[Bug] Wrong number of warmup steps in recipes with default Trainer.scheduler_after_epoch=True

[vanIvan]

Describe the bug

It seems like default number of warmup steps for lr scheduler in most recipes is wrong (or not?). It is set to 4000 for default NoamLR scheduler, however in Trainer, default value scheduler_after_epoch=True, and it will never get to highest lr value during training for 1000 epochs (also default value for num epochs in most configs).

To Reproduce

Just run:

python recipes/ljspeech/glow_tts/train_glowtts.py

Expected behavior

No response

Logs

No response

Environment

{
    "CUDA": {
        "GPU": [
            "GeForce RTX 3090",
            "GeForce RTX 3090"
        ],
        "available": true,
        "version": "11.1"
    },
    "Packages": {
        "PyTorch_debug": false,
        "PyTorch_version": "1.9.0+cu111",
        "TTS": "0.6.2",
        "numpy": "1.21.6"
    },
    "System": {
        "OS": "Linux",
        "architecture": [
            "64bit",
            "ELF"
        ],
        "processor": "x86_64",
        "python": "3.8.13",
        "version": "#170-Ubuntu SMP Mon Oct 18 11:38:05 UTC 2021"
    }
}

Additional context

No response

[erogol]

Can you check @loganhart420?

[erogol]

@loganhart420 ?

[loganhart02]

@vanIvan Can you try again on the latest version?

[stale[bot]]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions. You might also look our discussion channels.