Dev Builds » 20210618-2150

Use this dev build

NCM plays each Stockfish dev build 20,000 times against Stockfish 14. This yields an approximate Elo difference and establishes confidence in the strength of the dev builds.

Summary

Host Duration Avg Base NPS Games WLD Standard Elo Ptnml(0-2) Gamepair Elo
ncm-dbt-01 14:42:47 1136480 4942 1135 1233 2574 -6.89 +/- 4.73 8 616 1313 534 0 -12.66 +/- 9.38
ncm-dbt-02 14:38:35 1253960 4870 1090 1284 2496 -13.85 +/- 4.85 9 673 1261 487 5 -27.17 +/- 9.58
ncm-dbt-03 14:44:09 1264111 4942 1128 1304 2510 -12.38 +/- 4.85 13 669 1274 511 4 -23.52 +/- 9.54
ncm-dbt-04 00:30:49 1486302 172 34 38 100 -8.08 +/- 24.4 0 21 48 17 0 -16.17 +/- 49.12
ncm-dbt-05 14:38:48 1257928 4910 1080 1268 2562 -13.31 +/- 4.8 12 656 1301 480 6 -25.8 +/- 9.42
ncm-dbt-06 00:31:25 1497965 164 32 37 95 -10.6 +/- 24.52 0 20 47 15 0 -21.21 +/- 49.43
20000 4499 5164 10337 -11.56 +/- 2.38 42 2655 5244 2044 15 -22.2 +/- 4.7

Test Detail

ID Host Base NPS Games WLD Standard Elo Ptnml(0-2) Gamepair Elo CLI PGN
211318 ncm-dbt-02 1226945 208 47 54 107 -11.7 +/- 23.33 0 28 56 19 1 -26.78 +/- 45.61
211317 ncm-dbt-05 1219388 240 48 57 135 -13.03 +/- 19.36 0 28 73 19 0 -26.11 +/- 39.01
211316 ncm-dbt-03 1244868 276 64 67 145 -3.78 +/- 20.21 0 35 71 32 0 -7.55 +/- 40.57
211315 ncm-dbt-01 1114811 276 62 66 148 -5.04 +/- 20.65 0 37 68 33 0 -10.07 +/- 41.48
211314 ncm-dbt-02 1242011 500 113 138 249 -17.38 +/- 14.87 2 67 135 46 0 -32.05 +/- 29.26
211313 ncm-dbt-05 1218023 500 107 139 254 -22.27 +/- 16.3 3 78 119 48 2 -43.3 +/- 31.28
211312 ncm-dbt-01 1113812 500 124 134 242 -6.95 +/- 14.66 1 60 137 52 0 -12.51 +/- 29.02
211311 ncm-dbt-03 1223948 500 113 135 252 -15.3 +/- 14.96 1 69 131 49 0 -29.25 +/- 29.78
211310 ncm-dbt-02 1219091 500 114 136 250 -15.3 +/- 15.69 1 75 119 55 0 -29.25 +/- 31.27
211309 ncm-dbt-05 1232634 500 109 130 261 -14.6 +/- 14.0 1 61 146 42 0 -27.85 +/- 27.8
211308 ncm-dbt-01 1107220 500 125 118 257 +4.87 +/- 15.23 1 56 128 65 0 +11.12 +/- 30.16
211307 ncm-dbt-03 1230496 500 130 131 239 -0.69 +/- 15.24 0 63 125 62 0 -1.39 +/- 30.53
211306 ncm-dbt-02 1245990 500 116 113 271 +2.09 +/- 15.36 2 56 129 63 0 +6.95 +/- 30.04
211305 ncm-dbt-05 1245895 500 108 127 265 -13.21 +/- 16.01 0 78 114 57 1 -27.85 +/- 31.87
211304 ncm-dbt-01 1101171 500 119 124 257 -3.47 +/- 14.74 1 58 136 55 0 -5.56 +/- 29.15
211303 ncm-dbt-03 1245094 500 101 126 273 -17.39 +/- 14.75 2 65 140 42 1 -33.46 +/- 28.6
211302 ncm-dbt-02 1229221 500 106 136 258 -20.87 +/- 15.01 0 76 129 44 1 -43.3 +/- 30.02
211301 ncm-dbt-05 1226690 500 117 132 251 -10.43 +/- 14.57 0 65 135 50 0 -20.87 +/- 29.27
211300 ncm-dbt-01 1115612 500 118 117 265 +0.7 +/- 14.87 2 53 137 58 0 +4.17 +/- 29.02
211299 ncm-dbt-03 1231716 500 116 116 268 0.0 +/- 14.93 1 57 133 59 0 +1.39 +/- 29.53
211298 ncm-dbt-02 1213990 500 101 139 260 -26.45 +/- 14.78 2 74 134 40 0 -50.38 +/- 29.35
211297 ncm-dbt-05 1252196 500 123 123 254 0.0 +/- 15.54 1 62 123 64 0 +1.39 +/- 30.78
211296 ncm-dbt-01 1096659 500 118 132 250 -9.73 +/- 15.62 0 73 118 59 0 -19.48 +/- 31.39
211295 ncm-dbt-03 1231661 500 106 130 264 -16.69 +/- 15.56 2 72 124 52 0 -30.65 +/- 30.66
211294 ncm-dbt-02 1234108 500 108 132 260 -16.69 +/- 15.07 0 74 126 50 0 -33.46 +/- 30.41
211293 ncm-dbt-01 1102875 500 96 127 277 -21.57 +/- 14.81 0 76 129 45 0 -43.3 +/- 30.02
211292 ncm-dbt-05 1231274 500 105 131 264 -18.08 +/- 15.05 0 74 129 46 1 -37.67 +/- 30.03
211291 ncm-dbt-03 1262483 500 115 120 265 -3.47 +/- 14.86 0 62 131 57 0 -6.95 +/- 29.79
211290 ncm-dbt-02 1229636 500 125 137 238 -8.34 +/- 15.27 2 62 133 52 1 -15.3 +/- 29.53
211289 ncm-dbt-01 1098127 500 109 124 267 -10.43 +/- 14.7 0 66 133 51 0 -20.87 +/- 29.53
211288 ncm-dbt-05 1231523 500 114 122 264 -5.56 +/- 14.66 1 59 137 53 0 -9.73 +/- 29.02
211287 ncm-dbt-03 1245109 500 117 157 226 -27.85 +/- 16.67 4 83 114 47 2 -53.22 +/- 31.91
211286 ncm-dbt-02 1236808 500 116 129 255 -9.04 +/- 14.71 0 65 133 52 0 -18.08 +/- 29.53
211285 ncm-dbt-01 1092010 500 106 120 274 -9.73 +/- 14.25 1 59 143 47 0 -18.08 +/- 28.23
211284 ncm-dbt-05 1224450 500 102 126 272 -16.69 +/- 14.69 2 65 138 45 0 -30.65 +/- 28.87
211283 ncm-dbt-03 1235112 500 116 143 241 -18.78 +/- 14.73 1 69 137 42 1 -37.67 +/- 28.99
211282 ncm-dbt-03 1255851 500 110 135 255 -17.38 +/- 15.49 2 72 125 51 0 -32.05 +/- 30.53
211281 ncm-dbt-01 1070454 500 121 132 247 -7.64 +/- 14.84 2 59 137 52 0 -12.51 +/- 29.02
211280 ncm-dbt-05 1259855 500 108 135 257 -18.78 +/- 15.71 4 68 130 47 1 -33.46 +/- 29.9
211279 ncm-dbt-02 1225932 500 108 125 267 -11.82 +/- 15.43 0 72 124 53 1 -25.06 +/- 30.66
178228 ncm-dbt-02 1489831 162 36 45 81 -19.33 +/- 26.72 0 24 43 13 1 -43.11 +/- 52.17
178227 ncm-dbt-06 1497965 164 32 37 95 -10.6 +/- 24.52 0 20 47 15 0 -21.21 +/- 49.43
178226 ncm-dbt-04 1486302 172 34 38 100 -8.08 +/- 24.4 0 21 48 17 0 -16.17 +/- 49.12
178225 ncm-dbt-03 1498889 166 40 44 82 -8.37 +/- 25.95 0 22 43 18 0 -16.76 +/- 52.28
178224 ncm-dbt-05 1495280 170 39 46 85 -14.32 +/- 22.16 0 18 57 9 1 -32.8 +/- 42.41
178223 ncm-dbt-01 1488529 166 37 39 90 -4.19 +/- 24.64 0 19 47 17 0 -8.37 +/- 49.56

