Hypernetwork-generated LoRA adapters for knowledge injection show power-law scaling in width, depth, target size, and per-example fact count, and steeper out-of-distribution scaling than LoRA or full fine-tuning on MegaWikiQA.
Why Does New Knowledge Create Messy Ripple Effects in LLMs? , booktitle =
1 Pith paper cite this work, alongside 2 external citations. Polarity classification is still indexing.
1
Pith paper citing it
2
external citations · OpenAlex
fields
cs.CL 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Scaling Laws for Hypernetwork-Based Knowledge Injection in Large Language Models
Hypernetwork-generated LoRA adapters for knowledge injection show power-law scaling in width, depth, target size, and per-example fact count, and steeper out-of-distribution scaling than LoRA or full fine-tuning on MegaWikiQA.