from transformers import AutoModelForCausalLM, AutoTokenizer import gradio as gr import torch # Create the ChatBot class class ChatBot: def __init__(self): # Initialize the history self.history = [] # Initialize the tokenizer with SentencePiece support self.tokenizer = AutoTokenizer.from_pretrained("Open-Orca/Mistral-7B-OpenOrca", use_fast=True) # Initialize the model self.model = AutoModelForCausalLM.from_pretrained("Open-Orca/Mistral-7B-OpenOrca") def predict(self, input): # Encode user input new_user_input_ids = self.tokenizer.encode(input + self.tokenizer.eos_token, return_tensors="pt") # Flatten the conversation history flat_history = [item for sublist in self.history for item in sublist] flat_history_tensor = torch.tensor(flat_history).unsqueeze(dim=0) # Create the input tensor for the model bot_input_ids = torch.cat([flat_history_tensor, new_user_input_ids], dim=-1) if self.history else new_user_input_ids # Generate a response from the model chat_history_ids = self.model.generate(bot_input_ids, max_length=2000, pad_token_id=self.tokenizer.eos_token_id) # Update the history with the generated response self.history.append(chat_history_ids[:, bot_input_ids.shape[-1]:].tolist()[0]) # Decode the response response = self.tokenizer.decode(chat_history_ids[:, bot_input_ids.shape[-1]:][0], skip_special_tokens=True) return response # Example usage chatbot = ChatBot() user_input = "Hello, chatbot!" response = chatbot.predict(user_input) print(response) title = "👋🏻Welcome to Tonic's EZ Chat🚀" description = "You can use this Space to test out the current model (DialoGPT-medium) or duplicate this Space and use it for any other model on 🤗HuggingFace. Join me on [Discord](https://discord.gg/fpEPNZGsbt) to build together." examples = [["How are you?"]] iface = gr.Interface( fn=bot.predict, title=title, description=description, examples=examples, inputs="text", outputs="text", theme="ParityError/Anime" ) iface.launch()