import gradio as gr import torch from transformers import AutoTokenizer, AutoModelForCausalLM, BitsAndBytesConfig base_model_id = "mistralai/Mistral-7B-v0.1" bnb_config = BitsAndBytesConfig( load_in_4bit=True, bnb_4bit_use_double_quant=True, bnb_4bit_quant_type="nf4", bnb_4bit_compute_dtype=torch.bfloat16 ) # Load the fine-tuned model "Tonic/mistralmed" model = AutoModelForCausalLM.from_pretrained("Tonic/mistralmed", quantization_config=bnb_config) tokenizer = AutoTokenizer.from_pretrained("Tonic/mistralmed", trust_remote_code=True) tokenizer.pad_token = tokenizer.eos_token tokenizer.padding_side = 'left' class ChatBot: def __init__(self): self.history = [] def predict(self, input): new_user_input_ids = tokenizer.encode(input + tokenizer.eos_token, return_tensors="pt") flat_history = [item for sublist in self.history for item in sublist] flat_history_tensor = torch.tensor(flat_history).unsqueeze(dim=0) bot_input_ids = torch.cat([flat_history_tensor, new_user_input_ids], dim=-1) if self.history else new_user_input_ids chat_history_ids = model.generate(bot_input_ids, max_length=2000, pad_token_id=tokenizer.eos_token_id) self.history.append(chat_history_ids[:, bot_input_ids.shape[-1]:].tolist()[0]) response = tokenizer.decode(chat_history_ids[:, bot_input_ids.shape[-1]:][0], skip_special_tokens=True) return response bot = ChatBot() title = "👋🏻Welcome to Tonic's EZ Chat🚀" description = "You can use this Space to test out the current model (MistralMed) 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 = [["What is the boiling point of nitrogen"]] iface = gr.Interface( fn=bot.predict, title=title, description=description, examples=examples, inputs="text", outputs="text", theme="ParityError/Anime" ) iface.launch()