train.py

# /// script
# requires-python = ">=3.10"
# dependencies = [
#     "trl>=0.12.0",
#     "peft>=0.7.0",
#     "transformers>=4.36.0",
#     "accelerate>=0.24.0",
#     "trackio",
# ]
# ///

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

print("📦 Loading dataset...")
dataset = load_dataset("open-r1/codeforces-cots", "solutions_w_editorials", split="train")

print(f"✅ Dataset loaded: {len(dataset)} examples")

print("🔀 Creating train/eval split...")
dataset_split = dataset.train_test_split(test_size=0.05, seed=42)
train_dataset = dataset_split["train"].select_columns(["messages"])
eval_dataset = dataset_split["test"].select_columns(["messages"])

config = SFTConfig(
    output_dir="qwen3-0.6b-codeforces-cots",
    push_to_hub=True,
    hub_model_id="gengxin-zhang/qwen3-0.6b-codeforces-cots",
    hub_strategy="every_save",
    num_train_epochs=3,
    per_device_train_batch_size=4,
    gradient_accumulation_steps=4,
    learning_rate=2e-5,
    logging_steps=10,
    save_strategy="steps",
    save_steps=100,
    save_total_limit=2,
    eval_strategy="steps",
    eval_steps=100,
    warmup_steps=100,
    lr_scheduler_type="cosine",
    report_to="trackio",
    project="qwen3_codeforces",
    run_name="qwen3-0.6b-cots-sft",
    max_length=2048,
)

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

print("🎯 Initializing trainer...")
trainer = SFTTrainer(
    model="Qwen/Qwen3-0.6B",
    train_dataset=train_dataset,
    eval_dataset=eval_dataset,
    args=config,
    peft_config=peft_config,
)

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

print("💾 Pushing to Hub...")
trainer.push_to_hub()

trackio.finish()