from Hacker News

Text-to-LoRA: Instant Transformer Adaption

by yurimo on 6/12/25, 6:09 AM with 1 comments

  • by magicalhippo on 6/12/25, 6:43 AM

    From the abstract:

    [We] introduce Text-to-LoRA (T2L), a model capable of adapting large language models (LLMs) on the fly solely based on a natural language description of the target task. T2L is a hypernetwork trained to construct LoRAs in a single inexpensive forward pass.

    Code repository here: https://github.com/SakanaAI/text-to-lora