ADDINGMODELS.md

##Adding Models to models

In order to make ZazuML aware of a (model)[https://github.com/dataloop-ai/zoo] you have to append it to the models.json file.

Example key value in model.json

  "retinanet": {
    "task": "detection",
    "model_space": {
      "accuracy_rating": 8,
      "speed_rating": 2,
      "memory_rating": 4
    },
    "hp_search_space": [
      {
        "name": "learning_rate",
        "values": [
          5e-4,
          1e-5,
          5e-5
        ]
      },
      {
        "name": "anchor_scales",
        "values": [
          [1, 1.189207115002721, 1.4142135623730951],
          [1, 1.2599210498948732, 1.5874010519681994],
          [1, 1.5, 2.0]
        ]
      }
    ],
    "training_configs": {
      "epochs": 100,
      "depth": 50,
      "input_size": 608,
      "learning_rate": 1e-5,
      "anchor_scales": [1, 1.2599210498948732, 1.5874010519681994]
    }
  }

hp_search_space - is for defining hyper-parameters that will over-go optimization

training_configs - is where fixed hyper-parameters are defined

Which parameters will be frozen and which will be optimizable is a design decision and will be immutable once the model is pushed to the ZazuML model zoo.

model_space - is where you define the relative location of your model in a euclidean vector space

task - is the defining task of your model, currently you can choose from either classification, detection or instance segmentation

The json object key must match the model directory name exactly so that ZazuML knows what model to call upon, in our example the name of both will be "retinanet".