zbp-scaling.git/scripts/prepare_data.py, branch main Unnamed repository; edit this file 'description' to name the repository. prepare_data: hard-exit after DONE — HF streaming prefetch threads hang interpreter shutdown after break 2026-09-01T05:21:14+00:00 YurenHao0426 Blackhao0426@gmail.com 2026-09-01T05:21:14+00:00 8d0f5efe86dfa5dcf6b4135ad76a3a6f93535ac2 Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01GkgLsACEF6CCP7EUfA5fZe 
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01GkgLsACEF6CCP7EUfA5fZe
prepare_data: write .partial then rename, so interrupted prep can't leave a half train.bin that gets mistaken for complete 2026-09-01T03:56:32+00:00 YurenHao0426 Blackhao0426@gmail.com 2026-09-01T03:56:32+00:00 0a090e3f597ee0994ea16f0675135a33bf051441 Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01GkgLsACEF6CCP7EUfA5fZe 
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01GkgLsACEF6CCP7EUfA5fZe
data prep (FineWeb-Edu->GPT2 BPE), rho diagnostics, tests, measured-cost notes 2026-08-31T23:16:31+00:00 yurenh blackhao0426@gmail.com 2026-08-31T23:16:31+00:00 17a81b9c86cfedd70812a0e83f33798b64c1678e Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01GkgLsACEF6CCP7EUfA5fZe 
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01GkgLsACEF6CCP7EUfA5fZe