CloudyGo
Other Runs
v17 - 20x256 Squeeze-and-Excitation
v16 - 40x256 40 block!
v15 - 20x256 Q=loss
v14 - 20x256 Bigtable + Q=loss
v13 - 20x256 "Master" (SL)
v12 - 20x256 BS=2
v11 - 20x256 Q=draw
v10 - 20x256
v9 - 20x256
v7 - 20x128
v5 - 10x128
v3-9x9
leela-zero Official LZ games and eval
leela-zero-eval More eval games
leela-zero-eval-time Eval on even time
cross-eval (eval only)
KataGo KataGo games
Pages
Model list
Graphs
Model Evolution
Josekis
Figure 3
Evaluation
Model Details
Model Graphs
Newest Eval
Results
Puzzles
All Eval Graph
v3-9x9
Graphs
V3-9x9 Graphs
RESULTS
200
220
240
260
280
300
320
340
360
380
400
420
440
460
480
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
0.40
0.45
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
0.40
0.45
Model Number
Winrate
Winrate by Model
200
220
240
260
280
300
320
340
360
380
400
420
440
460
480
10
15
20
25
30
35
40
45
50
55
60
65
10
15
20
25
30
35
40
45
50
55
60
65
Model Number
Game length
Game length by Model
This aren't that useful and we have live data from logs now
200
220
240
260
280
300
320
340
360
380
400
420
440
460
480
0.000
0.001
0.002
0.003
0.004
0.005
0.006
0.007
0.008
0.009
0.010
Model Number
Bad resign rate
Bad Resign Rate by Model
466
468
470
472
474
476
478
480
482
484
486
488
490
492
494
496
0.50
0.55
0.60
0.65
0.70
0.75
0.80
0.50
0.55
0.60
0.65
0.70
0.75
0.80
Model Number
Resign rate for X% error
Threshold for 1%, 2% and 5% Bad Resign Rate
200
220
240
260
280
300
320
340
360
380
400
420
440
460
480
7,500
8,000
8,500
9,000
9,500
10,000
10,500
11,000
11,500
12,000
12,500
13,000
13,500
Model Number
Number of self-play games CloudyGo has ingested
Number of games (processed) by Model
Wed 07
Fri 09
Feb 11
Tue 13
Thu 15
Sat 17
Mon 19
Wed 21
Fri 23
Feb 25
Tue 27
March
Sat 03
0
20,000
40,000
60,000
80,000
100,000
120,000
140,000
160,000
180,000
200,000
220,000
240,000
0
20,000
40,000
60,000
80,000
100,000
120,000
140,000
160,000
180,000
200,000
220,000
240,000
Day
Number of Games
Games Per Day
200
220
240
260
280
300
320
340
360
380
400
420
440
460
480
0
10,000
20,000
30,000
40,000
50,000
60,000
70,000
80,000
90,000
100,000
110,000
0
10,000
20,000
30,000
40,000
50,000
60,000
70,000
80,000
90,000
100,000
110,000
Model Number
Sum visits
Sum of visits to top move per game by Model
200
220
240
260
280
300
320
340
360
380
400
420
440
460
480
1.8
1.9
2.0
2.1
2.2
2.3
2.4
2.5
Model Number
Soft-n of all better not-played moves (see early game temperature)
How badly soft-n messed up each game by Model
160
180
200
220
240
260
280
300
320
340
360
380
400
420
440
-400
-300
-200
-100
0
100
200
300
-400
-300
-200
-100
0
100
200
300
Model Number
Delta Rating
Rating delta from last model
-400
-300
-200
-100
0
100
200
300
400
0
10
20
30
40
50
60
70
80
90
100
Model number delta
elo delta
Win rate by difference in model number