机器之心报道


-
论文标题:Open-Reasoner-Zero: An Open Source Approach to Scaling Up Reinforcement Learning on the Base Model -
论文地址:https://github.com/Open-Reasoner-Zero/Open-Reasoner-Zero/blob/main/ORZ_paper.pdf -
项目地址:https://github.com/Open-Reasoner-Zero/Open-Reasoner-Zero -
Hugging Face:https://huggingface.co/Open-Reasoner-Zero










如图 6 所示,可以看到整个训练过程中响应长度持续增加,没有饱和迹象,类似于 DeepSeek-R1-Zero 中看到的行为。



©
(文:机器之心)