Dev Builds » 20230611-1323

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 09:57:00 1134001 3336 1362 317 1657 +112.62 +/- 5.42 0 61 531 1046 30 +245.48 +/- 14.79
ncm-dbt-02 09:49:44 1217584 3286 1382 309 1595 +117.76 +/- 5.42 0 55 492 1064 32 +259.67 +/- 15.37
ncm-dbt-03 09:56:24 1234988 3342 1359 322 1661 +111.48 +/- 5.41 0 59 548 1032 32 +241.6 +/- 14.56
ncm-dbt-04 09:56:07 1226663 3348 1407 323 1618 +116.69 +/- 5.33 0 46 536 1054 38 +254.64 +/- 14.71
ncm-dbt-05 09:52:10 1233732 3330 1389 308 1633 +117.02 +/- 5.21 0 36 546 1049 34 +256.92 +/- 14.54
ncm-dbt-06 09:57:49 1229515 3358 1392 285 1681 +118.98 +/- 5.44 3 48 507 1081 40 +261.82 +/- 15.14
20000 8291 1864 9845 +115.75 +/- 2.19 3 305 3160 6326 206 +253.25 +/- 6.06

Test Detail

ID Host Base NPS Games WLD Standard Elo Ptnml(0-2) Gamepair Elo CLI PGN
191671 ncm-dbt-02 1224074 286 113 21 152 +115.87 +/- 18.19 0 3 49 87 4 +249.3 +/- 49.07
191664 ncm-dbt-05 1238335 330 135 32 163 +112.18 +/- 16.93 0 5 55 102 3 +244.13 +/- 46.42
191663 ncm-dbt-01 1135650 336 144 33 159 +119.25 +/- 16.68 0 4 53 107 4 +261.6 +/- 47.32
191662 ncm-dbt-03 1222510 342 145 35 162 +115.86 +/- 16.52 0 5 54 109 3 +255.15 +/- 46.91
191661 ncm-dbt-04 1230461 348 144 38 166 +109.29 +/- 16.07 0 3 66 101 4 +233.43 +/- 41.81
191660 ncm-dbt-06 1212775 358 149 28 181 +122.24 +/- 15.68 1 4 47 127 0 +289.08 +/- 50.49
191653 ncm-dbt-02 1211356 500 210 42 248 +121.45 +/- 13.49 0 7 72 167 4 +273.0 +/- 40.54
191652 ncm-dbt-05 1229377 500 208 44 248 +118.33 +/- 13.0 0 5 79 163 3 +265.78 +/- 38.54
191651 ncm-dbt-01 1140802 500 201 51 248 +107.54 +/- 14.21 0 10 85 150 5 +230.16 +/- 37.2
191650 ncm-dbt-04 1232167 500 215 39 246 +127.76 +/- 12.65 0 3 72 171 4 +293.29 +/- 40.39
191649 ncm-dbt-03 1221881 500 200 49 251 +108.3 +/- 15.86 0 20 65 159 6 +230.16 +/- 41.63
191648 ncm-dbt-06 1244164 500 209 49 242 +115.22 +/- 14.37 0 9 79 155 7 +247.41 +/- 38.64
191641 ncm-dbt-02 1208571 500 210 55 235 +111.37 +/- 14.22 0 8 86 149 7 +236.51 +/- 36.93
191640 ncm-dbt-01 1127539 500 204 44 252 +115.22 +/- 13.55 0 8 77 162 3 +256.44 +/- 39.15
191639 ncm-dbt-05 1232600 500 215 45 240 +123.02 +/- 12.93 0 2 82 160 6 +273.0 +/- 37.5
191638 ncm-dbt-03 1236533 500 206 54 240 +109.07 +/- 13.73 0 7 89 149 5 +234.38 +/- 36.22
