5.4.2.2. Submit Model

If you want to submit a trained AI model to be displayed and used in the repository, you have to fill the fields shown in Figure below. It is important you to share with us the expected inputs and outputs of the model to allow us to check if it works properly. You can do it in the Comments field.

Figure: Submit model area

• Model Name:

  A short name to your model. It is recommended the name reflects basic information about the model.

• Model Description:

  A detailed explanation of the model (General description, meaning of inputs and outputs, information about the training data, etc.).

• Model Environment:

  Information about the programming (OS, programming language, used packages and its versions, type of serialization, etc.). This is extremely important information, so be careful when filling this field.

• Expected Model Input:

  Describe the expected inputs of the model and their formats (simple parameters, .csv archives, .png archives, etc.).

• Expected Model Output:

  Describe the outputs of the model and their formats (JSON, .txt, etc.).

• Model Weights:

  Upload in this field your trained model (.h5, .pkl, etc.) and the scripts needed to run it. You have to upload all of these at once.

• Comments:

  Fill this field with an example of valid inputs and their corresponding outputs. If some (or all) input and outputs are files, upload them in the Model Weights field and let us know their names in this comments section. In this field, you can also share with us all additional information you think is important to a better AI model submission.

When you have submitted your model, the SINFONIA repository’s technical team will add it to the AI Inference’s area according to your specifications. This process is still not automatic, so some delay is expected.