leading Llama 3 interruption root causes; network 8.4%, software 12.9% — yet >90% effective training time
148 faulty GPU / 72 GPU HBM3 (paper prints 30.1% / 17.2%)observed
| 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. |
|---|---|
| Caveat | Table 5 is internally inconsistent: its 17 listed counts sum to 417 although §3.3.4 states 419 unexpected interruptions; printed percentages sum to 94.4%; and 148/419 = 35.3%, not the printed 30.1%. Treat counts as auditable; label percentages as paper-printed. Network Switch/Cable is 35 (8.4% printed); NIC is classified Host and NCCL watchdog timeouts Unknown. |
| As of | 2024-07 |
| Source | Meta AI, The Llama 3 Herd of Models, §3.3.4/Table 5 · §3.3.4, Table 5: 466 total interruptions, 47 planned, 419 unexpected; rows for faulty GPU, HBM3, software, network, and the remaining categories |
| Review | checking…review by 2026-12-22 · standard cadence |
| Recorded changes | last 2026-08-24 · 2 revisions tracked |
| Claim id | leading-llama-3-interruption-root-causes |
Where the guide uses it
← Full numbers register — every date-stamped figure in the guide, with revision history.