from transformers import AutoModelForCausalLM, AutoTokenizer
from peft import PeftModel
# Option 1: Load the complete model directly
model = AutoModelForCausalLM.from_pretrained("leonzc/llama400m-wikitext2-merged")
tokenizer = AutoTokenizer.from_pretrained("leonzc/llama400m-wikitext2-merged")
# Option 2: Load just the adapter with the base model
base_model = AutoModelForCausalLM.from_pretrained("data4elm/Llama-400M-12L")
tokenizer = AutoTokenizer.from_pretrained("data4elm/Llama-400M-12L")
model = PeftModel.from_pretrained(base_model, "leonzc/llama400m-wikitext2-dora-adapter")
# Example usage
input_text = "What is the capital of France?"
inputs = tokenizer(input_text, return_tensors="pt")
outputs = model.generate(inputs.input_ids, max_length=50)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))