Revision History
Training · 9 revisions
Sizes are character counts of the article source. The signed number is the change from the previous revision.
Recent edit summaries
Detailed summaries recorded by editors. Generic maintenance summaries are omitted here; every recorded revision remains below. Dates describe the edit, not necessarily the event it covers.
- Update: fix version-number links (e.g. Claude 3.5 was linked as Claude 3 followed by .5)
Version 9 · Nov 5, 2026, 10:55 AM
- Correction: Corrected unsupported training-cost figures (2025 runs, Llama 3.1 405B, cost-per-compute trend), the Llama 3.1 405B GPU-hour total, DeepSeek V3 compute, the gradient-checkpointing overhead, AWS H100 per-GPU pricing, Epoch's language-model compute trend, and attributed GPT-4 hardware figures to an unconfirmed leak (plus 14 minor fixes)
Version 8 · Nov 2, 2026, 12:52 PM
- Improve internal links (2026-10-13)
Version 7 · Oct 13, 2026, 10:06 AM
- Added 6 contextual internal links
Version 6 · Jul 23, 2026, 03:03 PM