PlayTrain: An Efficient Reinforcement Learning Framework for LLM-Generated Adaptable JavaScript Games
PlayTrain turns a single LLM-written JavaScript game file into a trainable RL environment.
The paper says the framework can run generated JS games inside a standard gym setup, letting people play the same environments that agents train on. Its examples include simple JS clones of Atari and ProcGen games, with pixel-based agents trained end-to-end at more than 1M decisions per second on one GPU node. The authors also use it to modify games for new test sets, procedural-generation rules, and altered dynamics. ArXiv · AI/CL/LG's note
The paper says the framework can run generated JS games inside a standard gym setup, letting people play the same environments that agents train on. Its examples include simple JS clones of Atari and ProcGen games, with pixel-based agents trained end-to-end at more than 1M decisions per second on one GPU node. The authors also use it to modify games for new test sets, procedural-generation rules, and altered dynamics. ArXiv · AI/CL/LG's note
score 5