Revision History
PEFT · 8 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 LoRA GPT-3 configuration (rank 4 on query/value, not rank 8), the Houlsby adapter parameter share (2.1% at size 64; 3.6% was per-task sizes), the PEFT library's launch method list, the claim that Alpaca and Vicuna used PEFT (they used full fine-tuning), MoRA's and LoRA+'s reported results, an impossible single-RTX-4090 70B QLoRA claim, Apple's LoRA adapter use, and stale multi-LoRA serving claims (TGI and LoRAX are archived; Together AI's serverless multi-LoRA is discontinued; no source for OpenAI) (plus 34 minor fixes)
Version 8 · Nov 4, 2026, 06:08 PM
- Improve internal links (2026-10-13)
Version 7 · Oct 13, 2026, 10:08 AM
- Added 6 contextual internal links
Version 6 · Jul 23, 2026, 03:04 PM