Skip to content

My Models

In this section you can view and manage your models.

Models are added to this list in a few ways:

  • From Model Zoo

  • Training will automatically save last checkpoint

  • You can manually select checkpoint after training

  • Upload model manually

Models list

At "Neural networks" page you will see all models in your current workspace.

You can view the following fields:

  1. Creation date

  2. Model name

  3. Associated NN Architecture (Plugin)

  4. Storage - where model weights are stored: Supervisely Server or some Node from Cluster

Manage models

Click "three dots" icon to see available actions with model.

50%

 

  1. Goto task - redirects to task page and filter all tasks. So user will be able to get all necessary information about training: logs, charts, checkpoints, etc.

  2. Edit - change model name

  3. Upload to Supervisely - if model is stored on some Node from Cluster, user can upload this model to Supervisely Server. It allows to use this model on other computational Nodes. Read mode on NN lifecycle

  4. Archive

  5. Config - training configuration (it is useful to see all training parameters including classes model predicts).

Here is the example of entire training configuration for our custom YOLO V3.

{
  "settings": {
    "lr": 0.0001,
    "epochs": 15,
    "batch_size": {
      "train": 12
    },
    "input_size": {
      "width": 256,
      "height": 256
    },
    "bn_momentum": 0.07,
    "gpu_devices": [
      0,
      1,
      2,
      3
    ],
    "data_workers": {
      "train": 3
    },
    "dataset_tags": {
      "train": "train"
    },
    "subdivisions": {
      "train": 1
    },
    "print_every_iter": 10,
    "weights_init_type": "transfer_learning",
    "enable_augmentations": false
  },
  "out_classes": [
    {
      "color": "#82789c",
      "shape": "rectangle",
      "title": "kiwi_bbox"
    },
    {
      "color": "#57c0cb",
      "shape": "rectangle",
      "title": "lemon_bbox"
    }
  ],
  "class_title_to_idx": {
    "kiwi_bbox": 0,
    "lemon_bbox": 1
  }
}
  • settings - training parameters

  • out_classes - classes model predicts

  • class_title_to_idx - mapping from class title to class index

Download model

If model is stored on some Node from Cluster, you have to upload model to Supervisely Server (manage models). After that the "Download" button will be available. As a result user will download .tar archive with model weights.

Start training / inference

From this page you can start training or inference. Please, read corresponding chapters to get more information.