[KongchangAI]
Concept智能体强化学习 / 面向智能体的强化学习

Agentic RL

A reinforcement learning paradigm designed for AI agents that must handle longer decision chains, sparser reward signals, and more complex state spaces compared to standard reinforcement learning settings. It represents an evolutionary direction beyond RLHF, addressing the specific challenges of training autonomous agents in open-ended environments.

Timeline (last 90 days)

Nov 14

Agentic RL is a subdirection of reinforcement learning whose training target is agents capable of calling tools, executing multi-step actions, and continuously interacting with the environment.

Unverified50%
Oct 20

Agima's training is divided into two stages: the first stage uses supervised fine-tuning (SFT) to unify the policy and the world model into a single omni-model; the second stage uses agentic reinforcement learning (RL) to teach the model to reason over imagined futures and use them for planning.

Unverified50%
Sep 17

Xiaomi is live-streaming in real time an agentic reinforcement learning training process that will cost more than one million dollars.

Expired50%

All Facts (2)

Source Articles