MLFlow Logging Implementations

Hello all!
My group is using MLFlow for there experiment tracking and I am trying to figure out how to use MLFlow with Pytorch Lightning. Unfortunately the documentation is lacking.

I was wondering if someone could give the me implementation to log/save the following in MLFlow:

  1. A param of text: {key=“description”, value={“this model/experiment does the following:…”} (this is a single item)
  2. A param of a metric like training accuracy for every epoch (this is a multi-item)
  3. An artifact of the model definition/class (for inference later)
  4. An artifact of a checkpoint (for inference later)
  5. An artifact of the testing datasets labels y
  6. An artifact of the testing datasets prediction labels

Anyone have experience with this?

Hello, my apology for the late reply. We are slowly converging to deprecate this forum in favor of the GH build-in version… Could we kindly ask you to recreate your question there - Lightning Discussions