| Active : 4 Machines / 85 Threads / 69.33 MNPS | ||||||
|---|---|---|---|---|---|---|
| Priority 0 | ||||||
| Mhoupp | Stash | movepick_4ply_history | diff | 8.0+0.08 | LLR: 1.82 (-2.94, 2.94) [0.00, 5.00] Games: 45230 W: 11464 L: 11200 D: 22566 Ptnml(0-2): 544, 5362, 10579, 5546, 584 | Use 4-ply continuation history for scoring quiets |
| Jhonno | Berserk | feat/lmr-corr | diff | 60.0+0.60 | LLR: 0.98 (-2.94, 2.94) [0.00, 2.50] Games: 13806 W: 3276 L: 3186 D: 7344 Ptnml(0-2): 7, 1537, 3724, 1629, 6 | LMR: reduce less by |corrhist correction|, LMR_CORR=2048 |
| Vshche | Igel | nn-epoch999 | diff | 60.0+0.60 | LLR: 0.46 (-2.94, 2.94) [0.00, 2.00] Games: 4064 W: 1057 L: 1012 D: 1995 Ptnml(0-2): 1, 462, 1063, 503, 3 | LTC - rebased |
| Connor | Seer | main | diff | 40.0+0.40 | Elo: 47.72 +- 5.06 (95%) [N=20000] Games: 4938 W: 1573 L: 899 D: 2466 Ptnml(0-2): 5, 323, 1195, 885, 61 | 2.9.0 rc |
| Rosent | Winter | graph_history_hack * | diff | 8.0+0.08 | LLR: -0.97 (-2.94, 2.94) [-5.00, 0.00] Games: 13030 W: 3421 L: 3522 D: 6087 Ptnml(0-2): 254, 1607, 2881, 1532, 241 | Hack to handle graph history interactions related to the 50 move rule. This exists in other engines as well, such as Stockfish. Plan is to have this until I add 50 move rule to the static evaluation, at which point I will instead add 50 move rule to position hashing in a way that matches it in the eval. |
| Dseker | BlackMarlin | simplify-killer-lmr | diff | 40.0+0.40 | LLR: -1.63 (-2.94, 2.94) [-4.00, 0.00] Games: 28356 W: 6765 L: 6916 D: 14675 Ptnml(0-2): 59, 3327, 7557, 3176, 59 | Simplify killer LMR |
| Priority -1 | ||||||
| Dseker | BlackMarlin | quiet-bad-cap-lmr-2 | diff | 8.0+0.08 | LLR: 0.00 (-2.94, 2.94) [0.00, 4.00] Games: 0 W: 0 L: 0 D: 0 Ptnml(0-2): 0, 0, 0, 0, 0 | More LMR for quiet moves and bad captures excluding killers |
| Dseker | BlackMarlin | movegen-depth-see | diff | 8.0+0.08 | LLR: -0.39 (-2.94, 2.94) [0.00, 4.00] Games: 8842 W: 2206 L: 2197 D: 4439 Ptnml(0-2): 51, 1078, 2153, 1089, 50 | depth based see margin for good captures in move gen |
| Dseker | BlackMarlin | lmr-combined | diff | 8.0+0.08 | LLR: -1.53 (-2.94, 2.94) [0.00, 4.00] Games: 43980 W: 10993 L: 10928 D: 22059 Ptnml(0-2): 293, 5206, 10959, 5207, 325 | Reduce more when all 3 reduction conditions are met |
| Dseker | BlackMarlin | movegen-depth-capture-value | diff | 8.0+0.08 | LLR: -2.11 (-2.94, 2.94) [0.00, 4.00] Games: 13624 W: 3381 L: 3437 D: 6806 Ptnml(0-2): 94, 1592, 3487, 1554, 85 | Disregard the value of captured pieces in movegen at higher depths |
