Dev Builds » 20230606-1917

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 06:54:36 583716 4014 1313 688 2013 +54.54 ± 5.11 5 193 990 810 9 +111.14 ± 10.78
ncm-dbt-02 06:54:07 585626 4000 1302 715 1983 +51.36 ± 5.1 3 208 993 791 5 +104.68 ± 10.77
ncm-dbt-03 06:53:29 585337 4000 1312 719 1969 +51.89 ± 5.13 2 210 991 787 10 +104.68 ± 10.78
ncm-dbt-04 06:54:17 570597 4000 1306 741 1953 +49.41 ± 5.03 1 208 1022 763 6 +99.95 ± 10.59
ncm-dbt-05 06:54:41 582966 3986 1300 736 1950 +49.49 ± 5.09 3 207 1014 761 8 +100.13 ± 10.64
20000 6533 3599 9868 +51.34 ± 2.28 14 1026 5010 3912 38 +104.11 ± 4.79

Test Detail

ID Host Base NPS Games WLD Standard Elo Ptnml(0-2) Gamepair Elo CLI PGN
408680 ncm-dbt-01 582569 14 6 2 6 +102.04 ± 70.73 0 0 3 4 0 +225.59 ± 263.0
408679 ncm-dbt-05 585126 486 161 91 234 +50.39 ± 14.45 1 23 124 95 0 +104.56 ± 30.47
408678 ncm-dbt-02 587367 500 163 92 245 +49.67 ± 15.03 1 27 125 94 3 +98.44 ± 30.45
408677 ncm-dbt-04 569310 500 160 101 239 +41.19 ± 14.43 0 30 133 85 2 +80.63 ± 29.38
408676 ncm-dbt-03 587707 500 174 100 226 +51.8 ± 14.08 0 23 132 93 2 +102.97 ± 29.38
408675 ncm-dbt-01 585928 500 173 80 247 +65.38 ± 14.0 0 19 120 110 1 +134.15 ± 31.01
408674 ncm-dbt-05 582527 500 158 90 252 +47.54 ± 13.85 0 24 135 90 1 +95.44 ± 28.99
408673 ncm-dbt-02 587792 500 173 82 245 +63.95 ± 13.36 0 16 127 107 0 +132.54 ± 29.87
408672 ncm-dbt-03 585169 500 161 95 244 +46.13 ± 14.72 0 31 123 95 1 +92.46 ± 30.75
408671 ncm-dbt-04 570869 500 169 98 233 +49.67 ± 14.64 1 27 122 100 0 +102.97 ± 30.86
408670 ncm-dbt-01 584664 500 150 86 264 +44.72 ± 15.15 1 31 123 93 2 +89.48 ± 30.76
408669 ncm-dbt-05 580862 500 157 97 246 +41.89 ± 14.99 0 35 121 93 1 +83.57 ± 31.04
408668 ncm-dbt-02 584495 500 168 77 255 +63.95 ± 14.37 0 23 113 114 0 +132.54 ± 32.13
408667 ncm-dbt-03 586646 500 167 82 251 +59.64 ± 14.59 1 21 122 104 2 +121.46 ± 30.78
408666 ncm-dbt-04 570829 500 164 94 242 +48.96 ± 13.93 0 25 130 95 0 +99.95 ± 29.7
408665 ncm-dbt-01 581361 500 160 86 254 +51.8 ± 13.94 0 24 128 98 0 +106.01 ± 29.96
408664 ncm-dbt-05 583196 500 162 91 247 +49.67 ± 13.97 1 22 132 95 0 +102.97 ± 29.38
