RAG Retrieves. Fine-Tuning Forgets. HyperNetworks Inject - and Now We Have the Scaling Laws.

Researchers from Nace AI and Purdue University have introduced HyperNetwork-based knowledge injection as a third path for integrating proprietary data into LLMs, backed by the first systematic scaling laws. The approach trains a secondary network to generate LoRA adapters from facts, leaving the base model frozen, and achieves predictable scaling with out-of-distribution generalization. The team built MegaWikiQA, a dataset of 10M+ multi-hop QA pairs, and demonstrated that hypernetworks outperform RAG and fine-tuning in OOD scenarios.

There are two mainstream answers to "how do I get my LLM to reliably use proprietary data?" RAG retrieves relevant chunks at inference time. Fine-tuning bakes new knowledge into the weights. Both are widely deployed. Both have well-documented failure modes. A new paper from Nace AI and Purdue University proposes a third path backed by the first systematic scaling laws: HyperNetwork-based knowledge injection . Instead of modifying the LLM, you train a second network that generates LoRA adapters from a batch of facts. The base model never changes. LoRA reduces cost but shares the same structural problems. You still forget. A HyperNetwork is a secondary model conditioned on input facts. It generates LoRA adapters ΔW inserted into the target model at inference time. The base model is frozen. Always. Architecture: Fact batch enters a Transformer encoder no causal mask, Post-LayerNorm, RoPE - Mean pooling - Linear - ΔW LoRA rank 4, α=8 - adapters into upper half of target model layers. Only the HyperNetwork trains. The team built MegaWikiQA -- 10M+ multi-hop QA pairs from Wikidata5M 4.6M entities, 822 relations, 39 domains, 1-4 hops, explicit OOD splits . Hypernetworks from 167M to 2.8B parameters. Four findings: Most enterprise teams face this: proprietary knowledge that can't go into pretraining, too dynamic for static fine-tuning, too specific for general RAG. The hypernetwork approach: no base model modification, predictable scaling, OOD generalization to new entities and regulations, independently upgradable layers. RAG retrieves. Fine-tuning forgets. HyperNetworks inject -- learning to generate weight adaptations from facts, not memorizing facts themselves. The scaling laws make this a design principle: steeper OOD scaling exponents, gap widening at scale, smooth power laws you can engineer around. Paper: https://arxiv.org/abs/2607.19604 https://arxiv.org/abs/2607.19604 | Code and data: https://huggingface.co/collections/nace-ai/hypernetwork-datasets https://huggingface.co/collections/nace-ai/hypernetwork-datasets Originally published at vsebude.it