share of parameters trained by a LoRA adapter vs full fine-tune (model-dependent)

Name: share of parameters trained by a LoRA adapter vs full fine-tune (model-dependent)
License: https://aidatacenterguide.com/terms#intellectual-property

~0.1%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.
As of	2021
Source	LoRA (Hu et al.) / 2026 PEFT practitioner guides · Table 2 and GPT-3 results; LoRA configurations train roughly 0.01%–0.12% of parameters depending on rank and adapted matrices
Review	checking…review by 2026-11-23 · standard cadence
Recorded changes	last 2026-07-21 · 2 revisions tracked
Claim id	share-of-parameters-trained-by-a-lora-adapter

Where the guide uses it

1.4 Post-Training, Fine-Tuning & RL: The Hybrid Middle

← Full numbers register — every date-stamped figure in the guide, with revision history.