408663 ncm-dbt-04 571391 500 166 86 248 +56.07 ± 14.43 0 23 127 97 3 +110.6 ± 30.08
408662 ncm-dbt-02 585464 500 159 98 243 +42.6 ± 14.52 1 28 131 89 1 +86.52 ± 29.64
408661 ncm-dbt-03 585632 500 167 87 246 +56.07 ± 14.43 0 25 121 103 1 +113.68 ± 30.97
408660 ncm-dbt-01 583656 500 166 83 251 +58.21 ± 14.39 1 21 123 104 1 +119.89 ± 30.63
408659 ncm-dbt-05 581194 500 164 89 247 +52.51 ± 14.25 1 23 126 100 0 +109.07 ± 30.24
408658 ncm-dbt-04 570148 500 161 102 237 +41.19 ± 13.89 0 28 135 87 0 +83.57 ± 29.08
408657 ncm-dbt-02 585970 500 159 89 252 +48.96 ± 15.13 1 30 118 100 1 +99.95 ± 31.44
408656 ncm-dbt-03 585211 500 147 81 272 +46.13 ± 14.18 0 28 128 94 0 +93.95 ± 30.03
408655 ncm-dbt-01 581527 500 168 88 244 +56.07 ± 14.15 0 23 125 101 1 +113.68 ± 30.37
408654 ncm-dbt-05 586816 500 164 92 244 +50.38 ± 14.14 0 25 129 95 1 +101.46 ± 29.84
408653 ncm-dbt-02 584411 500 160 85 255 +52.51 ± 14.66 0 29 117 104 0 +107.54 ± 31.57
408652 ncm-dbt-04 568275 500 163 83 254 +56.07 ± 14.01 0 22 127 100 1 +113.68 ± 30.06
408651 ncm-dbt-01 585042 500 159 91 250 +47.55 ± 15.18 1 29 124 93 3 +93.95 ± 30.6
408650 ncm-dbt-03 584706 500 162 88 250 +51.8 ± 14.49 0 28 120 102 0 +106.01 ± 31.14
408649 ncm-dbt-05 582903 500 169 92 239 +53.93 ± 14.32 0 24 127 97 2 +107.54 ± 30.1
408648 ncm-dbt-02 582736 500 167 88 245 +55.36 ± 14.26 0 25 121 104 0 +113.68 ± 30.97
408647 ncm-dbt-04 570629 500 161 85 254 +53.22 ± 14.01 0 24 126 100 0 +109.07 ± 30.24
408646 ncm-dbt-01 585169 500 166 87 247 +55.36 ± 14.53 2 20 126 101 1 +115.23 ± 30.2
408645 ncm-dbt-03 581402 500 169 88 243 +56.78 ± 14.86 1 25 117 106 1 +116.78 ± 31.56
408644 ncm-dbt-05 581111 500 165 94 241 +49.67 ± 15.16 0 31 120 96 3 +96.94 ± 31.16
408643 ncm-dbt-02 586774 500 153 104 243 +34.16 ± 13.72 0 30 141 79 0 +68.99 ± 28.31
408642 ncm-dbt-04 573325 500 162 92 246 +48.96 ± 14.47 0 29 122 99 0 +99.95 ± 30.87
408641 ncm-dbt-01 583531 500 165 85 250 +56.07 ± 14.43 0 26 118 106 0 +115.23 ± 31.41
408640 ncm-dbt-03 586223 500 165 98 237 +46.84 ± 14.75 0 29 128 90 3 +90.97 ± 30.05

