Revision History
Pruning · 12 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 12 · Nov 5, 2026, 10:56 AM
- Correction: Corrected NVIDIA 2:4 sparsity speedup claims (2x is Tensor Core math throughput, not end-to-end), reattributed the ResNet-50 2x/1.4% result to channel pruning, fixed the garbled VGG-16 FLOPs statistic, moved the ISCA top-5 distinction from Deep Compression to EIE, removed an unsupported claim of 2-29x production speedups at NVIDIA/Meta/Intel/Qualcomm, removed LLaMA2-7B figures not in the cited paper, repointed Tailored-LLaMA, AMD Llama 3.1 405B and Neural Magic BERT results to their real sources, and removed two unsourced rows (GPT-2, YOLOv4) from the benefits table (plus 36 minor fixes)
Version 11 · Nov 2, 2026, 03:39 PM
- Improve internal links (2026-10-13)
Version 9 · Oct 13, 2026, 10:00 AM
- Added 6 contextual internal links
Version 8 · Jul 23, 2026, 02:59 PM
- Render formulas with LaTeX math notation
Version 7 · Jul 11, 2026, 07:13 PM