Finished
KamatwDualiir_depthlimitdiff8.0+0.08LLR: 2.97 (-2.94, 2.94) [0.00, 5.00]
Games: 2782 W: 749 L: 617 D: 1416
Ptnml(0-2): 28, 284, 641, 404, 34
Limit iir depth
KamatwDuali32_depthdiff8.0+0.08LLR: -0.89 (-2.94, 2.94) [-8.00, 0.00]
Games: 7178 W: 1757 L: 1826 D: 3595
Ptnml(0-2): 87, 853, 1770, 800, 79
Change depth to a signed value (preparation for fractional and iir :p) Edit: Fuck it, i need to merge it either way
KamatwDualmaindiff40.0+0.40Elo: 140.81 +- 19.67 (95%) [N=500]
Games: 502 W: 244 L: 51 D: 207
Ptnml(0-2): 0, 9, 81, 120, 41
progtest v0.4.2
KamatwDualmaindiff8.0+0.08Elo: 125.39 +- 15.97 (95%) [N=1000]
Games: 1000 W: 499 L: 153 D: 348
Ptnml(0-2): 2, 41, 171, 181, 105
progtest
KamatwDuallmr_improvingdiff8.0+0.08LLR: -2.96 (-2.94, 2.94) [0.00, 5.00]
Games: 4352 W: 1038 L: 1143 D: 2171
Ptnml(0-2): 72, 562, 1003, 477, 62
lmr improving modifier
KamatwDualmaindiff8.0+0.08Tuning 31 Parameters
1712/25000 Iterations
27402/400000 Games Played
merged
KamatwDualtune2diff8.0+0.08LLR: 2.95 (-2.94, 2.94) [0.00, 5.00]
Games: 4508 W: 1175 L: 1034 D: 2299
Ptnml(0-2): 47, 510, 1025, 599, 73
Tune on less impactful values
KamatwDualmaindiff8.0+0.08Elo: 133.37 +- 15.93 (95%) [N=1000]
Games: 1008 W: 505 L: 136 D: 367
Ptnml(0-2): 4, 38, 148, 213, 101
Progtest vs 0.4.2
KamatwDuallower_wdldiff8.0+0.08LLR: 3.05 (-2.94, 2.94) [0.00, 8.00]
Games: 4830 W: 1412 L: 1290 D: 2128
Ptnml(0-2): 99, 539, 1038, 619, 120
Lower the wdl from 0.25 -> 0.75 to 0.25 -> 0.5 stop reason: Better priorities to test before eep
KamatwDualcosdecaydiff8.0+0.08LLR: 2.97 (-2.94, 2.94) [0.00, 5.00]
Games: 6044 W: 1748 L: 1592 D: 2704
Ptnml(0-2): 101, 678, 1328, 794, 121
Cosine decay LR
KamatwDual512diff8.0+0.08LLR: 2.96 (-2.94, 2.94) [0.00, 8.00]
Games: 712 W: 256 L: 164 D: 292
Ptnml(0-2): 8, 63, 143, 113, 29
512HL network
KamatwDualnewgen_net_2diff8.0+0.08LLR: 2.97 (-2.94, 2.94) [0.00, 5.00]
Games: 1504 W: 510 L: 365 D: 629
Ptnml(0-2): 20, 150, 304, 221, 57
Network from gen 2
KamatwDualtuning *diff8.0+0.08Tuning 47 Parameters
3406/25000 Iterations
54506/400000 Games Played
Continue with a different tuning with lessons learned from this one
KamatwDualtuningdiff8.0+0.08LLR: 3.05 (-2.94, 2.94) [0.00, 8.00]
Games: 3084 W: 842 L: 740 D: 1502
Ptnml(0-2): 48, 322, 700, 424, 48
3406 tune iterations https://openbench.tstawowy.pl/tune/48/
KamatwDualtuningdiff8.0+0.08Tuning 47 Parameters
12/25000 Iterations
192/400000 Games Played
KamatwDualtuning *diff8.0+0.08Tuning 43 Parameters
875/25000 Iterations
14002/400000 Games Played
KamatwDualtuningdiff8.0+0.08Tuning 43 Parameters
0/25000 Iterations
0/400000 Games Played
KamatwDualtuningdiff8.0+0.08Tuning 43 Parameters
0/25000 Iterations
0/400000 Games Played
KamatwDualrazoring_checkdiff8.0+0.08LLR: -2.96 (-2.94, 2.94) [0.00, 5.00]
Games: 3938 W: 894 L: 1001 D: 2043
Ptnml(0-2): 63, 527, 886, 440, 53
Don't do razoring in check
KamatwDualage_updiff8.0+0.08LLR: -0.26 (-2.94, 2.94) [0.00, 5.00]
Games: 19652 W: 4878 L: 4804 D: 9970
Ptnml(0-2): 289, 2321, 4545, 2369, 302
Age up visited entries
KamatwDualtt_replacementdiff8.0+0.08LLR: -2.95 (-2.94, 2.94) [0.00, 5.00]
Games: 4416 W: 1036 L: 1138 D: 2242
Ptnml(0-2): 71, 552, 1047, 484, 54
Strength limit to adding tt move
KamatwDualtt_replacementdiff8.0+0.08LLR: -2.96 (-2.94, 2.94) [0.00, 5.00]
Games: 8478 W: 2033 L: 2121 D: 4324
Ptnml(0-2): 126, 1095, 1900, 977, 141
Update tt replacement scheme (add bestmove when not present)
KamatwDualnopvnmpdiff8.0+0.08LLR: -1.05 (-2.94, 2.94) [0.00, 8.00]
Games: 5740 W: 1630 L: 1617 D: 2493
Ptnml(0-2): 114, 664, 1309, 661, 122
Don't do nmp on PV nodes Stop reason: yolo
KamatwDualtwo_killers *diff8.0+0.08LLR: -0.85 (-2.94, 2.94) [0.00, 5.00]
Games: 22512 W: 5838 L: 5775 D: 10899
Ptnml(0-2): 370, 2727, 4999, 2790, 370
Two killer moves (fix overwrite issue)
KamatwDualCM_history *diff8.0+0.08LLR: -1.37 (-2.94, 2.94) [0.00, 5.00]
Games: 11630 W: 2931 L: 2938 D: 5761
Ptnml(0-2): 173, 1421, 2647, 1388, 186
countermove history Note: stall because worker Stop reason: I ain't fucking waiting for -0.2+-3 on sth that should be +20...