llm
Download
24 lines 857 BTextModified 2026-08-31 raw
123456789101112131415161718192021222324
# Python bytecode cache from the driver/plotter.
__pycache__/
# Downloaded reference implementations (other languages), never tracked.
fetched/
# Compiled binaries and per-language build scratch.
.build/
# Fortran module files (gfortran emits these; -J.build redirects them, this is
# a safety net in case a build runs without it).
*.mod
# Tomo build artifacts emitted next to the tracked sources.
tomo/.tomo/
tomo/*
!tomo/*.tm
# Raw measured timings (machine-specific) and the vector graphs. The PNG
# graphs ARE checked in (a representative snapshot embedded in README.md);
# everything else here regenerates locally.
results.json
results.svg
results-*.svg
# Binary-size comparison: machine-specific data + vector graph are ignored;
# the sizes.png snapshot IS checked in (embedded in README.md), like results.png.
sizes.json
sizes.svg
size-breakdown.svg