Revision History
Encoder · 10 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.
- Correction: Corrected the ViT-22B training-set size (about 4 billion images, not one billion), DistilBERT's size reduction (40 percent smaller and 60 percent faster, not 60 percent smaller), and the attribution of the first human-level SuperGLUE result to the original 1.5-billion-parameter DeBERTa rather than DeBERTa-v3 (plus 14 minor fixes)
Version 10 · Nov 5, 2026, 04:52 AM
- Improve internal links (2026-10-13)
Version 9 · Oct 13, 2026, 10:02 AM
- Added 6 contextual internal links
Version 8 · Jul 23, 2026, 03:01 PM