| Priority -5 | ||||||
| Terje | Weiss | Connectivity2 | diff | 60.0+0.60 | LLR: -0.21 (-2.94, 2.94) [0.00, 3.00] Games: 68704 W: 17741 L: 17594 D: 33369 Ptnml(0-2): 461, 8109, 17058, 8270, 454 | |
| Finished | ||||||
| Jhonno | Berserk | feat/lmr-corr | diff | 60.0+0.60 | LLR: -1.54 (-2.94, 2.94) [0.00, 2.50] Games: 3892 W: 881 L: 980 D: 2031 Ptnml(0-2): 4, 516, 1001, 425, 0 | LMR: reduce less by |corrhist correction| |
| Jhonno | Berserk | feat/lmr-corr | diff | 10.0+0.10 | LLR: 1.72 (-2.94, 2.94) [0.00, 2.00] Games: 73384 W: 18117 L: 17859 D: 37408 Ptnml(0-2): 127, 8764, 18669, 8988, 144 | LMR: reduce less by |corrhist correction| |
| Vshche | Igel | nn-epoch999 | diff | 60.0+0.60 | LLR: 0.71 (-2.94, 2.94) [0.00, 2.00] Games: 8390 W: 2186 L: 2114 D: 4090 Ptnml(0-2): 7, 940, 2225, 1020, 3 | LTC |
| Vshche | Igel | speedup2 | diff | 3.0+0.03 | LLR: 3.00 (-2.94, 2.94) [-3.00, 1.00] Games: 4374 W: 1257 L: 1123 D: 1994 Ptnml(0-2): 28, 452, 1099, 574, 34 | 3.8.6: speed up nnue and move ordering |
| Vshche | Igel | nn-epoch999 | diff | 90.0+0.90 | LLR: -0.08 (-2.94, 2.94) [0.00, 2.00] Games: 1538 W: 383 L: 388 D: 767 Ptnml(0-2): 0, 199, 377, 192, 1 | VLTC |
| Connor | Seer | qs0x2291e0ff.bin | diff | 40.0+0.40 | LLR: 2.98 (-2.94, 2.94) [-5.00, 0.00] Games: 20310 W: 4988 L: 4958 D: 10364 Ptnml(0-2): 37, 2297, 5452, 2337, 32 | |
| Vshche | Igel | nn-epoch999 | diff | 10.0+0.10 | LLR: 2.98 (-2.94, 2.94) [0.00, 2.00] Games: 34500 W: 9144 L: 8826 D: 16530 Ptnml(0-2): 105, 4005, 8718, 4311, 111 | STC |
| Vshche | Igel | nn-epoch679 | diff | 60.0+0.60 | LLR: -0.09 (-2.94, 2.94) [0.00, 2.00] Games: 10252 W: 2577 L: 2570 D: 5105 Ptnml(0-2): 8, 1186, 2729, 1197, 6 | |
| Jhonno | Berserk | 0047-pawnpairs-h512-41037-sorted.nn | diff | 60.0+0.60 | LLR: 2.91 (-2.25, 2.89) [0.00, 2.50] Games: 16654 W: 4050 L: 3823 D: 8781 Ptnml(0-2): 8, 1834, 4427, 2039, 19 | 0047: 0046 recipe + shard 41037 (pawn pairs, h512). Fixed nodes 40k: +4.17 +/- 2.59 (penta), 25,512 games. |
| Vshche | Igel | nn-epoch679 | diff | 10.0+0.10 | LLR: 2.95 (-2.94, 2.94) [0.00, 2.00] Games: 41850 W: 11070 L: 10745 D: 20035 Ptnml(0-2): 128, 4854, 10636, 5179, 128 | |
| Rosent | Winter | deeper_net * | diff | 40.0+0.40 | LLR: 3.00 (-2.94, 2.94) [0.00, 5.00] Games: 8314 W: 2153 L: 2001 D: 4160 Ptnml(0-2): 83, 931, 1971, 1095, 77 | Added correct weights file. |
| Vshche | Igel | speedup | diff | 3.0+0.03 | LLR: 3.08 (-2.94, 2.94) [-3.00, 1.00] Games: 5392 W: 1518 L: 1381 D: 2493 Ptnml(0-2): 33, 579, 1350, 686, 48 | 3.8.5: speed up nnue and prefetch |
