allenai / ai2thor

An open-source platform for Visual AI.
http://ai2thor.allenai.org
Apache License 2.0
1.15k stars 215 forks source link

Benchmarking #1121

Closed AlvaroHG closed 1 year ago

AlvaroHG commented 1 year ago

Setup to do daily benchmarks.

Runs using benchmark configuration, like one under ai2thor/benchmarking/benchmark_test.json

Benchmark configs are stored in S3 and picked up by build server and run daily.

To add new benchmarks that are run daily run the task, with the right S3 credentials, invoke task add_daily_benchmark_config <config_filepath>, with json file matching schema ai2thor/benchmarking/benchmark_config_schema.json.

Also by coincidence, includes new Windows build compatibility.

AlvaroHG commented 1 year ago

Done, addressed all comments. Thanks :)

Lucaweihs commented 1 year ago

Looks good! One last change request: please run black autoformatting on the new benchmarking.py script:

pip install black
black ai2thor/benchmarking.py