CalDec v1: a fully open decision model for personal assistants, dataset and recipe included
No_Contract_8296 · reddit · 2026-10-02
First-time releaser kgrozdanovski open-sourced CalDec v1, a fully open-weights decision model for personal assistants (CalDec Laya and CalDec GLiNER variants), with the assistant-decisions dataset and full recipe. The experiment tested how far open models like Laya trail the cheap but strong closed model Jev — fine-tuning closed much of the gap, with some checkpoints beating Jev on LocalLLaMA/typed-decisions and an internal test set, though Jev remains the better all-rounder. Built for a real-time, fully-local Jarvis-style assistant for privacy and zero latency; the author has no professional ML background.
More from Research
- AI 'speech clock' predicts how fast you're ageing from your voice, featured in Nature — AnnaCiaunica · 2026-10-02
- ScholarCatalyst: a new benchmark testing whether AI can pick research problems like humans — PangWeiKoh · 2026-10-02
- Clarification as Supervision lands NeurIPS Oral: denser training signal via model interaction — iatitov · 2026-10-02
- Ego-Exo4D-HM: SMPL-H reconstructions for 523 hours of egocentric-exocentric video, open-sourced — geopavlakos · 2026-10-02
- Cohere Labs Talk: Making AI Math Reasoning Machine-Checkable with Lean — Cohere_Labs · 2026-10-02
- New paper dissects only task-relevant network parts, making mechanisms inspectable and editable at far lower cost — leedsharkey · 2026-10-02