[HN Gopher] Text-to-LoRA: Hypernetwork that generates task-speci...
___________________________________________________________________
Text-to-LoRA: Hypernetwork that generates task-specific LLM
adapters (LoRAs)
Author : dvrp
Score : 91 points
Date : 2025-06-12 05:51 UTC (3 days ago)
(HTM) web link (github.com)
(TXT) w3m dump (github.com)
| vessenes wrote:
| Sounds like a good candidate for an mcp tool!
| gdiamos wrote:
| An alternative to prefix caching?
| phildini wrote:
| I got very briefly excited that this might be a new application
| layer on top of meshtastic.
| jph00 wrote:
| The paper link on that site doesn't work -- here's a working
| link:
|
| https://arxiv.org/abs/2506.06105
| npollock wrote:
| LoRA adapters modify the model's internal weights
| make3 wrote:
| not unless they're explicitly merged, which is not a
| requirement but a small speed only thing
| watkinss wrote:
| Interesting work to adapt LoRa adapters. Similar idea applied to
| VLMs: https://arxiv.org/abs/2412.16777
___________________________________________________________________
(page generated 2025-06-15 23:00 UTC)