| Jhonno | Berserk | exp/threats | diff | N=5000 * | Generated 13600002/13600000 Games Elo: 1.69 +- 0.17 (95%) [N=13600000] Games: 13600002 W: 5572274 L: 5506236 D: 2521492 | berserk-datagen round 3: 41017 settings, 0041 v 0038, 13.6M cap |
| Vshche | Igel | 5adc00b8 | diff | 90.0+0.90 | LLR: 2.95 (-2.94, 2.94) [0.00, 2.00] Games: 4004 W: 1119 L: 854 D: 2031 Ptnml(0-2): 2, 366, 1005, 623, 6 | VLTC |
| Vshche | Igel | 5adc00b8 | diff | 60.0+0.60 | LLR: 2.95 (-2.94, 2.94) [0.00, 2.00] Games: 4792 W: 1352 L: 1094 D: 2346 Ptnml(0-2): 4, 427, 1275, 687, 3 | LTC |
| Rosent | Winter | deeper_net * | diff | 8.0+0.08 | LLR: 2.99 (-2.94, 2.94) [-1.00, 4.00] Games: 4830 W: 1447 L: 1301 D: 2082 Ptnml(0-2): 86, 544, 1032, 644, 109 | Added correct weights file. |
| Vshche | Igel | 5adc00b8 | diff | 10.0+0.10 | LLR: 2.96 (-2.94, 2.94) [0.00, 2.00] Games: 5026 W: 1445 L: 1161 D: 2420 Ptnml(0-2): 18, 498, 1204, 768, 25 | STC |
| Rosent | Winter | deeper_net * | diff | 8.0+0.08 | LLR: 0.00 (-2.94, 2.94) [-1.00, 4.00] Games: 0 W: 0 L: 0 D: 0 Ptnml(0-2): 0, 0, 0, 0, 0 | Deeper net should have small N/s penalty, but should hopefully scale better. Testing against other network trained from scratch. |
| Connor | Seer | qs0x2291e0ff.bin | diff | 8.0+0.08 | LLR: -3.00 (-2.94, 2.94) [-3.00, 0.00] Games: 12916 W: 3098 L: 3315 D: 6503 Ptnml(0-2): 97, 1630, 3183, 1489, 59 | |
| Connor | Seer | misc-optimizations | diff | 40.0+0.40 | LLR: 2.96 (-2.94, 2.94) [0.00, 5.00] Games: 2650 W: 713 L: 607 D: 1330 Ptnml(0-2): 1, 233, 760, 321, 10 | test |
| Vshche | Igel | pawn_history4 | diff | 60.0+0.60 | LLR: 2.95 (-2.94, 2.94) [0.00, 2.00] Games: 35298 W: 8626 L: 8324 D: 18348 Ptnml(0-2): 15, 4051, 9224, 4335, 24 | 3.8.3: implement pawn history |
| Connor | Seer | misc-optimizations | diff | 8.0+0.08 | LLR: 2.97 (-2.94, 2.94) [0.00, 5.00] Games: 3338 W: 863 L: 745 D: 1730 Ptnml(0-2): 10, 324, 903, 402, 30 | test |
| Jhonno | Berserk | shared-corrhist-fixed | diff | 5.0+0.05 | LLR: 2.90 (-2.25, 2.89) [0.00, 2.00] Games: 23778 W: 5983 L: 5697 D: 12098 Ptnml(0-2): 23, 2711, 6172, 2923, 60 | SMP: share correction history across threads |
| Jhonno | Berserk | minor-corrhist | diff | 10.0+0.10 | LLR: -2.25 (-2.25, 2.89) [0.00, 2.00] Games: 9364 W: 2245 L: 2431 D: 4688 Ptnml(0-2): 21, 1220, 2392, 1022, 27 | Minor correction history |
| Jhonno | Berserk | major-corrhist | diff | 10.0+0.10 | LLR: -2.26 (-2.25, 2.89) [0.00, 2.00] Games: 8998 W: 2126 L: 2313 D: 4559 Ptnml(0-2): 24, 1165, 2305, 984, 21 | Major correction history |