File size: 2,496 Bytes
3c0443b |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 |
from datasets import load_dataset
from transformers import DataCollatorForLanguageModeling
from transformers import Trainer, TrainingArguments
import os
import torch
def main():
local_rank = int(os.environ['LOCAL_RANK'])
rank = int(os.environ['RANK'])
world_size = int(os.environ['WORLD_SIZE'])
torch.distributed.init_process_group("nccl")
print(f"Local Rank = {local_rank}/{world_size}")
# Load your JSONL file
dataset = load_dataset('json', data_files='../../data/m2_250514_1150.jsonl', split='train')
# Load a Model
from transformers import AutoTokenizer, AutoModelForCausalLM
model_name = "FacebookAI/roberta-base"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(model_name)
# Set pad token if not set
if tokenizer.pad_token is None:
tokenizer.pad_token = tokenizer.eos_token
# Tokenize the dataset
def tokenize_function(examples):
return tokenizer(examples["text"], truncation=True, max_length=512)
tokenized_dataset = dataset.map(tokenize_function, batched=True)
# Split the dataset into training and validation sets
split_dataset = tokenized_dataset.train_test_split(test_size=0.1)
# Data collator, pad the inputs to the maximum length in the batch
data_collator = DataCollatorForLanguageModeling(
tokenizer=tokenizer, mlm=False # mlm=False: causal language modeling
)
# Training
training_args = TrainingArguments(
output_dir="./results",
overwrite_output_dir=True,
num_train_epochs=3,
per_device_train_batch_size=4,
per_device_eval_batch_size=4,
dataloader_num_workers=8,
eval_steps=500,
save_steps=1000,
warmup_steps=500,
prediction_loss_only=True,
logging_dir="./logs",
logging_steps=100,
learning_rate=5e-5,
fp16=True, # true for GPU
)
trainer = Trainer(
model=model,
args=training_args,
train_dataset=split_dataset["train"],
eval_dataset=split_dataset["test"],
data_collator=data_collator,
)
# Start training
trainer.train()
torch.distributed.destroy_process_group()
# Save the model and tokenizer
model.save_pretrained("./fine_tuned_model")
tokenizer.save_pretrained("./fine_tuned_model")
if __name__ == "__main__":
main()
|