Search found 589 matches

by dkappe
Fri May 01, 2020 5:43 pm
Forum: Computer Chess Club: General Topics
Topic: New engine: a0lite
Replies: 38
Views: 9285

Re: New engine: a0lite

Fruit 2.1, once the Wunderkind of the chess engine world vs the pokey python script a0lite. [Event "?"] [Site "?"] [Date "2020.04.30"] [Round "2"] [White "Fruit-2.1"] [Black "A0Lite-Best"] [Result "0-1"] [ECO "E61"] [GameDuration "00:10:58"] [GameEndTime "2020-04-30T18:45:38.700 CDT"] [GameStartTime...
by dkappe
Thu Apr 30, 2020 6:35 pm
Forum: Computer Chess Club: General Topics
Topic: New engine: a0lite
Replies: 38
Views: 9285

Re: New engine: a0lite

I’ve polished up the branch batching-prune with little ender and badgyal 8 torchscript nets and a combination of the two switching at 12 pieces. Looks good so far. It’s less than 100 elo weaker (gtx 2070) than the latest Crafty on 1 CPU at 3+2, no TB. I think that’s about as good as it’s going to ge...
by dkappe
Mon Apr 20, 2020 6:25 pm
Forum: Computer Chess Club: General Topics
Topic: Lc0 soon to support chess960 ?
Replies: 9
Views: 1666

Re: Lc0 soon to support chess960 ?

Both MiniHuman and leela2200 on lichess play chess960 on lc0 0.24, so the capability has already been there.
by dkappe
Sat Apr 18, 2020 9:08 pm
Forum: Computer Chess Club: General Topics
Topic: New engine: a0lite
Replies: 38
Views: 9285

Re: New engine: a0lite

I think cutechess-cli gets confused by crafty. I’ve turned off the resigns adjudication. There’s a bit more of a gap. # PLAYER : RATING ERROR POINTS PLAYED (%) CFS(%) W D L D(%) 1 Crafty : 0 ---- 79.5 111 71.6 100 63 33 15 29.7 2 A0Lite-LE-Certainty : -165 62 28.5 100 28.5 52 15 27 58 27.0 3 A0Lite-...
by dkappe
Sat Apr 18, 2020 6:30 am
Forum: Computer Chess Club: General Topics
Topic: New engine: a0lite
Replies: 38
Views: 9285

Re: New engine: a0lite

I debugged the certainty propagation (essentially alpha/beta backup of terminals) in MCTS. Now that it's working, it seems to make a huge difference. No oops! checkmates or won endgames slipping away into a draw. This is with Bad Gyal 8 and Little Ender at 3+2. Will add tree reuse tomorrow. Stay saf...
by dkappe
Sun Apr 12, 2020 2:41 pm
Forum: Computer Chess Club: General Topics
Topic: New engine: a0lite
Replies: 38
Views: 9285

Re: New engine: a0lite

Using a endgame_net branch, I combined bad Gyal 8 and little Ender, switching at 16 pieces. (Need to update the badgyal package for this.) Getting closer. Next is basic certainty propagation. Will put the a0lite games up on lichess. TC 3 min + 2 sec # PLAYER : RATING ERROR POINTS PLAYED (%) CFS(%) W...
by dkappe
Thu Apr 09, 2020 5:52 pm
Forum: Computer Chess Club: General Topics
Topic: Which engines play the most human like at lower levels?
Replies: 15
Views: 2605

Re: Which engines play the most human like at lower levels?

Rather than jumping through hoops with an old lczero engine, simply try the small nets: 11258-16x2 distilled and tiny gyal (also 16x2). They can be used with the latest lc0 on cpu (or android). https://github.com/dkappe/leela-chess-weights/wiki/Distilled-Networks https://github.com/dkappe/leela-ches...
by dkappe
Thu Apr 02, 2020 12:14 pm
Forum: Computer Chess Club: General Topics
Topic: New engine: a0lite
Replies: 38
Views: 9285

Re: New engine: a0lite

Well, a description of the Bad Gyal nets can be found here: https://github.com/dkappe/leela-chess-weights/wiki/Bad-Gyal Bad Gyal data can be found here: https://github.com/dkappe/leela-chess-weights/wiki/Bad-Gyal-Data Training is a bit more challenging. Daniel Shawul’s Keras training code on github ...
by dkappe
Wed Apr 01, 2020 10:00 pm
Forum: Computer Chess Club: General Topics
Topic: New engine: a0lite
Replies: 38
Views: 9285

Re: New engine: a0lite

I added another branch, 'batching-prune,' that has smart pruning. Seems to make a fair bit of difference. [Event "?"] [Site "?"] [Date "2020.04.01"] [Round "5"] [White "A0Lite-Pruning"] [Black "A0Lite-Batch"] [Result "1-0"] [ECO "B06"] [GameDuration "00:02:33"] [GameEndTime "2020-04-01T16:49:34.165 ...
by dkappe
Wed Apr 01, 2020 1:41 pm
Forum: Computer Chess Club: General Topics
Topic: New engine: a0lite
Replies: 38
Views: 9285

Re: New engine: a0lite

I decided to test the virtual loss and tree reuse branch at 30 min + 5 sec against Crafty on 1 cpu. Using Noomen 3. # PLAYER : RATING ERROR POINTS PLAYED (%) CFS(%) W D L D(%) 1 Crafty : 0 ---- 26.5 40 66.2 100 14 25 1 62.5 2 A0Lite : -126 67 13.5 40 33.8 --- 1 25 14 62.5 Not quite there yet. :-)