Commit

Commit ID 373359b44d0947cce2628a9a8c9b432a458615a8
Author Linmiao Xu
Date 2023-06-06 19:17:36 UTC
Update default net to nn-0dd1cebea573.nnue Created by retraining an earlier epoch of the experiment leading to the first SFNNv6 net on a more-randomized version of the nn-e1fb1ade4432.nnue dataset mixed with unfiltered T80 apr2023 data. Trained using early-fen-skipping 28 and max-epoch 960. The trainer settings and epochs used in the 5-step training sequence leading here were: 1. train from scratch for 400 epochs, lambda 1.0, constant LR 9.75e-4, T79T77-filter-v6-dd.min.binpack 2. retrain ep379, max-epoch 800, end-lambda 0.75, T60T70wIsRightFarseerT60T74T75T76.binpack 3. retrain ep679, max-epoch 800, end-lambda 0.75, skip 28, nn-e1fb1ade4432 dataset 4. retrain ep799, max-epoch 800, end-lambda 0.7, skip 28, nn-e1fb1ade4432 dataset 5. retrain ep439, max-epoch 960, end-lambda 0.7, skip 28, shuffled nn-e1fb1ade4432 + T80 apr2023 This net was epoch 559 of the final (step 5) retraining: ```bash python3 easy_train.py \ --experiment-name L1-1536-Re4-leela96-dfrc99-T60novdec-v2-T80juntonovjanfebT79aprmayT78jantosepT77dec-v6dd-T80apr-shuffled-sk28 \ --training-dataset /data/leela96-dfrc99-T60novdec-v2-T80juntonovjanfebT79aprmayT78jantosepT77dec-v6dd-T80apr.binpack \ --nnue-pytorch-branch linrock/nnue-pytorch/misc-fixes-L1-1536 \ --early-fen-skipping 28 \ --start-lambda 1.0 \ --end-lambda 0.7 \ --max_epoch 960 \ --start-from-engine-test-net False \ --start-from-model /data/L1-1536-Re3-nn-epoch439.nnue \ --engine-test-branch linrock/Stockfish/L1-1536 \ --lr 4.375e-4 \ --gamma 0.995 \ --tui False \ --seed $RANDOM \ --gpus "0," ``` During data preparation, most binpacks were unminimized by removing positions with score 32002 (`VALUE_NONE`). This makes the tradeoff of increasing dataset filesize on disk to increase the randomness of positions in interleaved datasets. The code used for unminimizing is at: https://github.com/linrock/Stockfish/tree/tools-unminify For preparing the dataset used in this experiment: ```bash python3 interleave_binpacks.py \ leela96-filt-v2.binpack \ dfrc99-16tb7p-eval-filt-v2.binpack \ filt-v6-dd-min/test60-novdec2021-12tb7p-filter-v6-dd.min-mar2023.unmin.binpack \ filt-v6-dd-min/test80-aug2022-16tb7p-filter-v6-dd.min-mar2023.unmin.binpack \ filt-v6-dd-min/test80-sep2022-16tb7p-filter-v6-dd.min-mar2023.unmin.binpack \ filt-v6-dd-min/test80-jun2022-16tb7p-filter-v6-dd.min-mar2023.unmin.binpack \ filt-v6-dd/test80-jul2022-16tb7p-filter-v6-dd.binpack \ filt-v6-dd/test80-oct2022-16tb7p-filter-v6-dd.binpack \ filt-v6-dd/test80-nov2022-16tb7p-filter-v6-dd.binpack \ filt-v6-dd-min/test80-jan2023-3of3-16tb7p-filter-v6-dd.min-mar2023.unmin.binpack \ filt-v6-dd-min/test80-feb2023-16tb7p-filter-v6-dd.min-mar2023.unmin.binpack \ filt-v6-dd/test79-apr2022-16tb7p-filter-v6-dd.binpack \ filt-v6-dd/test79-may2022-16tb7p-filter-v6-dd.binpack \ filt-v6-dd-min/test78-jantomay2022-16tb7p-filter-v6-dd.min-mar2023.unmin.binpack \ filt-v6-dd/test78-juntosep2022-16tb7p-filter-v6-dd.binpack \ filt-v6-dd/test77-dec2021-16tb7p-filter-v6-dd.binpack \ test80-apr2023-2tb7p.binpack \ /data/leela96-dfrc99-T60novdec-v2-T80juntonovjanfebT79aprmayT78jantosepT77dec-v6dd-T80apr.binpack ``` T80 apr2023 data was converted using lc0-rescorer with ~2tb of tablebases and can be found at: https://robotmoon.com/nnue-training-data/ Local elo at 25k nodes per move vs. nn-e1fb1ade4432.nnue (L1 size 1024): nn-epoch559.nnue : 25.7 +/- 1.6 Passed STC: https://tests.stockfishchess.org/tests/view/647cd3b87cf638f0f53f9cbb LLR: 2.95 (-2.94,2.94) <0.00,2.00> Total: 59200 W: 16000 L: 15660 D: 27540 Ptnml(0-2): 159, 6488, 15996, 6768, 189 Passed LTC: https://tests.stockfishchess.org/tests/view/647d58de726f6b400e4085d8 LLR: 2.95 (-2.94,2.94) <0.50,2.50> Total: 58800 W: 16002 L: 15657 D: 27141 Ptnml(0-2): 44, 5607, 17748, 5962, 39 closes https://github.com/official-stockfish/Stockfish/pull/4606 bench 2141197
Copyright 2011–2025 Next Chess Move LLC