TPU Training: No TPU devices were found.

[sooftware]

Thanks for great framework.
I tried to train with tpu (Google Cloud Platform Environment). I encounter error like this:

kaki_ai@kaki-ins:~/kopite-bot$ python3 train_blender.py
16:14:31 | Overriding opt["no_cuda"] to True (previously: False)
16:14:31 | Loading model with `--beam-block-full-context false`
16:14:31 | loading dictionary from /home/kaki_ai/ParlAI/data/models/blender/blender_90M/model.dict
16:14:31 | num words = 54944
16:14:32 | DEPRECATED: XLM should only be used for backwards compatibility, as it involves a less-stable layernorm operation.
16:14:33 | Total parameters: 87,508,992 (87,508,992 trainable)
16:14:33 | Loading existing model params from /home/kaki_ai/ParlAI/data/models/blender/blender_90M/model
Traceback (most recent call last):
  File "train_blender.py", line 47, in <module>
    val_dataloader=test_loader,
  File "/home/kaki_ai/kopite-bot/training/lightning_base.py", line 135, in fit
    accumulate_grad_batches=self.accumulate_grad_batches,
  File "/opt/conda/lib/python3.7/site-packages/pytorch_lightning/trainer/connectors/env_vars_connector.py", line 39, in insert_env_defaults
    return fn(self, **kwargs)
  File "/opt/conda/lib/python3.7/site-packages/pytorch_lightning/trainer/trainer.py", line 321, in __init__
    replace_sampler_ddp, deterministic, precision, amp_backend, amp_level, plugins
  File "/opt/conda/lib/python3.7/site-packages/pytorch_lightning/trainer/connectors/accelerator_connector.py", line 91, in __init__
    self.tpu_cores = device_parser.parse_tpu_cores(tpu_cores)
  File "/opt/conda/lib/python3.7/site-packages/pytorch_lightning/utilities/device_parser.py", line 113, in parse_tpu_cores
    raise MisconfigurationException('No TPU devices were found.')
pytorch_lightning.utilities.exceptions.MisconfigurationException: No TPU devices were found.

If you have any doubts, please help me. Thank you!

[akihironitta]

See #6778. (just for the record)