scripts/job_trip.sh

#!/bin/bash
# Trip Plan half of the reproduction: Claim 3 (main table) + Claim 5 (buffer ablation).
# Resumable: finished configs are pulled from the results repo and skipped.
set -uo pipefail
pip install -q "transformers==4.46.2" "huggingface_hub<1.0" "datasets<4" "accelerate" 2>&1 | tail -1
python -c "
from huggingface_hub import snapshot_download
snapshot_download('ashishk1331/ccd-repro-code', repo_type='dataset', local_dir='/work')"
cd /work && mkdir -p outputs
nvidia-smi --query-gpu=name,memory.total --format=csv

# Fail fast if this GPU has no compiled kernels for our torch build (e.g. torch
# 2.5.1 on Blackwell/sm_120). Without this, every config dies one-by-one and the
# job burns wall-clock reporting the same CUDA error N times.
python - <<'EOF' || exit 1
import torch, sys
if not torch.cuda.is_available():
    print("ABORT: no CUDA"); sys.exit(1)
name = torch.cuda.get_device_name(0)
cap  = torch.cuda.get_device_capability(0)
try:
    (torch.zeros(8, 8, device="cuda", dtype=torch.bfloat16) @
     torch.zeros(8, 8, device="cuda", dtype=torch.bfloat16)).cpu()
except Exception as e:
    print(f"ABORT: {name} sm_{cap[0]}{cap[1]} unusable with torch {torch.__version__}: {e}")
    sys.exit(1)
print(f"GPU OK: {name} sm_{cap[0]}{cap[1]} torch {torch.__version__}")
EOF


N_TRIP=${N_TRIP:-64}
N_ABL=${N_ABL:-40}

python - <<'EOF'
from huggingface_hub import HfApi, snapshot_download
import glob, shutil
api = HfApi(); api.create_repo('ashishk1331/ccd-repro-results', repo_type='dataset', exist_ok=True)
try:
    snapshot_download('ashishk1331/ccd-repro-results', repo_type='dataset', local_dir='/work/_prev')
    n = 0
    for f in glob.glob('/work/_prev/outputs/*.json'):
        shutil.copy(f, '/work/outputs/'); n += 1
    print(f'resumed {n} finished configs')
except Exception as e:
    print('no previous results:', e)
EOF

push () {
  python - <<'EOF' 2>&1 | tail -1 || true
from huggingface_hub import HfApi
HfApi().upload_folder(folder_path="outputs", path_in_repo="outputs",
                      repo_id="ashishk1331/ccd-repro-results", repo_type="dataset")
print("pushed")
EOF
}

run () {
  out="outputs/$1"; shift
  if [ -f "$out" ]; then echo "SKIP $out"; return; fi
  echo "=========== RUN $out : $* ==========="
  python scripts/run_eval.py "$@" --out "$out" || echo "!!!!! FAILED: $out"
  push
}

########## Claim 3 — Trip Plan main table
# paper: baseline 15.10 | CCD 16.93 (+1.83) | CCD-DS 19.01 (+3.91) @ 75.20 steps (3.48x)
run "c3_trip_baseline.json" --task trip --method baseline --limit $N_TRIP
run "c3_trip_ccd.json"      --task trip --method ccd      --limit $N_TRIP
run "c3_trip_ccd_ds.json"   --task trip --method ccd_ds   --limit $N_TRIP
# "repaired" CCD-DS: V raised to the value the reported 3.48x actually requires
# at d=3 (V >= 13.9 by the k <= V/(d+1) bound). Tests whether the IDEA delivers
# even though the STATED configuration (V=4) cannot.
run "c3_trip_ccd_ds_V16.json" --task trip --method ccd_ds --limit $N_TRIP --buffer-V 16
run "c3_trip_ccd_V16.json"    --task trip --method ccd    --limit $N_TRIP --buffer-V 16

########## Claim 5 — buffer ablation on the City=3 subset (paper: peak 70% at size 4)
run "c5_abl_baseline.json" --task trip --method baseline --limit $N_ABL --num-cities 3
# V axis at d=3  (maps the k ~ max(1, V/(d+1)) law and the accuracy trade-off)
for V in 1 2 4 8 16; do
  run "c5_abl_V${V}.json" --task trip --method ccd_ds --limit $N_ABL --num-cities 3 --buffer-V $V --history-d 3
done
# d axis at V=4
for d in 1 2 5; do
  run "c5_abl_d${d}.json" --task trip --method ccd_ds --limit $N_ABL --num-cities 3 --buffer-V 4 --history-d $d
done

echo "=================== TRIP DONE ==================="
python - <<'EOF'
import json, glob
for f in sorted(glob.glob("outputs/c3_*.json")) + sorted(glob.glob("outputs/c5_*.json")):
    r = json.load(open(f))
    print(f"{f.split('/')[-1]:28s} score={r['score']:6.2f} steps={r['mean_steps']:7.2f} "
          f"speedup={r['speedup_vs_uniform']:5.2f}x V={r['buffer_V']} d={r['history_d']} n={r['n_examples']}")
EOF
push