LondonFrau Admin
Posts : 1315Reputation : 3534Join date : 2010-02-27Location : ??? Subject: Latest abrok Development Version August 20th (SF The gift that keeps on giving Sat Aug 24, 2024 12:15 pm !! latest version !! Windows x64 for Haswell CPUs Windows x64 for modern computers + AVX2 Windows x64 for modern computers Windows x64 + SSSE3 Windows x64 Linux x64 for Haswell CPUs Linux x64 for modern computers + AVX2 Linux x64 for modern computers Linux x64 + SSSE3 Linux x64 Author: Taras Vuk Date: Tue Aug 20 21:37:25 2024 +0200 Timestamp: 1724182645 Tweak late move extensions Allow late move extensions only for PV and cut nodes. Passed STC: LLR: 2.95 (-2.94,2.94) <0.00,2.00> Total: 44512 W: 11688 L: 11355 D: 21469 Elo +2.60 Ptnml(0-2): 167, 5180, 11229, 5513, 167 https://tests.stockfishchess.org/tests/view/66c0509d4ff211be9d4ef10e Passed LTC: LLR: 2.94 (-2.94,2.94) <0.50,2.50> Total: 152970 W: 39026 L: 38466 D: 75478 Elo +1.27 Ptnml(0-2): 102, 16792, 42164, 17298, 129 https://tests.stockfishchess.org/tests/view/66c0994d21503a509c13b2b6 closes https://github.com/official-stockfish/Stockfish/pull/5541 bench: 1484730 see source Windows x64 for Haswell CPUs Windows x64 for modern computers + AVX2 Windows x64 for modern computers Windows x64 + SSSE3 Linux x64 for Haswell CPUs Linux x64 for modern computers + AVX2 Linux x64 for modern computers Linux x64 + SSSE3 Linux x64 Author: Shawn Xu Date: Tue Aug 20 21:34:19 2024 +0200 Timestamp: 1724182459 Introduce Fail Low History Bonus When a node fails low, give TT move a small bonus 1/4 of normal value. Passed STC: LLR: 2.95 (-2.94,2.94) <0.00,2.00> Total: 92384 W: 24094 L: 23691 D: 44599 Elo +1.52 Ptnml(0-2): 323, 10852, 23465, 11203, 349 https://tests.stockfishchess.org/tests/view/66be80794ff211be9d4eed68 Passed LTC: LLR: 2.94 (-2.94,2.94) <0.50,2.50> Total: 114660 W: 29260 L: 28778 D: 56622 Elo +1.46 Ptnml(0-2): 97, 12506, 31653, 12966, 108 https://tests.stockfishchess.org/tests/view/66bf63ee4ff211be9d4eeef0 closes https://github.com/official-stockfish/Stockfish/pull/5539 bench 1463003 see source Windows x64 for Haswell CPUs Windows x64 + SSSE3 Windows x64 Linux x64 for Haswell CPUs Linux x64 for modern computers + AVX2 Linux x64 for modern computers Linux x64 + SSSE3 Linux x64 Author: Nonlinear2 Date: Tue Aug 20 21:31:33 2024 +0200 Timestamp: 1724182293 Simplify stand pat adjustement Remove && !PvNode condition for stand pat adjustement in quiescence search. Passed non-regression STC: LLR: 2.94 (-2.94,2.94) <-1.75,0.25> Total: 108544 W: 28228 L: 28085 D: 52231 Elo +0.46 Ptnml(0-2): 389, 12902, 27554, 13031, 396 https://tests.stockfishchess.org/tests/view/66bb402e4ff211be9d4ee688 Passed non-regression LTC: LLR: 2.94 (-2.94,2.94) <-1.75,0.25> Total: 193014 W: 48796 L: 48751 D: 95467 Elo +0.08 Ptnml(0-2): 188, 21481, 53116, 21542, 180 https://tests.stockfishchess.org/tests/view/66bc78774ff211be9d4ee88f closes https://github.com/official-stockfish/Stockfish/pull/5538 Bench 1787360 see source Windows x64 for Haswell CPUs Windows x64 for modern computers + AVX2 Windows x64 for modern computers Windows x64 + SSSE3 Windows x64 Linux x64 for Haswell CPUs Linux x64 for modern computers + AVX2 Linux x64 for modern computers Linux x64 + SSSE3 Linux x64 Author: Shawn Xu Date: Tue Aug 20 21:27:58 2024 +0200 Timestamp: 1724182078 Simplify doShallowerSearch Passed Non-regression STC: LLR: 2.94 (-2.94,2.94) <-1.75,0.25> Total: 242336 W: 62657 L: 62663 D: 117016 Elo -0.01 Ptnml(0-2): 941, 28949, 61418, 28895, 965 https://tests.stockfishchess.org/tests/view/66bc13c34ff211be9d4ee794 Passed Non-regression LTC: LLR: 2.94 (-2.94,2.94) <-1.75,0.25> Total: 128100 W: 32503 L: 32390 D: 63207 Elo +0.31 Ptnml(0-2): 106, 14319, 35113, 14380, 132 https://tests.stockfishchess.org/tests/view/66bdbb304ff211be9d4eec5d closes https://github.com/official-stockfish/Stockfish/pull/5537 bench 1586246 see source Windows x64 for Haswell CPUs Windows x64 for modern computers + AVX2 Windows x64 for modern computers Windows x64 Linux x64 for Haswell CPUs Linux x64 for modern computers + AVX2 Linux x64 for modern computers Linux x64 + SSSE3 Linux x64 Author: Shawn Xu Date: Tue Aug 20 21:24:04 2024 +0200 Timestamp: 1724181844 Simplify bestMove promotion Passed Non-regression STC: LLR: 2.95 (-2.94,2.94) <-1.75,0.25> Total: 216768 W: 56240 L: 56217 D: 104311 Elo +0.04 Ptnml(0-2): 794, 24900, 56956, 24957, 