Commit

Commit ID adfb23c029e54c7522aadca1adf3e0b15fdcebcd
Author Joost VandeVondele
Date 2021-06-18 21:50:26 UTC
Make net nn-50144f835024.nnue the default trained with the Python command c:\nnue>python train.py i:/bin/all.binpack i:/bin/all.binpack --gpus 1 --threads 4 --num-workers 30 --batch-size 16384 --progress_bar_refresh_rate 300 --smart-fen-skipping --random-fen-skipping 3 --features=HalfKAv2^ --lambda=1.0 --max_epochs=440 --seed %random%%random% --default_root_dir exp/run_8 --resume-from-model ./pt/nn-6ad41a9207d0.pt ` all.binpack equaled 4 parts Wrong_NNUE_2.binpack https://drive.google.com/file/d/1seGNOqcVdvK_vPNq98j-zV3XPE5zWAeq/view?usp=sharing plus two parts of Training_Data.binpack https://drive.google.com/file/d/1RFkQES3DpsiJqsOtUshENtzPfFgUmEff/view?usp=sharing Each set was concatenated together - make one large Wrong_NNUE 2 binpack and one large Training_Data of approximate size. They were then interleaved together. The idea was to give Wrong_NNUE.binpack closer to equal weighting with the Training _Data binpack . nn-6ad41a9207d0.pt was derived from a net vondele ran which passed STC quickly, but faltered in LTC. https://tests.stockfishchess.org/tests/view/60cba666457376eb8bcab443 STC: LLR: 2.95 (-2.94,2.94) <-0.50,2.50> Total: 18792 W: 2068 L: 1889 D: 14835 Ptnml(0-2): 82, 1480, 6117, 1611, 106 https://tests.stockfishchess.org/tests/view/60ccda8b457376eb8bcab568 LTC: LLR: 2.94 (-2.94,2.94) <0.50,3.50> Total: 11376 W: 574 L: 454 D: 10348 Ptnml(0-2): 4, 412, 4747, 510, 15 https://tests.stockfishchess.org/tests/view/60ccf952457376eb8bcab58d closes https://github.com/official-stockfish/Stockfish/pull/3568 Bench: 4900906
Copyright 2011–2024 Next Chess Move LLC