Skip to main content

Upload model

  • In order to start training, a crucial step is creating a model file and then uploading it.

  • If you want to learn how to create a compatible model file, have a look at our model guide.

Different ways in which a model file can be uploaded are mentioned below:

1. Uploading a Model without Weights

user.uploadModel('path/to/model/file')

A model file can be uploaded in 4 different ways:

  • Specifying the Model name with an extension (the model file being present in the current directory) Model Upload 1

  • Specifying the Model name without an extension (the model file being present in the current directory) Model Upload 2

  • Specifying the Model file Path with an extension (the model file being present in the current or different directory) Model Upload 3

  • Specifying the Model file Path without an extension (the model file being present in the current or different directory) Model Upload 4

2. Uploading a Model with Weights

user.uploadModel('path/to/model/file', weights=True)

There are certain pre-conditions to upload a model file with weights:

  • Please follow the file name convention: if the model file name is "model.py" or "model.zip" -> the weight file name should be "model_weights.pth" (pytorch) or "model_weights.pkl"(tensorflow).
  • Both should be present in the same directory.

A model file along the respective weights can be uploaded in 4 different ways.

  • Specifying Model Name with extension (model file and weights present in current directory) Model Upload 5

  • Specifying Model Name without extension (model file and weights present in current directory) Model Upload 6

  • Specifying Model File Path with extension (model file and weights present in different directory) Model Upload 7

  • Specifying Model File Path with extension (model file and weights present in different directory) Model Upload 8