论文速读 | 25年10月

Mastering the game of Go with deep neural networks and tree search

• AlphaGo 2016
• 人类数据训练网络 —— 自我对弈强化学习 —— MCTS(PUCT)

Mastering the game of Go without human knowledge

• AlphaZero 2017
• 完全摒弃人类数据

SmolVLA: A Vision-Language-Action Model for Affordable and Efficient Robotics

• 2024
• VLM前一半中间层feature质量更好
• action expert为flow matching transformer

\(π_0\): A Vision-Language-Action Flow Model for General Robot Control

• RSS 2025

Action Tokenizer Matters in In-Context Imitation Learning

• IROS 2025
• action tokenizer for Multitask
• In-context Imiation learning

In-Context Imitation Learning via Next-Token Prediction

• ICRA 2025