Nace.AI Releases Drex, Open Decision Models Returning Probability Distributions Without Autoregressive Decoding
nace-ai · hf · 2026-11-06
Nace.AI's research team published "Towards Foundation Decision Models: Drex", proposing an open "decision model" paradigm that challenges the default of long chain-of-thought autoregressive reasoning.
The paper argues that LLM autoregressive reasoning, while accurate, is inconsistent in production, requires output parsing, and adds cost and latency. Drex instead takes a question plus user-supplied options and directly returns a probability distribution over them — no autoregressive decoding needed.
Training combines supervised fine-tuning, RLAF, and consistency losses to improve accuracy, robustness, and latency. Paper, code, and models are open on GitHub and Hugging Face.
More from Research
- Discovery Loop member launches $DISCODEV open AI materials-discovery loop — arkitus · 2026-11-06
- Northwestern turns ordinary smartwatches into hand and gesture sensors with TransfHAR and electrode-based system — realkaranahuja · 2026-11-06
- Three researchers solo-prompting Bing Chat for four weeks gets roasted as a study — Sauers_ · 2026-11-06
- New attention kernels meet real models: where interpretability overlaps systems research — QuentinAnthon15 · 2026-11-06
- MoE All-to-All Comms Eat 60% of Training Steps; Correlated Placement Cuts 58% of Dispatched Rows — QuentinAnthon15 · 2026-11-06
- CMU and Bosch's BestRand cuts benchmark compression high-error risk by 50.2% — CarnegieMellonU · 2026-11-06