/ml-server

Tooling for training environment on machine learning server.

Primary LanguagePython

Machine Learning Server Training tools

Tools

  • clip-server - Simple flask server with /predict endpoint returns the prediction from CLIP
  • tfjs-fileserver - Builtin python3 -m http.server for serving models converted to tfjs format for use in js(AdCognition)
  • models-fileserver - Builtin python3 -m http.server for serving models trained
  • control-panel - Simple frontend for controlling the mlserver hyperparams, logs, training, converting, testing the models
  • tensorboard - Tensorboard for a more detailed information about the training process

Ports to expose for each tool

  • clip-server - 5000
  • tfjs-fileserver - 5500
  • models-fileserver - 8080
  • control-panel - 8888
  • tensorboard - 6006