Dev Builds » 20230227-2107

You are viewing an old NCM Stockfish dev build test. You may find the most recent dev build tests using Stockfish 15 as the baseline here.

Use this dev build

NCM plays each Stockfish dev build 20,000 times against Stockfish 7. This yields an approximate Elo difference and establishes confidence in the strength of the dev builds.

Summary

Host Duration Avg Base NPS Games Wins Losses Draws Elo
ncm-et-3 09:38:30 1955785 3337 2867 6 464 +445.86 ± 15.82
ncm-et-4 09:38:22 1957458 3355 2893 3 459 +451.23 ± 15.9
ncm-et-9 09:38:51 1950057 3324 2873 4 447 +453.56 ± 16.12
ncm-et-10 09:38:57 1949571 3317 2827 2 488 +438.54 ± 15.41
ncm-et-13 09:39:04 1956045 3338 2878 9 451 +448.68 ± 16.06
ncm-et-15 09:38:53 1954544 3329 2852 8 469 +441.9 ± 15.74
20000 17190 32 2778 +446.57 ± 6.45

Test Detail

ID Host Started (UTC) Duration Base NPS Games Wins Losses Draws Elo CLI PGN
172354 ncm-et-15 2023-03-04 23:44 00:27:19 1960949 161 137 1 23 +429.93 ± 74.39
172353 ncm-et-9 2023-03-04 23:43 00:27:39 1954572 152 134 0 18 +480.42 ± 85.44
172352 ncm-et-13 2023-03-04 23:43 00:28:23 1944645 158 136 1 21 +442.06 ± 78.29
172351 ncm-et-10 2023-03-04 23:42 00:29:14 1946469 169 132 0 37 +364.14 ± 56.96
172350 ncm-et-3 2023-03-04 23:41 00:30:14 1956177 172 145 1 26 +421.01 ± 69.53
172349 ncm-et-4 2023-03-04 23:38 00:33:24 1953140 188 166 0 22 +482.62 ± 76.32
172341 ncm-et-9 2023-03-04 22:16 01:26:59 1949254 500 445 1 54 +490.71 ± 47.28
172340 ncm-et-3 2023-03-04 22:15 01:24:23 1962007 500 419 1 80 +419.61 ± 38.47
172339 ncm-et-15 2023-03-04 22:15 01:27:46 1953447 500 434 0 66 +460.32 ± 42.48
172338 ncm-et-13 2023-03-04 22:15 01:27:20 1954508 500 431 2 67 +446.7 ± 42.25
172337 ncm-et-10 2023-03-04 22:15 01:26:25 1956790 500 429 0 71 +446.7 ± 40.89
172336 ncm-et-4 2023-03-04 22:12 01:25:23 1954657 500 437 0 63 +468.95 ± 43.54
172329 ncm-et-3 2023-03-04 20:49 01:25:31 1961271 500 431 0 69 +452.04 ± 41.5
172328 ncm-et-15 2023-03-04 20:48 01:26:19 1953150 500 426 1 73 +436.43 ± 40.36
172327 ncm-et-9 2023-03-04 20:48 01:27:12 1942388 500 432 0 68 +454.76 ± 41.82
172326 ncm-et-10 2023-03-04 20:47 01:27:17 1955414 500 436 0 64 +466.03 ± 43.18
172325 ncm-et-13 2023-03-04 20:47 01:27:29 1958803 500 439 0 61 +474.93 ± 44.28
172324 ncm-et-4 2023-03-04 20:45 01:25:43 1960318 500 441 0 59 +481.09 ± 45.07
172317 ncm-et-3 2023-03-04 19:20 01:28:36 1955420 500 432 0 68 +454.76 ± 41.82
172316 ncm-et-15 2023-03-04 19:20 01:27:39 1947066 500 430 0 70 +449.35 ± 41.19
172315 ncm-et-13 2023-03-04 19:20 01:25:54 1960925 500 429 1 70 +444.09 ± 41.26
172314 ncm-et-9 2023-03-04 19:20 01:27:14 1946915 500 438 1 61 +468.96 ± 44.35
172313 ncm-et-10 2023-03-04 19:20 01:26:02 1950509 500 421 0 79 +426.65 ± 38.66
172312 ncm-et-4 2023-03-04 19:20 01:24:29 1961120 500 426 0 74 +438.95 ± 40.01
172142 ncm-et-10 2023-03-04 04:29 00:25:49 1954201 148 131 0 17 +486.04 ± 88.33
172141 ncm-et-4 2023-03-04 04:25 00:28:31 1969240 167 143 1 23 +436.81 ± 74.42
172140 ncm-et-3 2023-03-04 04:25 00:29:03 1950395 165 139 1 25 +420.03 ± 71.03
172139 ncm-et-15 2023-03-04 04:25 00:29:19 1952683 168 144 2 22 +430.56 ± 75.95
172138 ncm-et-9 2023-03-04 04:24 00:30:25 1960320 172 143 0 29 +414.35 ± 65.23
172137 ncm-et-13 2023-03-04 04:24 00:30:47 1953132 180 154 0 26 +443.5 ± 69.39
172130 ncm-et-10 2023-03-04 02:59 01:29:27 1943467 500 427 1 72 +438.95 ± 40.66
172129 ncm-et-3 2023-03-04 02:58 01:26:14 1957118 500 427 2 71 +436.43 ± 40.99
172128 ncm-et-4 2023-03-04 02:58 01:26:41 1959252 500 434 0 66 +460.32 ± 42.48
172127 ncm-et-15 2023-03-04 02:57 01:26:56 1956025 500 420 2 78 +419.61 ± 39.03
172126 ncm-et-13 2023-03-04 02:57 01:25:59 1955595 500 417 0 83 +417.31 ± 37.67
172125 ncm-et-9 2023-03-04 02:56 01:26:55 1942399 500 426 1 73 +436.43 ± 40.36
172118 ncm-et-13 2023-03-04 01:31 01:25:45 1963357 500 438 2 60 +466.04 ± 44.75
172117 ncm-et-3 2023-03-04 01:31 01:27:08 1955575 500 441 1 58 +477.99 ± 45.54
172116 ncm-et-10 2023-03-04 01:30 01:27:41 1938094 500 424 1 75 +431.48 ± 39.8
172115 ncm-et-9 2023-03-04 01:30 01:25:47 1948278 500 435 0 65 +463.15 ± 42.83
172114 ncm-et-4 2023-03-04 01:30 01:27:45 1946949 500 421 1 78 +424.28 ± 38.99
172113 ncm-et-15 2023-03-04 01:29 01:27:28 1954659 500 427 0 73 +441.5 ± 40.29
172100 ncm-et-9 2023-03-04 00:03 01:26:40 1956331 500 420 1 79 +421.93 ± 38.73
172099 ncm-et-3 2023-03-04 00:03 01:27:21 1948320 500 433 0 67 +457.52 ± 42.15
172098 ncm-et-10 2023-03-04 00:03 01:27:02 1951626 500 427 0 73 +441.5 ± 40.29
172097 ncm-et-15 2023-03-04 00:03 01:26:07 1958376 500 434 2 64 +454.76 ± 43.27
172096 ncm-et-13 2023-03-04 00:03 01:27:27 1957402 500 434 3 63 +452.04 ± 43.62
172095 ncm-et-4 2023-03-04 00:03 01:26:26 1954994 500 425 1 74 +433.94 ± 40.08

