This is a streamlit app with the official Clarifai Python utilities. This repo includes higher level convencience classes, functions, and scripts to make using our API easier. This is built on top of the Clarifai Python gRPC Client.
- Try the Clarifai demo at: https://clarifai.com/demo
- Sign up for a free account at: https://clarifai.com/signup
- Read the documentation at: https://docs.clarifai.com/
This is to provide dashboard insights into a customer's usage of models within the Clarifai platform. In addition to basic information about the application where installed, it also provides a view at the total billable operations by type for the application. It also allows the user to explore model usage and determine how many predictions were made for each concept given a user-determined range of confidence levels. To that effect, Confidence Range Bucket Boundaries are defined in the app as Lower (0 - lower boundary threshold), middle (between lower and upper boundary thresholds), and upper (upper boundary threshold - 1) by using a two-point slider. The Detailed Data section allows the user to view and export the prediciton data for the examined model over the defined timeframe.
Currently there are some limitations:
- this needs to be configured by clarifai for a user or a plan for it to start collecting predictions from that user
- the analytics are not yet optimized so expect slow loading times if lots of predictions with a high sampling rate are stored
- this may not report properly yet for all combinations of collaborators, teammates and various model sharing situations in the platform. It should current work for the app owner calling their own models that live in one of their apps.
- this currently free while in beta testing but will be a paid feature in the future.