API reference
Every endpoint, with its parameters, a Python and a curl example, and the response it returns.
Base URL https://api.kaimei.ai. Authenticate every request with Authorization: Bearer km_live_…; see Authentication. Requests and responses are JSON, lists page with limit and offset, and every error shares one envelope; see pagination and Errors. Python examples use a client constructed as client = Kaimei(), which reads KAIMEI_API_KEY from the environment.
Register, update, version and run models, and manage the reference data they are explained against.
Upload a model's layers and weights, read them back, replay a prediction through them and see what every layer computes.
Log the predictions your models make, or read back the ones Kaimei made, for explanation and monitoring.
Attributions for one input: why the model produced this output, and the smallest change that would flip it.
Background studies of a whole model that write insights: importance, behaviour, interactions, robustness, drift and rules.
The findings analyses write about a model, each with a summary, a severity and the numbers behind it.
Group models into projects and read the roll-ups across them: traffic, alerts and the inputs each model relies on.
Your workspace's plan, quotas and settings, and the cross-model overview behind the dashboard home page.
Create, list and revoke the credentials your servers and notebooks use.
Sign-up, sign-in and profile endpoints for dashboard users and for front ends built on the API.