Commit

Commit ID 876906965b8d552866486c0e6eda1184fdb1d636
Author Linmiao Xu
Date 2023-02-27 21:07:52 UTC
Update default net to nn-52471d67216a.nnue Created by retraining the master net with modifications to the previous best dataset: * Improving T80 oct+nov 2022 endgame lambda accuracy by rescoring with 12-16tb of syzygy 7p tablebases * Filtering T78 jun+jul+aug 2022 with d6pv2 search to remove positions with bestmove captures or one good move * Adding T80 sep 2022 data, rescored with 16tb of 7p tablebases, unfiltered Trained with max-epoch 900, end-lambda 0.7, and early-fen-skipping 28. ``` python3 easy_train.py \ --experiment-name leela96-dfrc99-T80octnovT79aprmayT78junjulaugT60novdec-filt-v2-T78sep12tb7p-T77decT80sep16tb7p-lambda7-sk28 \ --training-dataset /data/leela96-dfrc99-T80octnovT79aprmayT78junjulaugT60novdec-filt-v2-T78sep12tb7p-T77decT80sep16tb7p.binpack \ --nnue-pytorch-branch linrock/nnue-pytorch/easy-train-early-fen-skipping \ --early-fen-skipping 28 \ --start-from-engine-test-net True \ --gpus "0," \ --max_epoch 900 \ --start-lambda 1.0 \ --end-lambda 0.7 \ --gamma 0.995 \ --lr 4.375e-4 \ --tui False \ --seed $RANDOM ``` Training data was rescored and d6pv2 filtered in the same way as recent best datasets. For preparing the merged training dataset: ``` python3 interleave_binpacks.py \ leela96-eval-filt-v2.binpack \ dfrc99-eval-filt-v2.binpack \ test80-oct2022-16tb7p-eval-filt-v2-d6.binpack \ test80-nov2022-12tb7p-eval-filt-v2-d6.binpack \ T79-apr2022-12tb7p-eval-filt-v2.binpack \ T79-may2022-12tb7p-eval-filt-v2.binpack \ test78-junjulaug2022-16tb7p-eval-filt-v2-d6.binpack \ T60-nov2021-12tb7p-eval-filt-v2.binpack \ T60-dec2021-12tb7p-eval-filt-v2.binpack \ T78-sep2022-12tb7p.binpack \ test77-dec2021-16gb7p.binpack \ test80-sep2022-16tb7p.binpack \ /data/leela96-dfrc99-T80octnovT79aprmayT78junjulaugT60novdec-filt-v2-T78sep12tb7p-T77decT80sep16tb7p.binpack ``` Links for downloading the training data components can be found at: https://robotmoon.com/nnue-training-data/ Local elo at 25k nodes per move: nn-epoch839.nnue : 0.6 +/- 1.4 Passed STC: https://tests.stockfishchess.org/tests/view/63f9ab4be74a12625bcdf02e LLR: 2.95 (-2.94,2.94) <0.00,2.00> Total: 84656 W: 22681 L: 22302 D: 39673 Ptnml(0-2): 271, 9343, 22734, 9696, 284 Passed LTC: https://tests.stockfishchess.org/tests/view/63fa3833e74a12625bce0c0e LLR: 2.94 (-2.94,2.94) <0.50,2.50> Total: 184664 W: 49933 L: 49344 D: 85387 Ptnml(0-2): 111, 17977, 55561, 18578, 105 closes https://github.com/official-stockfish/Stockfish/pull/4416 bench: 4814343
Copyright 2011–2024 Next Chess Move LLC