**SPRT-Matches** of 384 nets
**LC0 version:** v0.26.3
**LC0 options:** Threads=1, MinibatchSize=64, MaxPrefetch=32, MaxCollisionEvents=32, MaxOutOfOrderEvalsFactor=2.0, Backend=cuda-fp16, ScoreType=centipawn_2019, MoveOverheadMs=0,  SmartPruningMinimumBatches=2, RootHasOwnCpuctParams=false, CPuct=1.350, FpuValue=0.228, PolicyTemperature=1.497, CPuctFactor=3.468, CPuctBase=20253, MovesLeftMaxEffect=0.2, MovesLeftThreshold=0.0, MovesLeftSlope=0.007, MovesLeftScaledFactor=1.0, MovesLeftQuadraticFactor=0.0, MovesLeftConstantFactor=0.0
**Hardware:** Ryzen 9 3950X (3.5GHz) + RTX 3080
**Time control:** 1.5s/game+0.025s/move
**Speed:** ≈1000 nodes/move
**Book:** 3-moves unbalanced book, order=random
**Tablebases:** 6-man
**Adjudication:** 6-man TBs + -resign movecount=3 score=500, -draw movenumber=20 movecount=5 score=10
**Software:** Cutechess-CLI, time forfeits disabled
```diff
   # ENGINE            :  RATING  ERROR  CFS(%)   GAMES     W     D     L  DRAWS(%)
+  1 lc0.net.J94-80    :     7.3    5.6    99.4    5680  1142  3507  1031      61.7
   2 lc0.net.J94-40    :     0.0   ----     ---    5680  1031  3507  1142      61.7
White advantage = 86.5 +/- 2.9
Draw rate (equal opponents) = 67.6 % +/- 0.7```BayesElo bounds [0.0;5.0] scaled to logistic bounds based on drawrate:
`sprt.py --elo0 0.0 --elo1 3.094 --results 26 559 1579 630 46`:
```
[Elo0,Elo1]                 :  {0.000,3.094}
Elo                         :  6.3+/-4.7
Confidence interval         :  [1.5,11.0] (95.0%)
LLR [u,l]                   :  2.944  [-2.944,2.944]```
**Comment:** Latest tuned values updated.
**LC0 options:** CPuct=1.745, FpuValue=0.330, PolicyTemperature=1.359, CPuctFactor=3.894, CPuctBase=38739
```diff
   # ENGINE            :  RATING  ERROR  CFS(%)   GAMES    W     D    L  DRAWS(%)
   1 lc0.net.J94-80    :     0.0   ----    99.3    2642  549  1617  476      61.2
-  2 lc0.net.J96-28    :   -10.4    8.2     ---    2642  476  1617  549      61.2
White advantage = 91.4 +/- 4.2
Draw rate (equal opponents) = 67.7 % +/- 1.1```BayesElo bounds [0.0;5.0] scaled to logistic bounds based on drawrate:
`sprt.py --elo0 0.0 --elo1 3.127 --results 10 326 729 239 17`:
```
[Elo0,Elo1]                 :  {0.000,3.127}
Elo                         :  -9.1+/-6.8
Confidence interval         :  [-15.9,-2.3] (95.0%)
LLR [u,l]                   :  -2.944  [-2.944,2.944]```
```diff
   # ENGINE            :  RATING  ERROR  CFS(%)   GAMES    W     D    L  DRAWS(%)
   1 lc0.net.J94-80    :     0.0   ----    99.0    3030  624  1855  551      61.2
-  2 lc0.net.J96-44    :    -8.9    7.5     ---    3030  551  1855  624      61.2
White advantage = 84.6 +/- 3.8
Draw rate (equal opponents) = 66.6 % +/- 1.0```BayesElo bounds [0.0;5.0] scaled to logistic bounds based on drawrate:
`sprt.py --elo0 0.0 --elo1 3.126 --results 25 347 832 298 13`:
```
[Elo0,Elo1]                 :  {0.000,3.126}
Elo                         :  -7.8+/-6.5
Confidence interval         :  [-14.3,-1.4] (95.0%)
LLR [u,l]                   :  -2.944  [-2.944,2.944]```
```diff
   # ENGINE            :  RATING  ERROR  CFS(%)   GAMES     W     D     L  DRAWS(%)
   1 lc0.net.J94-80    :     0.0   ----    81.8    8902  1727  5497  1678      61.8
-  2 lc0.net.66615     :    -2.0    4.4     ---    8902  1678  5497  1727      61.8
White advantage = 83.8 +/- 2.3
Draw rate (equal opponents) = 67.2 % +/- 0.6```BayesElo bounds [0.0;5.0] scaled to logistic bounds based on drawrate:
`sprt.py --elo0 0.0 --elo1 3.093 --results 68 952 2448 927 56`:
```
[Elo0,Elo1]                 :  {0.000,3.093}
Elo                         :  -1.4+/-3.9
Confidence interval         :  [-5.2,2.5] (95.0%)
LLR [u,l]                   :  -2.944  [-2.944,2.944]```

Hier die Grafik:

https://cdn.discordapp.com/attachments/530486338236055583/787961838050213908/networks.png

Lothar