Player-1发布：用合成经验做游戏智能体后训练框架
Player-1: Learning to Play from Synthetic Experience
一篇教游戏智能体超越演示数据的新方法，用16个合成环境做强化学习，MineDojo-300成功率从12.63%拉到38.33%，做agent训练的可以看看。
Player-1 是一个通过合成经验学习游戏操作的后训练框架，扩展了计算机使用的动作空间，加入相对指针移动、持续输入和复合动作。团队先用 6,044 条精选游戏轨迹做监督微调，再构建 Player-1-Gym——16 个合成 2D/3D 环境，用可编程验证的奖励和 PPO 做多环境强化学习。Player-1-9B 和 Player-1-35B-A3B 在 MineDojo-300 上分别比 SFT 策略提升 27.8 和 17.7 个百分点，其中 Player-1-9B 达到 38.33% 成功率，远高于最强基线的 12.63%。
Player-1: Learning to Play from Synthetic Experience
Computer-use agents have demonstrated strong capabilities in interpreting screens and executing keyboard and mouse actions. However, games, with their visually rich worlds and diverse mechanics, remain challenging and offer a demanding next step toward general digital control. Current game-agent learning pipelines often rely on large-scale demonstrations with limited coverage of dynamic game states, while scalable environments for interactive training remain difficult to obtain. We introduce Player-1, a post-training framework for learning game control beyond demonstrations through synthetic experience. We extend the computer-use action space with relative pointer motion, sustained inputs, and composite actions. For supervised initialization, we fine-tune computer-use models on 6,044 curated gameplay trajectories that pair recorded actions with task instructions and visual reasoning. We construct Player-1-Gym, sixteen synthetic 2D and 3D environments designed around control demands shared across games. Programmatically verifiable rewards and randomized observations and control dynamics support multi-environment reinforcement learning with PPO. The resulting Player-1-9B and Player-1-35B-A3B improve MineDojo-300 success over their SFT policies by 27.8 and 17.7 percentage points, respectively. Player-1-9B reaches 38.33% success compared with 12.63% for the strongest evaluated baseline, with further gains over SFT on ViZDoom-10 and GameWorld. Further analyses show that broader synthetic environment coverage improves transfer, while the use of composite actions that combine multiple keyboard and mouse inputs increases during RL. We demonstrate that learning from synthetic experience substantially improves game control beyond supervised initialization, with gains that generalize beyond the RL training suite.
