Text-to-LoRA works: predicting LoRA distributions to scale inference by sampling weights
akyurekekin · x · 2026-10-10
Researcher Ramon Astudillo shares that he didn't believe text-to-LoRA worked—until experiments showed it does. Beyond that, the team demonstrates you can predict entire LoRA distributions given just the model input, enabling inference scaling by sampling models rather than tokens.
The original thread frames the core question: what if LLM inference could be scaled by sampling different model weights instead of more tokens? A single query suffices to generate useful LoRA updates, and predicting a distribution over them is key to making hypernets work.
More from Models
- Three criticisms of how a leaderboard treats Sol: version, timing metric, and API tuning — mgostIH · 2026-10-10
- Giffmana notices new models Argon and Astra share the same first letter and length — giffmana · 2026-10-10
- OpenAI says Codex usage data will never power its predictions, even after testing — btibor91 · 2026-10-10
- It's 2026 — Duke Libraries explainer on why LLMs still hallucinate, from guess-favoring benchmarks — ArtificialOther · 2026-10-10
- Dev's hands-on Opus 5.5 review: big coding leap, zero personality left — ryunuck · 2026-10-10
- Step 5 Preview, 600B MoE with 1M context, goes free and matches GPT-6 Luna — NousResearch · 2026-10-10