Evaluation code, configurations and results accompanying the AutoNIMA paper.
AutoNIMA itself lives in a separate repository: neurostuff/autonima (docs, PyPI).
projects/ |
one directory per benchmark project: configurations, run outputs, per-project reports |
scripts/ |
the general evaluation machinery — comparison against the benchmark, stage metrics, cross-project aggregation |
paper/ |
the figures and numbers for the manuscript, and reproduce.sh, which rebuilds them |
tools/ |
corpus-construction and retrieval utilities |
reports/ |
the evaluation tables the figures are built from |
Start with paper/README.md: it documents how to rebuild
every display item, and what the repository does and does not contain.
MIT. See LICENSE.