v0.0.4

What's Changed

• feat: init the first env by @lkevinzc in #2
• feat: add multi-turn api and stateful obs wrapper by @lkevinzc in #4
• feat: support reasoning_gym as single-turn environments by @lkevinzc in #6
• feat: python tool use by @anyasims in #8
• feat: textarena mastermind and minesweeper by @anyasims in #9
• feat: support vector env by @lkevinzc in #10
• fix: vec env reset & guess the number reward by @lkevinzc in #12
• feat: textarena wordle by @anyasims in #14
• feat: async vector env by @anyasims in #11
• fix: fix guess the number reward spike by @lkevinzc in #15
• fix mastermind by @vermouthdky in #18
• Remove distinction between single-step and multi-step envs and adding math_env by @anyasims in #17
• Wrappers etc by @anyasims in #23
• feat: support code environment with evaluation test by @lkevinzc in #25
• chore: add math12k dataset by @lkevinzc in #26
• fix: fix math verify error by @lkevinzc in #27
• feat: textarena refinements by @vermouthdky in #24
• feat: add code train data; refine async vec env by @lkevinzc in #28
• feat: add ta:FifteenPuzzle by @vermouthdky in #30
• feat: add QA env and support search tool by @cameron-chen in #31
• feat: add logic reasoning (ruletaker) environment by @lkevinzc in #32
• Search tool evaluation, single turn chat w/ tools and some QaEnv fixes by @cameron-chen in #34
• feat: add two qa environments (nq & hotpotqa); add search tool to wrapper factory by @cameron-chen in #35
• ref: improve search engine's perf; enhance qa env & search tool prompt by @cameron-chen in #37
• chore: clean py tool, use tool chat template by @lkevinzc in #41
• chore: clean search tool, sync the reward setting by @cameron-chen in #47
• feat: add oat examples by @lkevinzc in #49
• feat: add more textarena games and refinement by @vermouthdky in #48
• chore: fix imports, add math lvl3-4, better naming for game env by @lkevinzc in #50
• feat: integration with verl framework by @lkevinzc in #51
• feat: standardize python tool offline evaluation by @lkevinzc in #53
• doc: add the example of training on QA env with/without tool by @cameron-chen in #52
• feat: benchmark qa with search tool by @cameron-chen in #54
• chore: add license, prepare examples and package by @lkevinzc in #56
• refine online search server (ggl & serp) by @cameron-chen in #57
• doc: update readme by @cameron-chen in #55
• chore: update link by @lkevinzc in #58
• fix: readme typo by @lkevinzc in #59
• chore: refine example readme by @lkevinzc in #61
• chore: refine example readme by @lkevinzc in #62
• chore: refine example readme by @lkevinzc in #64
• chore: update readme and license by @lkevinzc in #66
• chore: add env_id property to vector_env by @vermouthdky in #67
• fix: randomness when seed is provided for .reset() by @lkevinzc in #68
• chore: fix random on reset and packaging by @lkevinzc in #69
• feat: support terminal environment via docker container by @lkevinzc in #72

New Contributors

• @lkevinzc made their first contribution in #2
• @anyasims made their first contribution in #8
• @vermouthdky made their first contribution in #18
• @cameron-chen made their first contribution in #31

Full Changelog: https://github.com/axon-rl/gem/commits/v0.0.4