from Hacker News

Text-to-LoRA: Hypernetwork that generates task-specific LLM adapters (LoRAs)

by dvrp on 6/12/25, 5:51 AM with 17 comments

• by phildini on 6/15/25, 8:19 PM

  I got very briefly excited that this might be a new application layer on top of meshtastic.

• by jph00 on 6/15/25, 9:44 PM

  The paper link on that site doesn't work -- here's a working link:

  https://arxiv.org/abs/2506.06105

• by smcleod on 6/16/25, 12:48 AM

  Out of interest, why does it depend on or at least recommend such an old version of Python? (3.10)

• by kixiQu on 6/17/25, 6:07 PM

  Can someone explain why this would be more effective than a system prompt? (Or just point me to it being tested out against that, I supposed)

• by watkinss on 6/15/25, 10:03 PM

  Interesting work to adapt LoRa adapters. Similar idea applied to VLMs: https://arxiv.org/abs/2412.16777

• by gdiamos on 6/15/25, 7:27 PM

  An alternative to prefix caching?

• by etaioinshrdlu on 6/16/25, 1:07 AM

  What is such a thing good for?

• by npollock on 6/15/25, 9:47 PM

  LoRA adapters modify the model's internal weights

• by vessenes on 6/15/25, 3:59 PM

  Sounds like a good candidate for an mcp tool!