linalg_zero/grpo/scripts/eval_linalg.py

from dotenv import load_dotenv

# For debugging purposes apply dotenv file
load_dotenv()

import asyncio

import art
import hydra
import torch
import wandb
from omegaconf import DictConfig, OmegaConf

from linalg_zero.grpo.run_rl import test
from linalg_zero.grpo.types import LinAlgPolicyConfig


@hydra.main(version_base=None, config_path="../../config/grpo/Qwen/Qwen2.5-3B/eval", config_name="linalgzero-sft.yaml")
def main(cfg: DictConfig) -> None:
    # Convert all configs to plain dicts
    init_config = OmegaConf.to_container(cfg.init, resolve=True)
    training_config = OmegaConf.to_container(cfg.training, resolve=True)
    run_config = OmegaConf.to_container(cfg.run, resolve=True)
    trainer_args = OmegaConf.to_container(cfg.trainer, resolve=True)
    engine_args = OmegaConf.to_container(cfg.engine, resolve=True)

    print(f"Evaluating model {cfg.run.base_model}")

    assert isinstance(init_config, dict), "Init config must be provided"
    assert isinstance(training_config, dict), "Training config must be provided"
    assert isinstance(run_config, dict), "Run config must be provided"
    assert isinstance(trainer_args, dict), "Trainer args must be provided"
    assert isinstance(engine_args, dict), "Engine args must be provided"

    # Set dynamic values
    if "tensor_parallel_size" not in engine_args:
        engine_args["tensor_parallel_size"] = torch.cuda.device_count()

    report_to = trainer_args.get("report_to") if isinstance(trainer_args, dict) else None
    if report_to:
        if isinstance(report_to, str):
            report_to = [report_to]
        if "wandb" in report_to and wandb.run is None:
            wandb.init(project=run_config["project"], name=run_config["project_id"], job_type="eval")

    # Build model and run training
    model = art.TrainableModel(
        name=run_config["project_id"],
        project=run_config["project"],
        base_model=run_config["base_model"],
        config=LinAlgPolicyConfig(
            training_config=training_config,
            run_config=run_config,
        ),
        _internal_config=art.dev.InternalModelConfig(
            init_args=init_config,
            engine_args=engine_args,
            trainer_args=trainer_args,
        ),
    )

    try:
        asyncio.run(test(model))
    finally:
        if wandb.run is not None:
            wandb.finish()


if __name__ == "__main__":
    main()