unplanned interruption rate, Llama 3 405B (419 over 54 days, 16,384 H100s); >90% effective training time with automation
1 every ~3 hrobserved
| Value kind | observed — Reported measurements, counts, and specifications keep the precision and scope stated by their source; an exact specification is not treated as a range. |
|---|---|
| As of | 2024 |
| Source | Meta (Llama 3 Herd of Models) |
| Review | checking…review by 2026-09-05 · fast cadence |
| Recorded changes | last 2026-06-29 |
| Claim id | unplanned-interruption-rate-llama-3-405b-419 |
Where the guide uses it
- 9.4 Checkpointing for Large-Scale Training
- 14.3 Component Failure Modes, Failure Rates & Fleet Reliability Data
← Full numbers register — every date-stamped figure in the guide, with revision history.