summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorYurenHao0426 <Blackhao0426@gmail.com>2026-09-01 00:21:14 -0500
committerYurenHao0426 <Blackhao0426@gmail.com>2026-09-01 00:21:14 -0500
commit8d0f5efe86dfa5dcf6b4135ad76a3a6f93535ac2 (patch)
tree17611ca7360be64af7302e99d66b7e5933375525
parent0a090e3f597ee0994ea16f0675135a33bf051441 (diff)
prepare_data: hard-exit after DONE — HF streaming prefetch threads hang interpreter shutdown after breakHEADmain
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01GkgLsACEF6CCP7EUfA5fZe
-rw-r--r--scripts/prepare_data.py4
1 files changed, 3 insertions, 1 deletions
diff --git a/scripts/prepare_data.py b/scripts/prepare_data.py
index b7684b8..8036a47 100644
--- a/scripts/prepare_data.py
+++ b/scripts/prepare_data.py
@@ -35,4 +35,6 @@ for i, ex in enumerate(ds):
break
ftr.close(); fva.close()
os.rename(train_path + ".partial", train_path); os.rename(val_path + ".partial", val_path)
-print(f"DONE train {n_tr/1e6:.1f}M val {n_va/1e6:.1f}M -> {a.out}")
+print(f"DONE train {n_tr/1e6:.1f}M val {n_va/1e6:.1f}M -> {a.out}", flush=True)
+sys.stdout.flush(); sys.stderr.flush()
+os._exit(0) # hard-exit: HF streaming leaves non-daemon prefetch threads after `break` that hang interpreter shutdown