Spaces:
Sleeping
Sleeping
| import gradio as gr | |
| from transformers import pipeline, AutoModelForCausalLM, AutoTokenizer | |
| import os | |
| from bertopic import BERTopic | |
| from sklearn.feature_extraction.text import CountVectorizer | |
| import numpy as np | |
| from sentence_transformers import SentenceTransformer | |
| # Retrieve the token from environment variables | |
| huggingface_token = os.getenv('LLAMA_ACCES_TOKEN') | |
| # Use the token with from_pretrained | |
| tokenizer = AutoTokenizer.from_pretrained("meta-llama/Llama-2-7b-chat-hf", token=huggingface_token) | |
| model = AutoModelForCausalLM.from_pretrained("meta-llama/Llama-2-7b-chat-hf", token=huggingface_token) | |
| # Function to load bad words from a file | |
| def load_bad_words(filepath): | |
| with open(filepath, 'r', encoding='utf-8') as file: | |
| return [line.strip().lower() for line in file] | |
| # Load bad words list | |
| bad_words = load_bad_words('badwords.txt') # Adjust the path to your bad words file | |
| # List of topics for the dropdown | |
| topics_list = ['Aviation', 'Science', 'Education', 'Air Force Pilot', 'Space Exploration', 'Technology'] | |
| #Load BerTopic model | |
| sentence_model = SentenceTransformer("all-MiniLM-L6-v2") | |
| topic_model = BERTopic(embedding_model=sentence_model) | |
| def is_inappropriate_or_offtopic(message, selected_topics): | |
| # Function logic as before | |
| def suggest_questions(): | |
| predefined_prompt = "Hello! My name is Andrea. So you want to learn how to become an airforce pilot? Select a question and let's start your learning journey." | |
| questions = [ | |
| "What are the basic requirements to become an airforce pilot?", | |
| "How long does it take to train as an airforce pilot?", | |
| "Can you describe a day in the life of an airforce pilot?" | |
| ] | |
| return predefined_prompt, questions | |
| def generate_response(selected_question, selected_topics): | |
| # Function logic for generating a response based on the selected question | |
| # Example of generating a response based on the selected question | |
| response = "This is a placeholder response for the question: " + selected_question | |
| # In a real scenario, you'd integrate this with the Llama model to generate a detailed response | |
| return response | |
| def main(): | |
| with gr.Blocks() as demo: | |
| gr.Markdown("### Child-Safe Chatbot | BETA") | |
| gr.Markdown("This chatbot uses BERTopic to identify topics in your messages and ensures the conversation stays relevant and safe.") | |
| with gr.Row(): | |
| start_btn = gr.Button("Start Conversation") | |
| questions_output = gr.Radio(label="Select a question", choices=[]) | |
| submit_btn = gr.Button("Submit Question") | |
| response_output = gr.Textbox(label="Bot Response") | |
| def update_questions(): | |
| prompt, questions = suggest_questions() | |
| questions_output.update(choices=questions) | |
| return prompt | |
| start_btn.click( | |
| fn=update_questions, | |
| inputs=[], | |
| outputs=response_output | |
| ) | |
| submit_btn.click( | |
| fn=generate_response, | |
| inputs=[questions_output, topics_list], # Assuming topics_list is used to filter topics or for future use | |
| outputs=response_output | |
| ) | |
| demo.launch() | |
| if __name__ == "__main__": | |
| main() | |