dragon_lora: LoRA settings

In dragonfarm: Fine-Tune Small Language Models with LoRA from R

LoRA settings

Description

LoRA settings

Usage

dragon_lora(r = 16, alpha = 32, dropout = 0.05, target_modules = "auto")

Arguments

r	Rank of the adapter matrices. Higher learns more, costs more memory.
alpha	Scaling factor. A common rule is alpha = 2 * r.
dropout	Dropout applied to the adapter input.
target_modules	"auto" adapts every linear layer except the output head (peft's "all-linear"). Or a character vector of module names such as c("q_proj", "v_proj").

Value

A dragon_lora object.

Examples

dragon_lora(r = 8, alpha = 16)

dragonfarm documentation built on Oct. 12, 2026, 5:07 p.m.