191637 ncm-dbt-04 1216671 500 212 41 247 +123.81 +/- 14.32 0 9 68 166 7 +273.0 +/- 41.7
191636 ncm-dbt-06 1230972 500 203 39 258 +118.33 +/- 14.2 0 10 71 164 5 +261.07 +/- 40.78
191629 ncm-dbt-02 1201030 500 206 47 247 +114.45 +/- 13.89 0 9 77 160 4 +251.89 +/- 39.16
191628 ncm-dbt-05 1241231 500 204 44 252 +115.22 +/- 14.21 0 6 87 148 9 +243.0 +/- 36.62
191627 ncm-dbt-01 1134216 500 212 48 240 +118.33 +/- 14.03 0 7 79 157 7 +256.44 +/- 38.62
191626 ncm-dbt-04 1243234 500 210 64 226 +104.49 +/- 14.35 0 7 99 135 9 +213.85 +/- 34.13
191625 ncm-dbt-03 1234768 500 204 49 247 +111.37 +/- 14.06 0 10 79 157 4 +243.0 +/- 38.64
191624 ncm-dbt-06 1261869 500 206 42 252 +118.33 +/- 12.81 0 3 84 159 4 +263.42 +/- 37.11
191617 ncm-dbt-02 1213681 500 211 45 244 +119.89 +/- 13.33 0 8 70 170 2 +273.0 +/- 41.13
191616 ncm-dbt-05 1223661 500 207 50 243 +112.91 +/- 13.22 0 5 87 154 4 +247.41 +/- 36.56
191615 ncm-dbt-01 1138507 500 203 47 250 +112.14 +/- 14.22 0 11 76 159 4 +245.2 +/- 39.38
191614 ncm-dbt-03 1253475 500 204 48 248 +112.14 +/- 13.22 0 6 85 156 3 +247.41 +/- 37.09
191613 ncm-dbt-04 1223958 500 206 42 252 +118.33 +/- 13.35 0 6 78 162 4 +263.42 +/- 38.85
191612 ncm-dbt-06 1203397 500 204 48 248 +112.14 +/- 15.3 1 10 80 150 9 +236.51 +/- 38.38
191605 ncm-dbt-02 1225993 500 206 47 247 +114.45 +/- 13.89 0 9 77 160 4 +251.89 +/- 39.16
191604 ncm-dbt-05 1221227 500 210 48 242 +116.77 +/- 13.71 0 7 79 159 5 +256.44 +/- 38.62
191603 ncm-dbt-01 1138011 500 186 46 268 +99.95 +/- 13.36 0 9 93 147 1 +217.85 +/- 35.44
191602 ncm-dbt-03 1245403 500 203 50 247 +109.83 +/- 14.37 0 8 89 145 8 +230.16 +/- 36.26
191601 ncm-dbt-04 1208561 500 204 44 252 +115.22 +/- 13.88 0 9 76 161 4 +254.16 +/- 39.42
191600 ncm-dbt-06 1217379 500 209 47 244 +116.77 +/- 14.53 0 9 78 155 8 +249.64 +/- 38.9
191593 ncm-dbt-02 1238384 500 226 52 222 +126.17 +/- 14.62 0 11 61 171 7 +280.42 +/- 43.82
191592 ncm-dbt-05 1249694 500 210 45 245 +119.11 +/- 13.34 0 6 77 163 4 +265.78 +/- 39.12
191591 ncm-dbt-04 1231594 500 216 55 229 +116.0 +/- 14.21 0 9 77 158 6 +251.89 +/- 39.16
191590 ncm-dbt-03 1230352 500 197 37 266 +115.22 +/- 12.67 0 3 87 157 3 +256.44 +/- 36.38
191589 ncm-dbt-01 1123285 500 212 48 240 +118.33 +/- 14.68 0 12 68 164 6 +258.75 +/- 41.54
191588 ncm-dbt-06 1236051 500 212 32 256 +130.94 +/- 13.69 1 3 68 171 7 +298.62 +/- 41.71

