Models
A key Component
type in Superduper is Model
and its descendants.
The intended usage is that Model
wraps classical AI and machine learning models,
AI APIs, as well as important processing steps involved in building such models,
such as feature-computation.
See here for basic usage. This section gives detailed usage information as well as information about building your own model types.
Read more
📄️ Key methods of Model
All usage in superduper proceeds by changing or setting the attributes of a Component
📄️ Computing model outputs with listeners
Usage
📄️ Configuring models to ingest features from other models
There are two ways to connect models in Superduper:
📄️ Training models directly on your datastore
Model instances may be trained if a trainer is set on the Model when db.apply is called.
📄️ Evaluating models
See here.
📄️ LLMs
Superduper allows users to work with LLM services and models
📄️ Creating embeddings
Superduper supports a number of embedding models which may be used to create
📄️ Bring your own models
There are two ways to bring your own computations