Function trainDiffusionLora

trainDiffusionLora(
    model: RuntimeModel,
    pairs: DiffusionPair[],
    cfg?: DiffusionLoraConfig,
): { lora: TrainableLora; losses: number[] }
Train a LoRA adapter on DiffusionGemma with the denoising objective. Returns the TrainableLora (caller saves it via saveAdapter) + the loss history.
Parameters
- model: RuntimeModel
- pairs: DiffusionPair[]
- cfg: DiffusionLoraConfig = {}
Returns { lora: TrainableLora; losses: number[] }
- Defined in packages/training/src/diffusion-lora.ts:146