Intern-Decision: Efficient Multimodal Decision Making with Autoregressive Masked Language Modeling
We present Intern-Decision, a family of 0.8B, 2B, and 4B multimodal decision models that return structured choices and candidate probabilities from text and optional images.
ProofPaper ↗
Key points
- Interactive agents repeatedly choose among a finite set of actions, labels, or ordinal outcomes.
- Generating a free-form response for each choice adds latency and complicates the interface between a model and its environment.
- Our autoregressive masked language modeling objective places one decision placeholder per field in a complete assistant JSON skeleton.
- These results support compact decision models as a practical component of larger agent systems.
Sources (2)
- [1]Intern-Decision: Efficient Multimodal Decision Making with Autoregressive Masked Language ModelingHugging Face Daily Papers · Oct 15, 12:00 AM
We present Intern-Decision, a family of 0.8B, 2B, and 4B multimodal decision models that return structured choices and candidate probabilities from text and optional images.
Interactive agents repeatedly choose among a finite set of actions, labels, or ordinal outcomes.
- [2]Intern-Decision: Efficient Multimodal Decision Making with Autoregressive Masked Language ModelingarXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 15, 04:05 PM · same content
Extractive summary: sentences quoted from the sources.
Before this
- Oct 14, 2026ibm-granite/granite-switch-4.1-30b
- Oct 14, 2026I trained a 15M "decision-only" model two ways — linear recurrence vs Transformer attention. Accuracy was a coin flip; latency and training cost weren't.[R]
- Oct 14, 2026The Design Space of Tri-Modal Masked Diffusion Models
- Oct 14, 2026CoTranslator: Reverse to Forward Chain-of-Thought Translator for Reasoning Distillation
- Oct 13, 2026DecisionRobust: Benchmarking Robustness for JEV Style Decision Models Across Text and Vision
- Oct 13, 2026Recalibrating Many Questions with Few Labels: Minimax Rates and Identification Limits
