Convex Markets / Datasets / RL environments

Dataset catalog

Filter by type, access, and pricing. Specs show before you open the product page.

35 results
t(
External · RLE-1206

tau-bench (τ-bench)

Tool-agent-user benchmark of 165 customer-service tasks (retail 115, airline 50) with policy-following and pass@k evaluation.

Type RL environmentsVolume 165 tasks (retail 115 + airline 50)Format Python package + JSON domain dataAccess PUBLIC LICENSE
Free
open-source license
View
W
External · RLE-1101

WebArena

Realistic self-hosted web environment: 812 long-horizon tasks over shopping, forum, GitLab, CMS and maps with execution-based evaluation.

Type RL environmentsVolume 812 tasks (from 241 intent templates)Format JSON task configs + Python (gym) + self-hosted Docker sitesAccess PUBLIC LICENSE
Free
open-source license
View
M
External · RLE-1105

MiniWoB++

Farama-maintained suite of 128 synthetic mini web-interaction tasks (button/form/date/inbox) with DOM+screenshot observations.

Type RL environmentsVolume 128 registered Gymnasium environments (docs: 'over 100 tasks')Format Python package (pip install miniwob-plusplus) + Gymnasium + Selenium/Chrome; HTML task pagesAccess PUBLIC LICENSE
Free
open-source license
View
S
External · RLE-1303

ScienceWorld

Text-based interactive environment covering elementary-science curriculum tasks (30 task types) with a dense reward in [0,1].

Type RL environmentsVolume 30 task typesFormat Python (pip: scienceworld) + Scala/JVM simulatorAccess PUBLIC LICENSE
Free
open-source license
View
W
External · RLE-1103

WebShop

Simulated e-commerce shopping environment: 1.18M real products and 12,087 instructions with a search/click action space and dense reward.

Type RL environmentsVolume 12087 crowd-sourced instructions (1.18M products; 500-task common test set)Format Python (Flask web app + gym text interface) + downloadable product data & search indexAccess PUBLIC LICENSE
Free
open-source license
View
C
External · RLE-1502

Crafter

Crafter is an open-world 2D survival reinforcement-learning environment where agents are evaluated by the 22 semantically meaningful achievements they can unlock in each procedurally generated episode.

Type RL environmentsVolume 22 achievements (evaluation tasks)Format Python package / OpenAI Gym environment; uint8 image observations, procedurally generated worldsAccess PUBLIC LICENSE
Free
open-source license (MIT)
View
M
External · RLE-1504

MiniHack

MiniHack is an open-source sandbox framework, built on the NetHack Learning Environment, for designing and running procedurally generated NetHack-based reinforcement-learning tasks (navigation and skill-acquisition), exposed as pre-registered Gymnasium environments.

Type RL environmentsVolume 159 pre-registered environments (Gymnasium env ids)Format Gymnasium (gym) RL environment Python package; per-step observations are a dict of NumPy arrays; levels defined in the NetHack des-file DSLAccess PUBLIC LICENSE
Free
open-source license (Apache-2.0)
View
A
External · RLE-1301

ALFWorld

Text-based embodied household RL environment aligning ALFRED tasks with TextWorld; 134 unseen eval games across 6 task types with an optional vision variant.

Type RL environmentsVolume 134 unseen eval gamesFormat Python (pip: alfworld) + PDDL/game filesAccess PUBLIC LICENSE
Free
open-source license
View
NL
External · RLE-1503

NetHack Learning Environment (NLE)

A Gymnasium reinforcement-learning environment that wraps the roguelike game NetHack 3.6.7, exposing symbolic game-state observations and a set of programmatically-rewarded goal tasks.

Type RL environmentsVolume 9 registered Gymnasium environments (base NetHack + 8 task variants)Format Gymnasium (Gym) reinforcement-learning environment (Python package with C/NetHack backend)Access PUBLIC LICENSE
Free
open-source license
View
P(
External · RLE-1407

PettingZoo (Farama)

The standard Python API and reference environment suite for multi-agent RL, covering Atari (multiplayer), Classic board/card games, MPE, SISL, and Butterfly.

Type RL environmentsVolume dozens reference environmentsFormat Python (pip: pettingzoo; AEC / Parallel multi-agent API)Access PUBLIC LICENSE
Free
open-source license
View
PB
External · RLE-1505

Procgen Benchmark

OpenAI's Procgen Benchmark is a suite of 16 procedurally-generated, Atari-like Gym/Gym3 reinforcement-learning environments built to measure sample efficiency and generalization in RL.

Type RL environmentsVolume 16 procedurally-generated game environments (each with up to 2^31 unique levels)Format Gym / gym3 RL environment installed as a Python package (pip install procgen); observations are (64,64,3) uint8 NumPy RGB arraysAccess PUBLIC LICENSE
Free
open-source software (MIT license)
View
B
External · RLE-1304

BabyAI

Grid-world instruction-following platform with a compositional synthetic 'Baby Language'; 19 levels of increasing difficulty, now part of Minigrid.

Type RL environmentsVolume 19 levelsFormat Python (Minigrid / Gym API)Access PUBLIC LICENSE
Free
open-source license
View
J
External · RLE-1508

Jericho

Open-source Python RL environment from Microsoft Research that connects agents to 57 human-made interactive-fiction text games via a modified Frotz Z-machine interpreter, with reward defined as in-game score deltas.

Type RL environmentsVolume 57 supported interactive-fiction games (Z-machine)Format Z-machine story files (.z3/.z5/.z6/.z8) played through the Python FrotzEnv wrapper; observations and actions are plain textAccess PUBLIC LICENSE
Free
open-source license
View
O
External · RLE-1507

Overcooked-AI

A two-agent gridworld cooking environment for benchmarking human-AI and multi-agent coordination, where agents cooperatively prepare and deliver soups for a shared sparse reward.

Type RL environmentsVolume 49 layout files (kitchen configurations; 5 are the canonical benchmark layouts)Format Python simulator (installable overcooked_ai_py package with Gym/PettingZoo-style envs) plus .layout config files (Python-dict text)Access PUBLIC LICENSE
Free
open-source license
View
T
External · RLE-1302

TextWorld

Microsoft's sandbox engine for procedurally generating and playing text-adventure games to train and evaluate RL agents; underlies ALFWorld.

Type RL environmentsVolume Unlimited procedurally-generated gamesFormat Python (pip: textworld) + generated game filesAccess PUBLIC LICENSE
Free
open-source license
View