SKU RLE-1306 · Sold by External

TextArena

Product specifications

SKURLE-1306
Data typeRL environments
Volume100+ games
Size on diskNot published by source
FormatPython (pip: textarena) + Gym-style API
Access modelPUBLIC LICENSE
PricingFree · open-source license
Quality score
LicenseMIT License
TextArena is an open-source collection of 100+ single-, two-, and multi-player text-based games for training and evaluating LLM agents on skills such as negotiation, theory of mind, deception, spatial reasoning, and long-term planning. It follows an OpenAI-Gym-style interface (reset/step/render) where the agent is a call function mapping a string observation to a string action, with bracketed action tokens per game. It provides online play and a TrueSkill-based leaderboard, enabling effectively unlimited self-play training data.