Dev Builds » 20230101-1128

Use this dev build

NCM plays each Stockfish dev build 20,000 times against Stockfish 15. 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 07:00:45 583607 3998 1168 880 1950 +25.07 ± 5.25 9 314 1061 610 5 +51.12 ± 10.42
ncm-dbt-02 07:00:31 586536 4000 1186 856 1958 +28.73 ± 5.15 6 290 1078 620 6 +57.86 ± 10.31
ncm-dbt-03 07:00:33 584125 4014 1213 890 1911 +28.02 ± 5.2 6 309 1050 640 2 +57.12 ± 10.48
ncm-dbt-04 07:00:15 569715 3988 1191 888 1909 +26.45 ± 5.14 1 311 1071 606 5 +52.49 ± 10.36
ncm-dbt-05 06:59:27 577842 4000 1143 903 1954 +20.87 ± 5.15 8 318 1106 562 6 +42.25 ± 10.16
20000 5901 4417 9682 +25.83 ± 2.32 30 1542 5366 3038 24 +52.16 ± 4.63

Test Detail

ID Host Base NPS Games WLD Standard Elo Ptnml(0-2) Gamepair Elo CLI PGN
430954 ncm-dbt-03 584285 14 4 4 6 -0.0 ± 69.83 0 1 5 1 0 -0.0 ± 145.53 ↓
430953 ncm-dbt-04 569829 488 153 98 237 +39.32 ± 13.9 0 27 135 82 0 +79.68 ± 28.98 ↓
430952 ncm-dbt-05 577520 500 145 127 228 +12.51 ± 14.74 1 46 138 64 1 +25.06 ± 28.88 ↓
430951 ncm-dbt-02 587749 500 158 108 234 +34.86 ± 14.31 0 34 132 84 0 +70.44 ± 29.57 ↓
430950 ncm-dbt-01 582235 498 146 92 260 +37.82 ± 15.06 0 36 125 86 2 +73.64 ± 30.5 ↓
430949 ncm-dbt-03 580572 500 156 105 239 +35.56 ± 15.12 1 37 122 90 0 +73.34 ± 30.91 ↓
430948 ncm-dbt-04 570789 500 147 108 245 +27.15 ± 14.31 0 38 135 77 0 +54.65 ± 29.21 ↓
430947 ncm-dbt-02 587409 500 151 104 245 +32.75 ± 15.2 0 40 125 83 2 +63.23 ± 30.52 ↓
430946 ncm-dbt-05 573770 500 142 108 250 +23.66 ± 14.71 1 39 136 73 1 +47.55 ± 29.1 ↓
430945 ncm-dbt-01 582819 500 138 113 249 +17.39 ± 14.49 3 36 144 67 0 +39.08 ± 28.04 ↓
430944 ncm-dbt-03 583489 500 157 122 221 +24.36 ± 14.37 1 37 138 74 0 +50.38 ± 28.82 ↓
430943 ncm-dbt-04 570789 500 148 118 234 +20.87 ± 15.62 0 51 119 79 1 +40.48 ± 31.28 ↓
430942 ncm-dbt-05 580406 500 145 105 250 +27.85 ± 15.12 3 33 137 75 2 +57.5 ± 28.93 ↓
430941 ncm-dbt-02 582318 500 143 103 254 +27.85 ± 13.96 0 34 143 72 1 +54.65 ± 28.11 ↓
430940 ncm-dbt-01 583070 500 144 114 242 +20.87 ± 15.01 1 43 132 73 1 +41.89 ± 29.63 ↓
430939 ncm-dbt-03 586012 500 136 96 268 +27.85 ± 15.13 0 44 122 84 0 +56.07 ± 30.91 ↓
430938 ncm-dbt-04 570629 500 146 116 238 +20.87 ± 14.24 0 41 138 71 0 +41.89 ± 28.84 ↓
430937 ncm-dbt-05 581569 500 140 115 245 +17.39 ± 14.75 0 47 131 72 0 +34.86 ± 29.77 ↓
430936 ncm-dbt-02 586054 500 151 111 238 +27.85 ± 14.62 0 39 133 77 1 +54.65 ± 29.48 ↓
430935 ncm-dbt-01 582027 500 136 113 251 +15.99 ± 14.76 1 45 134 70 0 +33.46 ± 29.39 ↓
430934 ncm-dbt-03 584453 500 149 121 230 +19.48 ± 14.39 1 39 142 67 1 +39.08 ± 28.31 ↓
430933 ncm-dbt-04 568951 500 154 114 232 +27.85 ± 15.25 0 43 126 79 2 +53.22 ± 30.4 ↓
430932 ncm-dbt-05 574135 500 138 103 259 +24.36 ± 13.84 0 35 146 68 1 +47.55 ± 27.73 ↓
430931 ncm-dbt-02 588771 500 150 105 245 +31.35 ± 14.85 2 32 137 77 2 +63.23 ± 28.9 ↓
430930 ncm-dbt-01 583070 500 131 108 261 +15.99 ± 15.14 2 45 131 72 0 +34.86 ± 29.77 ↓
430929 ncm-dbt-03 585632 500 155 117 228 +26.46 ± 14.9 1 40 129 80 0 +54.65 ± 30.01 ↓
430928 ncm-dbt-04 569669 500 149 114 237 +24.36 ± 13.84 0 35 146 68 1 +47.55 ± 27.73 ↓
430927 ncm-dbt-05 578424 500 141 117 242 +16.69 ± 14.82 1 45 133 71 0 +34.86 ± 29.51 ↓
430926 ncm-dbt-01 583824 500 156 112 232 +30.65 ± 14.81 1 37 129 83 0 +63.23 ± 29.99 ↓
430925 ncm-dbt-02 587240 500 145 112 243 +22.97 ± 14.13 2 33 145 70 0 +48.96 ± 27.86 ↓
430924 ncm-dbt-03 583740 500 150 106 244 +30.65 ± 14.42 0 37 132 81 0 +61.79 ± 29.59 ↓
430923 ncm-dbt-04 569629 500 149 110 241 +27.16 ± 14.7 1 37 135 76 1 +54.65 ± 29.21 ↓
430922 ncm-dbt-05 575843 500 144 113 243 +21.57 ± 14.29 0 40 140 69 1 +41.89 ± 28.57 ↓
430921 ncm-dbt-01 585253 500 162 124 214 +26.46 ± 14.9 1 39 132 77 1 +53.22 ± 29.61 ↓
430920 ncm-dbt-03 585000 500 154 105 241 +34.16 ± 14.52 1 33 132 84 0 +70.44 ± 29.57 ↓
430919 ncm-dbt-02 587452 500 137 106 257 +21.57 ± 14.81 0 45 129 76 0 +43.3 ± 30.02 ↓
430918 ncm-dbt-04 567442 500 145 110 245 +24.36 ± 14.24 0 39 137 74 0 +48.96 ± 28.96 ↓
430917 ncm-dbt-05 581070 500 148 115 237 +22.97 ± 14.13 2 33 145 70 0 +48.96 ± 27.86 ↓
430916 ncm-dbt-01 586562 500 155 104 241 +35.56 ± 14.35 0 33 134 82 1 +70.44 ± 29.29 ↓
430915 ncm-dbt-03 583950 500 152 114 234 +26.46 ± 15.15 1 41 128 79 1 +53.22 ± 30.14 ↓
430914 ncm-dbt-02 585295 500 151 107 242 +30.65 ± 14.68 2 33 134 81 0 +64.66 ± 29.31 ↓

