LLaMA-Factory-Mirror/scripts/cal_lr.py

78 lines
2.9 KiB
Python

# coding=utf-8
# Calculates the optimal learning rate for 7B/13B models using LLaMA's hyper-parameters.
# Usage: python cal_lr.py --model_name_or_path path_to_model --dataset alpaca_en --cutoff_len 1024 --batch_size 16
# Inspired by: https://github.com/imoneoi/openchat/blob/master/ochat/training_deepspeed/train.py
import math
from typing import Optional
import fire
import torch
from torch.utils.data import DataLoader
from tqdm import tqdm
from transformers import DataCollatorForLanguageModeling, DataCollatorForSeq2Seq
from llmtuner.data import get_dataset
from llmtuner.extras.constants import IGNORE_INDEX
from llmtuner.hparams import get_train_args
from llmtuner.model import load_model_and_tokenizer
BASE_LR = 3e-4 # 1.5e-4 for 30B-70B models
BASE_BS = 4_000_000 # from llama paper
def calculate_lr(
model_name_or_path: str,
batch_size: int, # total batch size, namely (batch size * gradient accumulation * world size)
stage: Optional[str] = "sft",
dataset: Optional[str] = "alpaca_en",
dataset_dir: Optional[str] = "data",
template: Optional[str] = "default",
cutoff_len: Optional[int] = 1024, # i.e. maximum input length during training
is_mistral: Optional[bool] = False, # mistral model uses a smaller learning rate,
):
model_args, data_args, training_args, finetuning_args, _ = get_train_args(
dict(
stage=stage,
model_name_or_path=model_name_or_path,
dataset=dataset,
dataset_dir=dataset_dir,
template=template,
cutoff_len=cutoff_len,
output_dir="dummy_dir",
overwrite_cache=True,
)
)
_, tokenizer = load_model_and_tokenizer(model_args, finetuning_args, is_trainable=False, add_valuehead=False)
trainset = get_dataset(tokenizer, model_args, data_args, training_args, stage=stage)
if stage == "pt":
data_collator = DataCollatorForLanguageModeling(tokenizer=tokenizer, mlm=False)
elif stage == "sft":
data_collator = DataCollatorForSeq2Seq(tokenizer=tokenizer, label_pad_token_id=IGNORE_INDEX)
else:
raise NotImplementedError
dataloader = DataLoader(
dataset=trainset, batch_size=batch_size, shuffle=True, collate_fn=data_collator, pin_memory=True
)
valid_tokens, total_tokens = 0, 0
for batch in tqdm(dataloader):
valid_tokens += torch.sum(batch["labels"] != IGNORE_INDEX).item()
total_tokens += torch.numel(batch["labels"])
batch_max_len = cutoff_len * batch_size # max tokens in a batch
valid_ratio = valid_tokens / total_tokens
batch_valid_len = batch_max_len * valid_ratio
lr = BASE_LR * math.sqrt(batch_valid_len / BASE_BS) # lr ~ sqrt(batch_size)
lr = lr / 6.0 if is_mistral else lr
print(
"Optimal learning rate is {:.2e} for valid ratio% {:.2f} and effective batch size {:.2f}".format(
lr, valid_ratio * 100, batch_valid_len
)
)
if __name__ == "__main__":
fire.Fire(calculate_lr)