Mixtral_ether / Peft.train
jeduardogruiz's picture
Create Peft.train
92ee7ff verified
raw
history blame
917 Bytes
https://github.com/huggingface/peft
from transformers import AutoModelForSeq2SeqLM
from peft import get_peft_config, get_peft_model, LoraConfig, TaskType
model_name_or_path = "bigscience/mt0-large"
tokenizer_name_or_path = "bigscience/mt0-large"
peft_config = LoraConfig(
task_type=TaskType.SEQ_2_SEQ_LM, inference_mode=False, r=8, lora_alpha=32, lora_dropout=0.1
)
model = AutoModelForSeq2SeqLM.from_pretrained(model_name_or_path)
model = get_peft_model(model, peft_config)
model.print_trainable_parameters()
"trainable params: 2359296 || all params: 1231940608 || trainable%: 0.19151053100118282"
@Misc{peft,
title = {PEFT: State-of-the-art Parameter-Efficient Fine-Tuning methods},
author = {Sourab Mangrulkar and Sylvain Gugger and Lysandre Debut and Younes Belkada and Sayak Paul and Benjamin Bossan},
howpublished = {\url{https://github.com/huggingface/peft}},
year = {2022}
}