From f68f6b00f047d9412113d7c2627bd7f6e96f2d62 Mon Sep 17 00:00:00 2001 From: yurenh Date: Mon, 31 Aug 2026 18:33:06 -0500 Subject: env bootstrap (optional venv) + self-check in the ladder runner Co-Authored-By: Claude Fable 5 Claude-Session: https://claude.ai/code/session_01GkgLsACEF6CCP7EUfA5fZe --- scripts/run_ladder.sh | 10 +++++++++- 1 file changed, 9 insertions(+), 1 deletion(-) (limited to 'scripts/run_ladder.sh') diff --git a/scripts/run_ladder.sh b/scripts/run_ladder.sh index 65b43a9..17219bd 100755 --- a/scripts/run_ladder.sh +++ b/scripts/run_ladder.sh @@ -30,7 +30,15 @@ EXTRA="--set ${TOKENS:+tokens=$TOKENS} ${SET:-}" echo "== zbp-scaling ladder: sizes=[$SIZES] arms=[$ARMS] nproc=$NPROC data=$DATA out=$OUT ${TOKENS:+tokens=$TOKENS}" [ "$NPROC" -ge 1 ] || { echo "no GPUs detected; set NPROC"; exit 1; } -python -c "import zbp_scaling" 2>/dev/null || pip install -e . -q +# --- environment bootstrap: current env first; VENV=1 creates ./.venv (fresh nodes) --- +PY=${PY:-python3} +if [ "${VENV:-0}" = 1 ]; then + [ -d .venv ] || "$PY" -m venv .venv + . .venv/bin/activate +fi +python -c "import zbp_scaling" 2>/dev/null || pip install -e . -q || { + echo "pip install failed; on managed nodes run with VENV=1 (creates ./.venv) or install into your conda env"; exit 1; } +python scripts/env_check.py || exit 1 if [ ! -f "$DATA/train.bin" ]; then echo "== preparing data ($PREP_TOKENS tokens -> $DATA) — run this on the training node" -- cgit v1.2.3