gemma-7b / example_fsdp.py
milandean's picture
Upload example_fsdp.py with huggingface_hub
8490370 verified
raw
history blame
1.74 kB
# Make sure to run the script with the following envs:
# PJRT_DEVICE=TPU XLA_USE_SPMD=1
import torch
import torch_xla
import torch_xla.core.xla_model as xm
from datasets import load_dataset
from peft import LoraConfig, get_peft_model
from transformers import AutoTokenizer, AutoModelForCausalLM, TrainingArguments
from trl import SFTTrainer
# Set up TPU device.
device = xm.xla_device()
model_id = "google/gemma-7b"
# Load the pretrained model and tokenizer.
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(model_id, torch_dtype=torch.bfloat16)
# Set up PEFT LoRA for fine-tuning.
lora_config = LoraConfig(
r=8,
target_modules=["k_proj", "v_proj"],
task_type="CAUSAL_LM",
)
# Load the dataset and format it for training.
data = load_dataset("Abirate/english_quotes", split="train")
max_seq_length = 1024
# Set up the FSDP config. To enable FSDP via SPMD, set xla_fsdp_v2 to True.
fsdp_config = {"fsdp_transformer_layer_cls_to_wrap": [
"GemmaDecoderLayer"
],
"xla": True,
"xla_fsdp_v2": True,
"xla_fsdp_grad_ckpt": True}
# Finally, set up the trainer and train the model.
trainer = SFTTrainer(
model=model,
train_dataset=data,
args=TrainingArguments(
per_device_train_batch_size=64, # This is actually the global batch size for SPMD.
num_train_epochs=100,
max_steps=-1,
output_dir="./output",
optim="adafactor",
logging_steps=1,
dataloader_drop_last = True, # Required for SPMD.
fsdp="full_shard",
fsdp_config=fsdp_config,
),
peft_config=lora_config,
dataset_text_field="quote",
max_seq_length=max_seq_length,
packing=True,
)
trainer.train()