app.py

import gradio as gr
from transformers import AutoTokenizer, AutoModelForCausalLM
from peft import PeftModel
import torch

# Base model and adapter paths
base_model_name = "unsloth/mistral-7b-instruct-v0.3-bnb-4bit"
adapter_model_name = "jimy26/Chatbot"

# Load tokenizer and base model
tokenizer = AutoTokenizer.from_pretrained(base_model_name)
base_model = AutoModelForCausalLM.from_pretrained(
    base_model_name,
    device_map="auto",
    torch_dtype=torch.float16
)

# Load LoRA adapter on top
model = PeftModel.from_pretrained(base_model, adapter_model_name)

# Chat function
def chat(prompt):
    inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
    outputs = model.generate(**inputs, max_new_tokens=100)
    return tokenizer.decode(outputs[0], skip_special_tokens=True)

# Gradio interface
gr.Interface(fn=chat, inputs="text", outputs="text").launch()