+        max_tokens=512  # Control the maximum number of tokens generated in the response
     )
 
     agent = Agent(llm, db_path, system_prompt)
     user_id = str(uuid.uuid4())  # Generate a unique user ID for each session
 
-    try:
-        # Truncate the context to fit within the model's context window
-        max_context_tokens = 500  # Adjust this based on your model's context window
-        context = f"{system_prompt}\nUser: {message}\nAssistant: "
-        truncated_context = truncate_context(context, max_context_tokens)
-
-        response = agent.process_query(user_id, message, truncated_context)
-    except ValueError as e:
-        logging.error(f"Error during processing: {e}")
-        response = "Désolé, il y a eu une erreur lors du traitement de votre requête. Veuillez essayer à nouveau."
-
+    response = agent.process_query(user_id, message)
     return response
 
 """