This repository has been archived by the owner on May 6, 2024. It is now read-only.
v0.8.0
Installing NLE
Install with pip: pip install nle==0.8.0
.
Or pull the docker image: docker pull fairnle/nle:0.8.0
.
See README.md and/or docker/README.md for further instructions.
New in NLE v0.8.0
- Update tty tools (#283, @heiner)
- Fix NetHackScout environment following #179. (#277, @heiner)
- Alternative fix for gym 0.21 changes. (#271, @heiner)
- Follow https://apt.kitware.com in Dockerfiles (#266, @heiner)
- ttyplay: Don't read from stdin by default. Print help when called w/o… (#262, @heiner)
- Update plotting script to include xkey and ykey (#259, @dmadeka)
- Two misc changes: Fix error message and fix copyright (with check) (#252, @heiner)
- Run finalizer correctly even when _pynethack.Nethack creation fails. (#250, @heiner)
- Expose Inventory Size (#251, @dmadeka)
- Test and fix for "NLE with wrong HACKDIR after spawn" (#249, @heiner)
- Drop GIL on reset. (#242, @heiner)
- Disable dumplog files. (#240, @heiner)
- Close Nethack instance via finalizer. (#241, @heiner)
- Update "difficulty" comment to align with permonst.h. (#237, @heiner)
- Expose Monster Difficulty attribute of permonst struct through FFI (#236, @dmadeka)
- Add symdef Python class. (#231, @heiner)
- Use memfd_create + /proc/self/fd/{FD} for temporary DL on Linux. (#228, @heiner)
- Add wizard mode option to play.py, refactor rendering logic. (#226, @heiner)
- Fix black color handling (#225, @heiner)