Comments (5)
Mate Test を行った所、昔より判定不足が増えている(ちゃんと探索して調べる方の合法手生成も変わっているので全く同じにになるかはわからないが)
昔(Glicine大会版)
➜ Glicine_submitted git:(master) ✗ ./mate_test -l /Users/ohto/documents/data_research/uecda/log/wisteriax5_50000.dat
passed case test.
using log file [ /Users/ohto/documents/data_research/uecda/log/wisteriax5_50000.dat ]
start reading log...
50000 games were loaded.
judge result (hand) =
447199 528
25957 313960
judge time (hand) = 30.2089
judge time (pw-slow) = 4478.54
check result (hand) =
1625145 1606
3608 551851
check time (hand) = 90.9669
check time (pw-slow) = 1577.8
search result (hand) =
1636888 13589
213171 192646
search time (hand) = 291.79
search time (pw-slow) = 1665.95
passed record move mate judge test.
number of mate moves = {759357, 24535, 8226, 688, 21, 1, 0, 0, 0, 0, 0, 0}
finished analyzing mate moves distribution.
今
➜ FujiGokoroUECda git:(develop) ✗ ./out/release/mate_test -l /Users/ohto/documents/data_research/uecda/log/wisteriax5_50000.dat
passed case test.
using log file [ /Users/ohto/documents/data_research/uecda/log/wisteriax5_50000.dat ]
start reading log...
50000 games were loaded.
1 matches were loaded.
judge result (hand) =
447198 528
26597 313321
judge time (hand) = 33.3491
judge time (pw-slow) = 4981.71
check result (hand) =
1625844 906
5269 550191
check time (hand) = 99.1627
check time (pw-slow) = 1708.98
search result (hand) =
1637852 13733
212266 192443
search time (hand) = 328.407
search time (pw-slow) = 1890.52
passed record move mate judge test.
number of mate moves = {759309, 24542, 8256, 687, 19, 2, 0, 0, 0, 0, 0, 0}
finished analyzing mate moves distribution.
from fujigokorouecda.
L2 Test こちらはほとんど変化なし
昔
➜ Glicine_submitted git:(master) ✗ ./l2_test -l /Users/ohto/documents/data_research/uecda/log/wisteriax5_50000.dat
passed case test.
using log file [ /Users/ohto/documents/data_research/uecda/log/wisteriax5_50000.dat ]
start reading log...
50000 games were loaded.
judge result (hand) =
18547 321 2
2 260 30868
judge time (hand) = 201952
passed record L2 judge test.
今
➜ FujiGokoroUECda git:(develop) ✗ ./out/release/l2_test -l /Users/ohto/documents/data_research/uecda/log/wisteriax5_50000.dat
{ S3 S4 H7 H9 DT HQ SK }(7){ H4 C7 D7 S7 H8 C9 ST }(7) -> 0
{ H4 C7 D7 S7 H8 C9 ST }(7){ S3 S4 H7 H9 DT HQ SK }(7) -> 0
passed case test.
using log file [ /Users/ohto/documents/data_research/uecda/log/wisteriax5_50000.dat ]
start reading log...
50000 games were loaded.
1 matches were loaded.
judge result (hand) =
18549 320 1
2 260 30868
judge time (hand) = 234993
passed record L2 judge test.
from fujigokorouecda.
必勝判定のバグ修正と仕様元に戻す(5枚以上階段を支配とする)によりMate Testの結果がだいたい元に戻った
➜ FujiGokoroUECda git:(develop) ✗ ./out/release/mate_test -l /Users/ohto/documents/data_research/uecda/log/wisteriax5_50000.dat
passed case test.
using log file [ /Users/ohto/documents/data_research/uecda/log/wisteriax5_50000.dat ]
start reading log...
50000 games were loaded.
1 matches were loaded.
judge result (hand) =
447198 528
25958 313960
judge time (hand) = 31.7658
judge time (pw-slow) = 5142.38
check result (hand) =
1625143 1607
3609 551851
check time (hand) = 91.4954
check time (pw-slow) = 1613.77
search result (hand) =
1636948 13540
213170 192636
search time (hand) = 301.459
search time (pw-slow) = 1736.06
passed record move mate judge test.
number of mate moves = {759309, 24542, 8256, 687, 19, 2, 0, 0, 0, 0, 0, 0}
finished analyzing mate moves distribution.
from fujigokorouecda.
必勝判定をいじった結果強さは元に戻ったが、
================ game 100000
daihugou 3.piccoloGlcn
fugo 1.piccoloGlcn
heimin 4.kleinBlauw
himmin 2.piccoloGlcn
daihinmin 0.piccoloGlcn
--------- total point
298035 (2.980) 0.piccoloGlcn
297531 (2.975) 1.piccoloGlcn
299251 (2.993) 2.piccoloGlcn
298546 (2.985) 3.piccoloGlcn
306637 (3.066) 4.kleinBlauw
L2判定の結果かその記録ががおかしくなってしまったのでチェックする
L2 Result : DCL_WIN (17738-93) DCL_LOSE (362-10120) NODCL (514-8887)
Declared Class Miss : ROSE (0) FELL (102)
from fujigokorouecda.
L2判定の結果まわりのバグの原因はFieldAddInfoのどのインスタンスを使うかがおかしくなっていたことだった。他のバグも修正し、とりあえず移植は問題なさそうであった
from fujigokorouecda.
Related Issues (15)
- 世界生成の構造化
- READMEの記述とGPL3ライセンスの矛盾 HOT 2
- 分割ビルドにする HOT 1
- ルール記述部分を一箇所にまとめる
- 人数に対して可変にする
- 少ない試合数で強さを検証する対戦実験モード
- 任意の棋譜から着手決定できるようにする HOT 2
- java版サーバー仕様への対処 HOT 2
- 学習並列化とエポックの定義変更 HOT 4
- 学習時に読み込んだ棋譜でメモリオーバーにならないようにする
- 手札内にある札による役をジョーカーを用いて作成することがどの程度重要か調べる HOT 1
- simd演算による線形パターンマッチ方策計算の高速化 HOT 1
- Policyの計算結果を保存して再利用で高速化するか調べる HOT 5
- パーツの性能評価等test/下の実行ファイル統一すれば? HOT 1
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from fujigokorouecda.