Check Training/Inference values of each benchmark

[samueljackson92]

Each benchmark should produce metrics measuring quality (e.g. accuracy, loss, etc.) and time. From this we can compute time to metric etc. We need to collect the current state of these for each benchmark in the repo. Each of these should be collected into a documentation page.

• mnist
  • training
  • inference
• em_denoise
  • training
  • inference
• dms_structure
  • training
  • inference
• slstr_cloud
  • training
  • inference
• optical_damage
  • training
  • inference
• stemdl
  • training
  • inference
• hydronet
  • training
  • inference