scripts/sft_qwen_uv.py

# /// script
# dependencies = [
#     "trl>=0.12.0",
#     "peft>=0.7.0",
#     "transformers>=4.36.0",
#     "accelerate>=0.24.0",
#     "datasets>=2.16.0",
#     "trackio",
# ]
# ///

import os
import trackio
from datasets import load_dataset
from peft import LoraConfig
from trl import SFTTrainer, SFTConfig


def main() -> None:
    base_model = "Qwen/Qwen2.5-0.5B"
    hub_model_id = os.environ.get("HUB_MODEL_ID", "davidsmts/qwen25-0_5b-sft-demo")
    project = os.environ.get("TRACKIO_PROJECT", "qwen25_sft_demo")
    run_name = os.environ.get("TRACKIO_RUN", "qwen25-0_5b-sft-lora")

    print("Loading dataset...")
    dataset = load_dataset("trl-lib/Capybara", split="train")
    print(f"Loaded {len(dataset)} examples")

    print("Creating train/eval split...")
    dataset_split = dataset.train_test_split(test_size=0.1, seed=42)
    train_ds = dataset_split["train"]
    eval_ds = dataset_split["test"]
    print(f"Train {len(train_ds)}, Eval {len(eval_ds)}")

    trackio.init(
        project=project,
        run_name=run_name,
        config={"model": base_model, "dataset": "trl-lib/Capybara"},
    )

    peft_config = LoraConfig(
        r=16,
        lora_alpha=32,
        lora_dropout=0.05,
        bias="none",
        task_type="CAUSAL_LM",
        target_modules=["q_proj", "v_proj"],
    )

    training_args = SFTConfig(
        output_dir="qwen25-0_5b-sft-demo",
        push_to_hub=True,
        hub_model_id=hub_model_id,
        hub_strategy="every_save",
        num_train_epochs=1,
        per_device_train_batch_size=4,
        gradient_accumulation_steps=4,
        learning_rate=2e-5,
        logging_steps=10,
        save_strategy="steps",
        save_steps=50,
        save_total_limit=2,
        eval_strategy="steps",
        eval_steps=50,
        warmup_ratio=0.1,
        lr_scheduler_type="cosine",
        gradient_checkpointing=True,
        fp16=True,
        report_to="trackio",
        project=project,
        run_name=run_name,
    )

    print("Initializing trainer...")
    trainer = SFTTrainer(
        model=base_model,
        args=training_args,
        train_dataset=train_ds,
        eval_dataset=eval_ds,
        peft_config=peft_config,
    )

    print("Starting training...")
    trainer.train()

    print("Pushing to Hub...")
    trainer.push_to_hub()
    print(f"Complete! Model available at https://huggingface.co/{hub_model_id}")


if __name__ == "__main__":
    main()