Editing Openai/693c0f4f-255c-8008-92e9-0cd44c6d6226 (section)

===== 1.1 Parameter-efficient finetuning and data-to-LoRA generation =====

Large language models (LLMs) are commonly adapted via LoRA: low-rank adapters ΔW=BA\Delta W = BAΔW=BA added to a frozen base model, trained on task data DDD. 2506.16406v1

Recent work like Drag-and-Drop LLMs (DnD) treats LoRA updates themselves as generative targets: a hyper-network takes a batch of unlabeled prompts from a dataset and directly outputs the LoRA weights for that dataset. This collapses the classical data→gradients→weights loop into a single forward pass and removes per-task finetuning.

DnD trains the hyper-network with MSE on tokenized LoRA weights between generated and reference adapters. 2506.16406v1

Limitations of current data-to-LoRA training:
* The target space (LoRA weights) is huge, so pure MSE is noisy and high-variance.
* Weight space is multi-modal: many different LoRAs can yield similar task behavior, but MSE to a single reference penalizes all but one.
* The supervision is purely parametric, not directly aligned with model behavior.

This makes training the generator harder, and probably limits generalization to unseen tasks.