Push Model Image | Documentation

To upload a model image to ⚗️ Instill Model, follow these steps:

Install the latest version of the Python 📦 Instill SDK:

pip install instill-sdk

Then, log in to Docker on the Instill Core instance:

# for Instill Cloud
docker login api.instill.tech
# username: USER_ID
# password: API_TOKEN
# for a local Instill Core instance
docker login localhost:8080
# username: admin
# password: API_TOKEN

The USER_ID, MODEL_ID, VERSION_TAG and API_TOKEN correspond to the user namespace, the ID of the model, the version tag of the model and the API token the user has generated. To learn how to create and manage your API Tokens, refer to the API Token Management page.

Now you can upload the model to ⚗️ Instill Model:

# for Instill Cloud
instill push USER_ID/MODEL_ID -t VERSION_TAG -u api.instill.tech
# for a local Instill Core instance
instill push USER_ID/MODEL_ID -t VERSION_TAG -u localhost:8080

Once you see the following message, the model image has been successfully uploaded:

2024-05-23 23:05:03,484.484 INFO     [Instill Builder] USER_ID/MODEL_ID:VERSION_TAG  pushed
2024-05-23 23:05:03,485.485 INFO     [Instill Builder] Done

As the model image is uploaded, ⚗️ Instill Model will automatically allocate the necessary resources for your model and deploy it. Depending on the size of the model image, and the type of hardware specified for the model, it could take some time for the model to go online. In the console, you can select the Versions tab under your model namespace to see the current state of the model version. To learn more about the different states a model version can assume, please refer to the Model State page.

To perform an inference using your model, please see the Model Inference page.