Original article can be found here (source): Artificial Intelligence on Medium
In the upload model form, you are asked to specify both the model and version name. Both model and version names must start with a letter and contain only letters, numbers and underscores.
The model is the overall prediction solution you are trying to achieve, while the versions are the various iterations on that solution, which you may create using different techniques. For example, let’s say you are trying to predict what a baby’s weight will be given certain information about the pregnancy. You may create different versions of the “Babyweight” model by tuning hyperparameters, varying the dataset or modifying the inputs. When you create an app to host your model, you can easily switch between model versions depending on which iteration you want to deploy.
You can also provide descriptions for both the model and version to help yourself keep track of the different models and versions you create.
You are also asked to specify the framework and framework version as well as the python version you used to train the model.
Depending on which framework you select, different options will appear to upload your model files. mia currently supports the SavedModel format in Tensorflow and the model.bst format in XGBoost (support for other frameworks and formats are coming soon!). Follow these links to read about how to export model files for predictions:
Once you have completed all of the required fields and selected the appropriate model files, hit the UPLOAD button at the bottom of the form.
If your model has been successfully uploaded to mia, you will be redirected to the following page:
You will also see the model appear in the model summary table on your profile page.