Commit

Commit ID 932f5a2d657c846c282adcf2051faef7ca17ae15
Author Linmiao Xu
Date 2023-06-11 13:23:52 UTC
Update default net to nn-ea57bea57e32.nnue Created by retraining an earlier epoch (ep659) of the experiment that led to the first SFNNv6 net: - First retrained on the nn-0dd1cebea573 dataset - Then retrained with skip 20 on a smaller dataset containing unfiltered Leela data - And then retrained again with skip 27 on the nn-0dd1cebea573 dataset The equivalent 7-step training sequence from scratch that led here was: 1. max-epoch 400, lambda 1.0, constant LR 9.75e-4, T79T77-filter-v6-dd.min.binpack ep379 chosen for retraining in step2 2. max-epoch 800, end-lambda 0.75, T60T70wIsRightFarseerT60T74T75T76.binpack ep679 chosen for retraining in step3 3. max-epoch 800, end-lambda 0.75, skip 28, nn-e1fb1ade4432 dataset ep799 chosen for retraining in step4 4. max-epoch 800, end-lambda 0.7, skip 28, nn-e1fb1ade4432 dataset ep759 became nn-8d69132723e2.nnue (first SFNNv6 net) ep659 chosen for retraining in step5 5. max-epoch 800, end-lambda 0.7, skip 28, nn-0dd1cebea573 dataset ep759 chosen for retraining in step6 6. max-epoch 800, end-lambda 0.7, skip 20, leela-dfrc-v2-T77decT78janfebT79aprT80apr.binpack ep639 chosen for retraining in step7 7. max-epoch 800, end-lambda 0.7, skip 27, nn-0dd1cebea573 dataset ep619 became nn-ea57bea57e32.nnue For the last retraining (step7): python3 easy_train.py --experiment-name L1-1536-Re6-masterShuffled-ep639-sk27-Re5-leela-dfrc-v2-T77toT80small-Re4-masterShuffled-ep659-Re3-sameAs-Re2-leela96-dfrc99-16t-v2-T60novdecT80juntonovjanfebT79aprmayT78jantosepT77dec-v6dd-Re1-LeelaFarseer-new-T77T79 \ --training-dataset /data/leela96-dfrc99-T60novdec-v2-T80juntonovjanfebT79aprmayT78jantosepT77dec-v6dd-T80apr.binpack \ --nnue-pytorch-branch linrock/nnue-pytorch/misc-fixes-L1-1536 \ --early-fen-skipping 27 \ --start-lambda 1.0 \ --end-lambda 0.7 \ --max_epoch 800 \ --start-from-engine-test-net False \ --start-from-model /data/L1-1536-Re5-leela-dfrc-v2-T77toT80small-epoch639.nnue \ --lr 4.375e-4 \ --gamma 0.995 \ --tui False \ --seed $RANDOM \ --gpus "0," For preparing the step6 leela-dfrc-v2-T77decT78janfebT79aprT80apr.binpack dataset: python3 interleave_binpacks.py \ leela96-filt-v2.binpack \ dfrc99-16tb7p-eval-filt-v2.binpack \ test77-dec2021-16tb7p.no-db.min-mar2023.binpack \ test78-janfeb2022-16tb7p.no-db.min-mar2023.binpack \ test79-apr2022-16tb7p-filter-v6-dd.binpack \ test80-apr2022-16tb7p.no-db.min-mar2023.binpack \ /data/leela-dfrc-v2-T77decT78janfebT79aprT80apr.binpack The unfiltered Leela data used for the step6 dataset can be found at: https://robotmoon.com/nnue-training-data Local elo at 25k nodes per move: nn-epoch619.nnue : 2.3 +/- 1.9 Passed STC: https://tests.stockfishchess.org/tests/view/6480d43c6e6ce8d9fc6d7cc8 LLR: 2.94 (-2.94,2.94) <0.00,2.00> Total: 40992 W: 11017 L: 10706 D: 19269 Ptnml(0-2): 113, 4400, 11170, 4689, 124 Passed LTC: https://tests.stockfishchess.org/tests/view/648119ac6e6ce8d9fc6d8208 LLR: 2.94 (-2.94,2.94) <0.50,2.50> Total: 129174 W: 35059 L: 34579 D: 59536 Ptnml(0-2): 66, 12548, 38868, 13050, 55 closes https://github.com/official-stockfish/Stockfish/pull/4611 bench: 2370027
Copyright 2011–2024 Next Chess Move LLC