files / stage1v2_final /check_progress.sh
milkyroad's picture
Upload folder using huggingface_hub (part 3)
b83ee78 verified
Raw
History Blame Contribute Delete
1.16 kB
#!/bin/bash
# Check if stage1v2_final training is still running and report progress
RUN_PID=$(pgrep -f "run_all.sh" 2>/dev/null | head -1)
TRAIN_PID=$(pgrep -f "stage1v2_final/train_cls\|stage1v2_final/train_crt\|stage1v2_final/ensemble" 2>/dev/null | head -1)
if [ -z "$RUN_PID" ] && [ -z "$TRAIN_PID" ]; then
echo "TRAINING_COMPLETE"
echo "Backbones with final results:"
for bb in convnext swin mamba resnet; do
f="/home/kccitadmin/jupyterlab/sandbox/stage1v2_final/results/letterbox/${bb}/target3_${bb}_final.json"
if [ -f "$f" ]; then echo " ✓ $bb plain"; else echo " ✗ $bb plain"; fi
f="/home/kccitadmin/jupyterlab/sandbox/stage1v2_final/results/letterbox/${bb}/crt_${bb}_final.json"
if [ -f "$f" ]; then echo " ✓ $bb cRT"; else echo " ✗ $bb cRT"; fi
done
echo "Ensemble results:"
ls /home/kccitadmin/jupyterlab/sandbox/stage1v2_final/results/letterbox/ensemble_*.json 2>/dev/null || echo " none"
else
echo "RUNNING"
echo "run_all PID: ${RUN_PID:-none}, train PID: ${TRAIN_PID:-none}"
tail -10 /home/kccitadmin/jupyterlab/sandbox/stage1v2_final/stage1v2_run.log 2>/dev/null
fi