from transformers import AutoModelForCausalLM, AutoTokenizer
from peft import PeftModel

model_id = "Qwen/Qwen3-8B-Base"
lora_id = "ZihengZhang/WARDEN-Qwen3"

# Load tokenizer and base model
tokenizer = AutoTokenizer.from_pretrained(lora_id)
model = AutoModelForCausalLM.from_pretrained(model_id)

# Load LoRA adapter
model = PeftModel.from_pretrained(model, lora_id)

# Inference
inputs = tokenizer("Hello, how are you?", return_tensors="pt")
outputs = model.generate(**inputs, max_new_tokens=50)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))