Commit

Commit ID be9bc420afa11314c886c0aede4c4ae3d76f8a50
Author Linmiao Xu
Date 2023-01-01 11:28:51 UTC
Update default net to nn-60fa44e376d9.nnue Created by retraining the master net on the previous best dataset with additional filtering. No new data was added. More of the Leela-dfrc_n5000.binpack part of the dataset was pre-filtered with depth6 multipv2 search to remove bestmove captures. About 93% of the previous Leela/SF data and 99% of the SF dfrc data was filtered. Unfiltered parts of the dataset were left out. The new Leela T80 oct+nov data is the same as before. All early game positions with ply count <= 28 were skipped during training by modifying the training data loader in nnue-pytorch. Trained in a similar way as recent master nets, with a different nnue-pytorch branch for early ply skipping: python3 easy_train.py \ --experiment-name=leela93-dfrc99-filt-only-T80-oct-nov-skip28 \ --training-dataset=/data/leela93-dfrc99-filt-only-T80-oct-nov.binpack \ --start-from-engine-test-net True \ --nnue-pytorch-branch=linrock/nnue-pytorch/misc-fixes-skip-ply-lteq-28 \ --gpus="0," \ --start-lambda=1.0 \ --end-lambda=0.75 \ --gamma=0.995 \ --lr=4.375e-4 \ --tui=False \ --seed=$RANDOM \ --max_epoch=800 \ --network-testing-threads 20 \ --num-workers 6 For the exact training data used: https://robotmoon.com/nnue-training-data/ Details about the previous best dataset: #4295 Local testing at a fixed 25k nodes: experiment_leela93-dfrc99-filt-only-T80-oct-nov-skip28 Local Elo: run_0/nn-epoch779.nnue : 5.1 +/- 1.5 Passed STC https://tests.stockfishchess.org/tests/view/63adb3acae97a464904fd4e8 LLR: 2.94 (-2.94,2.94) <0.00,2.00> Total: 36504 W: 9847 L: 9538 D: 17119 Ptnml(0-2): 108, 3981, 9784, 4252, 127 Passed LTC https://tests.stockfishchess.org/tests/view/63ae0ae25bd1e5f27f13d884 LLR: 2.94 (-2.94,2.94) <0.50,2.50> Total: 36592 W: 10017 L: 9717 D: 16858 Ptnml(0-2): 17, 3461, 11037, 3767, 14 closes https://github.com/official-stockfish/Stockfish/pull/4314 bench 4015511
Copyright 2011–2025 Next Chess Move LLC
Feedback & Support Terms of Service Privacy Policy Desktop Version