Usage

from peft import PeftModel
from transformers import LLaMATokenizer, LLaMAForCausalLM, GenerationConfig

tokenizer = LLaMATokenizer.from_pretrained("22h/open-cabrita3b")
model = LLaMAForCausalLM.from_pretrained(
    "22h/open-cabrita3b",
    load_in_8bit=True,
    device_map="auto",
)
model = PeftModel.from_pretrained(model, "jcfneto/lora-cabrita-tv-ptbr")