Hugging Face Introduces Decision Models That Output Probability Distributions Without Generating Text
ariG23498 · x · 2026-11-14
Hugging Face published a deep dive on "Decision Models": instead of prompting an LLM to generate JSON or text for classification/routing tasks, decision models directly evaluate candidate options and return a probability distribution — no text generation involved.
- Demoed on a customer-support ticket routing case (which team, urgency, refund request) via a new pipeline("decision", ...) API that takes a state plus questions with choice criteria.
- Example model: autotrust/GEV-26B-Decide, with runnable transformers code included.
- Positioned for any "pick an answer" workload where parsing generated text is fragile.
The post is part of the author's weekly roundup, which also covers token pooling-aware training, llama.cpp v1, TRELLIS fine-tuning, and the Inkling 2 Small release.
More from coding & agent
- Free open-source "AI Engineering from Scratch" course teaches algorithms from raw math — ghumare64 · 2026-11-14
- Inside the MI300X: 23-min visual guide from compute chiplets to a Gluon GEMM — saurabhtwq · 2026-11-14
- Job hunter's trick: fix a real issue in a startup repo, send the PR as evidence — minhnguyen · 2026-11-14
- Screenshot shows multi-agent, multi-user chat with realtime 3D video and long-term memory — nptacek · 2026-11-14
- DeepSeek opens its entire app stack via API: search, voice, OCR, ASR, TTS — churchkey · 2026-11-14
- Garry Tan's 52-task eval: Fable 5.2 matches Opus 5.5 on pass rate but costs 2.80x more — garrytan · 2026-11-14