Views
No views yet
1from peft import AutoPeftModelForCausalLM
2from transformers import AutoTokenizer
3
4model = AutoPeftModelForCausalLM.from_pretrained(
5 "atsuki-yamaguchi/Mistral-7B-v0.1-focus-sw"
6)
7tokenizer = AutoTokenizer.from_pretrained(
8 "atsuki-yamaguchi/Mistral-7B-v0.1-focus-sw"
9)
10
11# w/ GPU
12model = AutoPeftModelForCausalLM.from_pretrained(
13 "atsuki-yamaguchi/Mistral-7B-v0.1-focus-sw",
14 device_map="auto",
15 load_in_8bit=True,
16)@article{yamaguchi2024empirical,
title={An Empirical Study on Cross-lingual Vocabulary Adaptation for Efficient Generative {LLM} Inference},
author={Atsuki Yamaguchi and Aline Villavicencio and Nikolaos Aletras},
journal={ArXiv},
year={2024},
volume={abs/2402.10712},
url={https://arxiv.org/abs/2402.10712}
}