OpenBench

OpenBench Testing Framework

Author	Menkib
Creation	1778787116
Source	https://github.com/Menkib64/lc0
Sha	1e0a7566840760e3ff649c94b9a072b3d554fa15
Branch	uncertainty_weighting_logistic
Bench	1
Network	tf13-375
Options	Threads=2 Hash=1
Time	20.0+0.20
Opening Book	UHO_Lichess_4852_v1.epd
Upload PGNs	FALSE
Syzygy WDL	DISABLED
Syzygy ADJ.	OPTIONAL
Win ADJ.	movecount=3 score=400
Draw ADJ.	movenumber=40 movecount=8 score=10
Scaling	DEV 6250 NPS
Alpha	0.602
Gamma	0.101
A-Ratio	0.1
Reporting	BATCHED
Distribution	SINGLE
Pairs Per Point	8

uncertainty_weighting_lo... vs uncertainty_weighting_lo...

Approve Stop Delete

UncertaintyWeightingMidPoint, float, 0.03, 0.0, 1.0, 0.0045, 0.002
UncertaintyWeightingMinusExponent, float, 114.0, 0.0, 100000.0, 10.0, 0.002
UncertaintyWeightingScale, float, 0.42, 0.0, 10.0, 0.06, 0.002
UncertaintyWeightingBase, float, 0.71, 0.0, 10.0, 0.1, 0.002
UncertaintyWeightingSkewRoot, float, 1.88, 0.01, 10000.0, 0.3, 0.002
UncertaintyWeightingCap, float, 1.0, 0.0, 10.0, 0.15, 0.002

Copy SPSA Inputs

UncertaintyWeightingMidPoint, 0.036653239978303734
UncertaintyWeightingMinusExponent, 115.82138498394087
UncertaintyWeightingScale, 0.47418203143210685
UncertaintyWeightingBase, 0.6668184551030224
UncertaintyWeightingSkewRoot, 1.813089104156836
UncertaintyWeightingCap, 1.2167527469057744

Copy SPSA Outputs

Tuning 6 Parameters
3405/25000 Iterations
54486/400000 Games Played

Name	Curr	Start	Min	Max	C	C_end	R	R_end
UncertaintyWeightingMidPoint	0.0367	0.0300	0.0000	1.0000	0.0055	0.0045	0.0034	0.0020
UncertaintyWeightingMinusExponent	115.8214	114.0000	0.0000	100000.0000	12.2301	10.0000	0.0034	0.0020
UncertaintyWeightingScale	0.4742	0.4200	0.0000	10.0000	0.0734	0.0600	0.0034	0.0020
UncertaintyWeightingBase	0.6668	0.7100	0.0000	10.0000	0.1223	0.1000	0.0034	0.0020
UncertaintyWeightingSkewRoot	1.8131	1.8800	0.0100	10000.0000	0.3669	0.3000	0.0034	0.0020
UncertaintyWeightingCap	1.2168	1.0000	0.0000	10.0000	0.1835	0.1500	0.0034	0.0020

#	User	Updated	Games	Won	Lost	Drawn
415	Kovax	1778893852	1580	358	380	842
1276	Kovax	1778893842	1932	471	473	988
1339	Kovax	1778841774	1040	280	260	500
1268	Kovax	1778893931	1910	477	434	999
1264	Kovax	1778893938	1848	447	447	954
1345	Kovax	1778893919	1942	467	469	1006
1311	Kovax	1778893895	1926	476	449	1001
1307	Kovax	1778893877	1850	449	460	941
1261	Kovax	1778893914	1920	481	505	934
1273	Kovax	1778893897	1940	507	476	957
1252	Kovax	1778893918	1880	473	474	933
1280	Kovax	1778893864	1940	491	466	983
1259	Kovax	1778893941	1870	451	475	944
1279	Kovax	1778893943	1910	467	484	959
1344	Kovax	1778825472	720	182	184	354
1260	Kovax	1778893881	1608	408	395	805
1256	Kovax	1778893850	1472	342	358	772
1312	Kovax	1778893895	1428	327	354	747
1338	Kovax	1778893882	1880	468	473	939
1253	Kovax	1778893921	1952	501	494	957
1342	Kovax	1778893907	1864	444	458	962
1308	Kovax	1778893875	1934	476	477	981
416	Kovax	1778893882	1214	292	269	653
1306	Kovax	1778893839	1472	369	358	745
1310	Kovax	1778841869	608	149	149	310
1267	Kovax	1778893946	740	172	177	391
1309	Kovax	1778893906	742	181	182	379
414	Kovax	1778841779	496	127	117	252
1358	Kovax	1778879381	1020	251	252	517
413	Kovax	1778893950	1238	308	301	629
1313	Kovax	1778842153	624	160	154	310
1361	Kovax	1778879371	1002	248	256	498
1360	Kovax	1778879283	988	237	229	522
1341	Kovax	1778893928	1500	354	341	805
1343	Kovax	1778893912	894	221	229	444
1272	Kovax	1778893872	848	217	207	424
1340	Kovax	1778893861	888	208	224	456
1359	Kovax	1778879308	492	118	129	245
1266	Kovax	1778893913	872	221	224	427
1362	Kovax	1778893835	128	36	34	58
1363	Kovax	1778893886	120	27	24	69
1364	Kovax	1778893866	128	38	36	54
1365	Kovax	1778893894	126	36	37	53