Hanashi a293d0a2c4 reliability: failure-stats + output-validation tooling; persistent output dir ...

- bench/failures.py: per-GPU success rates and error-class breakdown by config
- bench/validate_outputs.py: ffprobe every recorded output via /view
- finding: 24GB cards fail on multi-ref (TE vision OOM), not duration/size;
  100% of successful runs produce valid videos
- outputs now written to /workspace (survives pod resets)

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

2026-08-06 14:41:00 -04:00

..

BENCHMARKS.md

reliability: failure-stats + output-validation tooling; persistent output dir

2026-08-06 14:41:00 -04:00

CONCURRENCY.md

concurrency experiment measured: co-scheduling 13% slower than sequential — one worker per GPU confirmed

2026-08-06 14:27:32 -04:00

RESEARCH.md

docs: pad markdown table columns for raw-text readability

2026-08-06 00:38:51 -04:00

WEIGHTS.md

WEIGHTS.md: all 9 files persistent on 1TB volume, tmpfs scheme retired

2026-08-06 02:12:42 -04:00