Unexpected dataloader_idx_X added after log_dict.

[george-gca]

🐛 Bug

Trainer is adding dataloader_idx_X to metrics, but it is not supposed to since the name of my metrics are data1/x and data2/x. It currently displays as data1/x/dataloader_idx_0=3.46e-14, data2/x/dataloader_idx_1=3.49e-14 in progress bar.

To Reproduce

https://colab.research.google.com/drive/1L_0LOl7pUeL7bzh6uZiRa3-WXYQQwirN?usp=sharing

Expected behavior

It should appear in progress bar the information as data1/x=3.46e-14, data2/x=3.49e-14 and ModelCheckpoint should find data1/x to save based on it.

Environment

• CUDA:
  • GPU:
    • Tesla T4
  • available: True
  • version: 10.1
• Packages:
  • numpy: 1.19.5
  • pyTorch_debug: True
  • pyTorch_version: 1.7.0+cu101
  • pytorch-lightning: 1.2.0
  • tqdm: 4.41.1
• System:
  • OS: Linux
  • architecture:
    • 64bit
  • processor: x86_64
  • python: 3.6.9
  • version: Proposal for help #1 SMP Thu Jul 23 08:00:38 PDT 2020

Additional context

This was not happening before (I was using another version, but recently updated my pytorch lightning docker container), so I don't have working version information.

[george-gca]

Looking for it in older versions, I just realized this bug always existed, it was me that was not logging the metrics for the progress bar.