{"full_name":"leela-zero/leela-zero","owner":"leela-zero","name":"leela-zero","shard":"1","url":"https://github.com/leela-zero/leela-zero","description":"Go engine with no human-provided knowledge, modeled after the AlphaGo Zero paper.","homepage":null,"topics":[],"tag_pages":[],"language":"C++","license":"GPL-3.0","avatar":"https://avatars.githubusercontent.com/u/46999504?v=4","created_at":"2017-10-24T18:19:43Z","pushed_at":"2024-05-02T18:00:51Z","metrics":{"stars":5594,"forks":1013,"watchers":283,"issues":2007,"prs":659,"releases":18,"issues_open":363,"issues_closed":1644,"prs_open":12,"prs_closed":207,"prs_merged":440,"discussions":0},"growth":{"day":{"stars":0,"forks":0,"base_day":"2026-11-15"}},"history":{"dates":["2026-11-10","2026-11-11","2026-11-12","2026-11-13","2026-11-14","2026-11-15","2026-11-16"],"stars":[5595,5593,5592,5592,5592,5594,5594],"forks":[1014,1012,1012,1012,1013,1013,1013],"watchers":[null,283,283,283,283,283,283],"issues_open":[null,363,363,363,363,363,363],"prs_open":[null,11,11,11,11,12,12]},"boards":{},"same_owner":[],"issues":[{"number":78,"title":"How to find Information about the new best network","comments":1992,"created_at":"2017-11-17T09:30:12Z"},{"number":591,"title":"Version 0.10 released - Next steps","comments":732,"created_at":"2018-01-07T20:20:41Z"},{"number":1311,"title":"Facebook open sources elf opengo ","comments":416,"created_at":"2018-05-02T19:08:16Z"},{"number":1030,"title":"Bigger, stronger, not faster (4dc12a8e etc)","comments":378,"created_at":"2018-03-13T13:10:53Z"},{"number":696,"title":"Tune first play urgency ?","comments":369,"created_at":"2018-01-20T14:00:10Z"},{"number":1681,"title":"40x256 training note.","comments":305,"created_at":"2018-08-02T12:20:59Z"},{"number":965,"title":"128x10 transition (6615567e)","comments":280,"created_at":"2018-03-02T15:30:20Z"},{"number":1599,"title":"Dynamical \"komi\" to keep winrate within a certain range","comments":215,"created_at":"2018-07-03T07:27:23Z"}],"discussions":[],"latest_release":{"name":"Leela Zero 0.17 + AutoGTP v18","tag":"v0.17","published_at":"2019-04-04T07:17:05Z","body":"* Implemented batching for GPUs, increasing speed.\r\n* Added support for Tensor Cores on NVIDIA cards.\r\n* Optimized Winograd code for CPU and GPU.\r\n* Improved UCT formula.\r\n* Improved root move selection by using Lower Confidence Bounds.\r\n* Improved passing heuristics (disabled in self-play).\r\n* Improved autotuner behavior in edge cases.\r\n* The training code now supports mixed precision fp32/fp16 training.\r\n* AutoGTP can now start from an arbitrary SGF position.\r\n* AutoGTP can now play matches with different white/black parameters.\r\n* AutoGTP can now send initial commands to the engine.\r\n* AutoGTP now provides the full engine options in the SGF comments.\r\n* AutoGTP now uses batching during self-play.\r\n* GTP: Add support for ```gomill-explain_last_move```.\r\n* GTP: ```lz-analyze``` and ```lz-genmove_analyze``` now support an ```avoid``` tag.\r\n* GTP: ```lz-analyze``` and ```lz-genmove_analyze``` now support a ```minmoves``` tag.\r\n* GTP: ```lz-analyze``` and ```lz-genmove_analyze``` now support a (post) ```interval``` tag.\r\n* GTP: ```loadsgf``` and ```printsgf``` can now save and load time control information.\r\n* Bugfix: ```lz-setoption name playouts``` now works correctly.\r\n* Bugfix: in analysis mode, provide output even if we did not have to search.\r\n* Bugfix: detect if the weights file is for the wrong board format.\r\n* Bugfix: bail out if OpenCL doesn't find a GPU (instead of using OpenCL-on-CPU).","is_prerelease":false},"readme":true,"readme_zh":false,"news":[],"stale":false,"snapshot_at":"2026-11-16T09:23:14Z","history_start":"2026-09-06"}