777 https://tests.stockfishchess.org/tests/view/66bc11324ff211be9d4ee78b Passed Non-regression LTC: LLR: 2.94 (-2.94,2.94) <-1.75,0.25> Total: 44970 W: 11391 L: 11199 D: 22380 Elo +1.48 Ptnml(0-2): 44, 4596, 13002, 4810, 33 https://tests.stockfishchess.org/tests/view/66bdbb1b4ff211be9d4eec5a closes https://github.com/official-stockfish/Stockfish/pull/5535 bench: 1613043 see source Windows x64 for Haswell CPUs Windows x64 for modern computers + AVX2 Windows x64 for modern computers Windows x64 + SSSE3 Windows x64 Linux x64 for Haswell CPUs Linux x64 for modern computers + AVX2 Linux x64 for modern computers Linux x64 + SSSE3 Linux x64 Author: Linmiao Xu Date: Tue Aug 20 20:59:36 2024 +0200 Timestamp: 1724180376 Update default main net to nn-1111cefa1111.nnue Created from 2 distinct spsa tunes of the latest main net (nn-31337bea577c.nnue) and applying the params to the prior main net (nn-e8bac1c07a5a.nnue). This effectively reverts the modifications to output weights and biases in https://github.com/official-stockfish/Stockfish/pull/5509 SPSA: A: 6000, alpha: 0.602, gamma: 0.101 1st - 437 feature transformer biases where values are < 25 54k / 120k games at 180+1.8 https://tests.stockfishchess.org/tests/view/66af98ac4ff211be9d4edad0 nn-808259761cca.nnue 2nd - 208 L2 weights where values are zero 112k / 120k games at 180+1.8 https://tests.stockfishchess.org/tests/view/66b0c3074ff211be9d4edbe5 nn-a56cb8c3d477.nnue When creating the above 2 nets (nn-808259761cca.nnue, nn-a56cb8c3d477.nnue), spsa params were unintentionally applied to nn-e8bac1c07a5a.nnue rather than nn-31337bea577c.nnue due to an issue in a script that creates nets by applying spsa results to base nets. Since they both passed STC and were neutral or slightly positive at LTC, they were combined to see if the elo from each set of params was additive. The 2 nets can be merged on top of nn-e8bac1c07a5a.nnue with: https://github.com/linrock/nnue-tools/blob/90942d3/spsa/combine_nnue.py ``` python3 combine_nnue.py \ nn-e8bac1c07a5a.nnue \ nn-808259761cca.nnue \ nn-a56cb8c3d477.nnue ``` Merging yields nn-87caa003fc6a.nnue which was renamed to nn-1111cefa1111.nnue with an updated nnue-namer around 10x faster than before by: - using a prefix trie for efficient prefix matches - modifying 4 non-functional bytes near the end of the file instead of 2 https://github.com/linrock/nnue-namer Thanks to @MinetaS for pointing out in #nnue-dev what the non-functional bytes are: L3 is 32, 4 bytes for biases, 32 bytes for weights. (fc_2) So -38 and -37 are technically -2 and -1 of fc_1 (type AffineTransform<30, 32>) And since InputDimension is padded to 32 there are total 32 of 2 adjacent bytes padding. So yes, it's non-functional whatever values are there. It's possible to tweak bytes at -38 - 32 * N and -37 - 32 * N given N = 0 ... 31 The net renamed with the new method passed non-regression STC vs. the original net: https://tests.stockfishchess.org/tests/view/66c0f0a821503a509c13b332 To print the spsa params with nnue-pytorch: ``` import features from serialize import NNUEReader feature_set = features.get_feature_set_from_name("HalfKAv2_hm") with open("nn-31337bea577c.nnue", "rb") as f: model = NNUEReader(f, feature_set).model c_end = 16 for i,ft_bias in enumerate(model.input.bias.data[:3072]): value = int(ft_bias * 254) if abs(value) < 25: print(f"ftB[{i}],{value},-1024,1024,{c_end},0.0020") c_end = 6 for i in range(8): for j in range(32): for k in range(30): value = int(model.layer_stacks.l2.weight.data[32 * i + j, k] * 64) if value == 0: print(f"twoW[{i}][{j}][{k}],{value},-127,127,{c_end},0.0020") ``` New params found with the same method as: https://github.com/official-stockfish/Stockfish/pull/5459 Passed STC: https://tests.stockfishchess.org/tests/view/66b4d4464ff211be9d4edf6e LLR: 2.94 (-2.94,2.94) <0.00,2.00> Total: 136416 W: 35753 L: 35283 D: 65380 Elo +1.20 Ptnml(0-2): 510, 16159, 34416, 16597, 526 Passed LTC: https://tests.stockfishchess.org/tests/view/66b76e814ff211be9d4ee1cc LLR: 2.95 (-2.94,2.94) <0.50,2.50> Total: 159336 W: 40753 L: 40178 D: 78405 Elo +1.25 Ptnml(0-2): 126, 17497, 43864, 18038, 143 closes https://github.com/official-stockfish/Stockfish/pull/5534 bench 1613043 see source
_________________ Bettina...............The greatest happiness of life is the conviction that we are loved. Victor Hugo
vv56 likes this post
Like 1 Dislike