Announcement
LFM2.5-24B harness adapter announced for Hermes agent
The adapter's makers report a 20% efficiency gain with no accuracy drop on held-out test tasks.
TLDR
The makers say they finished LoRA training an adapter for Nous Research's Hermes agent on Liquid AI's LFM2.5-24B model. They report a 20% efficiency gain with no accuracy drop on held-out test tasks. Harness Hub lists a collection of LFM2.5-24B harness adapters on Hugging Face.
Combined views
16.3K
3 Sources, first seen ago
