dm_nevis/dm_nevis/benchmarker at master · google-deepmind/dm_nevis · GitHub
Skip to content

Latest commit

 

History

History

Folders and files

README.md

Benchmarker

The benchmarker library is a platform-agnostic collection of modules for training and evaluating the performance of learners as they interact with long task streams.

This package defines the key APIs that learners and streams must satisfy to be evaluable using the Nevis evaluation protocol.

Streams

The most important stream can be found at streams/nevis_stream.py. This stream contains the 100+ dataset image classification stream reported in the paper.

Learners

The key learner interface can be found at learners/learner_interface.py. This is the interface that learners must satisfy in order to for the benchmarker to evaluate them.