Computer Chess Club Archives

Search

Terms

Messages

Subject: Offer (eval testing)

Author: Sergei S. Markoff

Date: 22:31:37 09/26/04

In new SmarThink I implemented such a command:

epdevl <f> ................ Produces static evaluations for EPD-file <f>

The idea is to compare our engines on special testsets. It can help all
programmers to debug their eval.

I see, that some programmers keeping ther eval functions in secret. For example,
Ruffian produces first score only after big time interval. Anyway, I think that
there are a lot of people interested in some common research.

You can compare your eval and find positions for which your eval is
significantly different to other engines eval.

Example of using command for
http://www.jakob.at/steffen/hossa/testsuites/wac.epd:

========== begin ============
SmarThink v0.18a[lpha]
Error opening "Book.smt"!
Transposition size is 524288 entries (10485760 bytes)
Pawn hash size is 262144 entries (5767168 bytes)
EGTBHashSize=1048576 bytes
Aggressivity=100%
Cowardice=100%
ResignValue=800
ResignTimes=3
UseEGTB=on
EGTB path="c:\tablebases\"
UseBookWeights=on
UseBookLearning=on
BookLearningMode=KILL
VarietyOfPlay=10
BookLearningBound=400
UseLearning=off
UseLearningData=off
LogLearningIntoFile=off
Pondering=on

EGTB files not found

> epdevl wac.epd

-153
88
-71
18
-493
-122
-81
408
-422
63
20
-575
-104
57
353
-17
-7
77
-167
237
186
-27
111
15
-35
-170
-470
-40
85
-54
89
-12
200
93
162
-111
-113
-414
-39
24
621
54
47
-176
-70
-1
-136
-25
-14
-640
-298
300
-150
206
-40
14
-225
3
-6
-422
-76
-316
182
194
39
178
-221
203
148
-13
-145
86
123
404
-384
33
280
221
-45
37
252
-27
57
-330
323
32
235
-159
502
60
70
-27
-5
338
-303
132
282
38
54
383
56
43
-192
-33
354
3
-73
-133
49
-44
73
485
-199
-68
-98
58
62
-56
-38
238
-69
277
-183
-35
4
-233
128
-447
102
-96
-196
308
105
-185
-80
216
30
-256
-218
150
-253
-50
66
67
30
175
-29
-14
17
236
6
13
128
383
13
-241
-212
43
57
358
152
-233
-15
175
-137
-40
183
312
-7
-124
-73
221
-112
121
-243
-88
-369
-137
125
16
366
68
24
54
-192
-121
-222
-246
272
-225
204
-415
28
191
-94
-123
-640
-113
15
154
71
34
23
-289
138
173
-257
55
172
-261
178
-220
-555
20
-429
69
-167
-58
-457
171
330
75
-397
117
310
-291
12
38
184
221
17
-44
200
59
166
-19
-67
-72
-91
-171
138
-78
236
70
312
154
254
-156
5
-68
302
-176
-202
-20
18
119
41
-62
-123
272
120
-104
147
-162
18
-197
45
-5
54
28
-100
124
6
110
82
-241
-220
97
-10
70
375
-168
-813
205
-871
-387
340
53
120
140
16
-259
-20
323
-232
57
-4
179
201
-120
339
=========== end =============

Re: Offer (eval testing) Peter Fendrich 07:42:27 09/29/04
CRAFTY David B Weller 16:00:15 09/27/04
Re: Offer (eval testing) Sune Fischer 15:36:24 09/27/04
WACNEW by GES David B Weller 14:08:45 09/27/04
Re: Offer (eval testing) GeoffW 12:19:29 09/27/04
Re: Offer (eval testing) Robert Hyatt 09:57:25 09/27/04
- Re: Offer (eval testing) GeoffW 12:37:05 09/27/04
  - eval testing (Quick graph) GeoffW 15:34:09 09/27/04
    - says no access n/t David B Weller 15:54:48 09/27/04
    - (Quick graph) 2nd Attempt GeoffW 15:54:22 09/27/04
      - Re: (Quick graph) 2nd Attempt Rick Bischoff 21:44:46 09/27/04
      - very neat! n/t David B Weller 16:01:16 09/27/04
  - Re: Offer (eval testing) David B Weller 15:05:25 09/27/04
  - Re: Offer (eval testing) Robert Hyatt 14:02:04 09/27/04
Re: Offer (eval testing) Bas Hamstra 23:54:44 09/26/04
Re: Offer (eval testing) Bas Hamstra 23:53:51 09/26/04
Re: Offer (eval testing) Jeremiah Penery 22:46:25 09/26/04

This page took 0.01 seconds to execute

Last modified: Thu, 15 Apr 21 08:11:13 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.