README.md

metadata

language:
  - en
tags:
  - llama
  - peft
  - dora
  - lora
license: apache-2.0
base_model: YongganFu/Llama-400M-12L

dora_model

DoRA-finetuned Llama-400M model

Model Details

This model is a DoRA-finetuned version of YongganFu/Llama-400M-12L. The standalone adapter is available at lxaw/dora_model-adapter.

Usage

from transformers import AutoModelForCausalLM, AutoTokenizer
from peft import PeftModel
# Option 1: Load the complete model directly
model = AutoModelForCausalLM.from_pretrained("lxaw/dora_model")
tokenizer = AutoTokenizer.from_pretrained("lxaw/dora_model")
# Option 2: Load just the adapter with the base model
base_model = AutoModelForCausalLM.from_pretrained("YongganFu/Llama-400M-12L")
tokenizer = AutoTokenizer.from_pretrained("YongganFu/Llama-400M-12L")
model = PeftModel.from_pretrained(base_model, "lxaw/dora_model-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))