share of GPT-3 parameters trained in the original LoRA paper's two Table 4 configurations
~0.003–0.022%derived
| Value kind | derived — Derived values are guide calculations; their result is only as fixed as the stated inputs, scope, and method. |
|---|---|
| Scope | two original GPT-3 attention-adapter configurations; not a general LoRA percentage |
| As of | 2021-10-16 |
| Source | Hu et al., LoRA, Table 4; Guide percentage calculation · 4.7M and 37.7M trainable parameters versus 175,255.8M full-model parameters |
| Derivation | 100 × 4.7 / 175255.8 ≈ 0.00268%; 100 × 37.7 / 175255.8 ≈ 0.02151% |
| Review | checking…review by 2027-03-31 · standard cadence |
| Recorded changes | last 2026-10-05 · 3 revisions tracked |
| Claim id | share-of-parameters-trained-by-a-lora-adapter |
Where the guide uses it
← Full numbers register — every date-stamped figure in the guide, with revision history.