From 906d17e896329e4d2dd8d7f728e4d5a234afe0e4 Mon Sep 17 00:00:00 2001 From: YurenHao0426 Date: Thu, 30 Jul 2026 12:15:40 -0500 Subject: experiment: serialize crossover launch locks --- experiments/handoff_resnet_p1_to_transformer_p1.sh | 6 ++++++ 1 file changed, 6 insertions(+) (limited to 'experiments/handoff_resnet_p1_to_transformer_p1.sh') diff --git a/experiments/handoff_resnet_p1_to_transformer_p1.sh b/experiments/handoff_resnet_p1_to_transformer_p1.sh index c3424d0..7376bf3 100755 --- a/experiments/handoff_resnet_p1_to_transformer_p1.sh +++ b/experiments/handoff_resnet_p1_to_transformer_p1.sh @@ -113,6 +113,12 @@ fi mkdir -p "$transformer_root/results/transformer_crossover" tmux new-session -d -s transformer_p1_gpu5 \ "cd '$transformer_root' && env CUDA_VISIBLE_DEVICES=5 '$python_bin' experiments/transformer_crossover_grid.py --stage p1 --num-shards 2 --shard-index 0 > results/transformer_crossover/p1_gpu5.log 2>&1" +until "$python_bin" "$validator" \ + --path "$transformer_launch" \ + --kind transformer_p1_launch \ + --expected-commit "$expected_transformer_commit" >/dev/null 2>&1; do + sleep 1 +done tmux new-session -d -s transformer_p1_gpu7 \ "cd '$transformer_root' && env CUDA_VISIBLE_DEVICES=7 '$python_bin' experiments/transformer_crossover_grid.py --stage p1 --num-shards 2 --shard-index 1 > results/transformer_crossover/p1_gpu7.log 2>&1" tmux has-session -t transformer_p1_gpu5 -- cgit v1.2.3