Ray Tune: Load a checkpoints and continue to train PPO

Question

I train a ppo in the following way, with my config, custom Env and action masking:

 config = (
        ppo.PPOConfig()
        .environment(env=CustomEnv))

tuner = tune.Tuner(
        "PPO",
        param_space=config,
        run_config=air.RunConfig(
            stop=stop,
            verbose=1,
            checkpoint_config=air.CheckpointConfig(
                num_to_keep=3, checkpoint_frequency=5),
            local_dir=localDir)
    )

    result = tuner.fit()

From the checkpoints how can I restart a training session? I need to train in multiple step (easier to harder).

I am able to restore checkpoint to make inferences

I tried to use restore and but it does not to work, the only way it's to use a Trainable?