Here a description of the things I did :
I first downloaded latest cutechess and stockfish from github and clop from Remis site and compiled .I compiled clop using python2.7 instead of python 2.6 but that should not be a problem and later compiled cutechess. Then I went into stockfish's ucioption.cpp and changed ponder and own book to false and increased hash 128 and compiled it.
I configured clop-cutechess-cli.py according to https://github.com/mcostalba/Stockfish/ ... a2c669a9db
Code: Select all
#!/usr/bin/python
# -*- coding: utf-8 -*-
"""
Usage: clop-cutechess-cli.py CPU_ID SEED [PARAM_NAME PARAM_VALUE]...
Run cutechess-cli with CLOP_PARAM(s).
CPU_ID Symbolic name of the CPU or machine that should run the game
SEED Running number for the game to be played
PARAM_NAME Name of a parameter that's being optimized
PARAM_VALUE Integer value for parameter PARAM_NAME
CLOP is a black-box parameter tuning tool designed and written by Rémi Coulom.
More information about CLOP can be found at the CLOP website:
http://remi.coulom.free.fr/CLOP/
This script works between CLOP and cutechess-cli. The path to this script,
without any parameters, should be on the "Script" line of the .clop file.
'Replications' in the .clop file should be set to 2 so that this script can
alternate the engine's playing side correctly.
In this script the variables 'cutechess_cli_path', 'engine', 'engine_param_cmd',
'opponents' and 'options' must be modified to fit the test environment and
conditions. The default values are just examples.
When the game is completed the script writes the game outcome to its
standard output:
W = win
L = loss
D = draw
"""
from subprocess import Popen, PIPE
import sys
# Path to the cutechess-cli executable.
# On Windows this should point to cutechess-cli.exe
cutechess_cli_path = './home/quantum/Downloads/Testing/cutechess/projects/cli/cutechess-cli'
# The engine whose parameters will be optimized
engine = 'cmd=stockfish'
# Format for the commands that are sent to the engine to
# set the parameter values. When the command is sent,
# {name} will be replaced with the parameter name and {value}
# with the parameter value.
engine_param_cmd = 'setoption name {name} value {value}'
# A pool of opponents for the engine. The opponent will be
# chosen based on the seed sent by CLOP.
opponents = [
'cmd=stockfish_master'
]
# Additional cutechess-cli options, eg. time control and opening book
options = '-each tc=0:1+0.05 proto=uci restart=on -openings file=book5.epd format=epd order=random start=1 -recover -resign movecount=5 score=500'
def main(argv = None):
if argv is None:
argv = sys.argv[1:]
if len(argv) == 0 or argv[0] == '--help':
sys.stdout.write(__doc__)
return 0
argv = argv[1:]
if len(argv) < 3 or len(argv) % 2 == 0:
sys.stderr.write('Too few arguments\n')
return 2
clop_seed = 0
try:
clop_seed = int(argv[0])
except ValueError:
sys.stderr.write('invalid seed value: %s\n' % argv[0])
return 2
fcp = engine
scp = opponents[(clop_seed >> 1) % len(opponents)]
# Parse the parameters that should be optimized
for i in range(1, len(argv), 2):
# Make sure the parameter value is numeric
try:
float(argv[i + 1])
except ValueError:
sys.stderr.write('invalid value for parameter %s: %s\n' % (argv[i], argv[i + 1]))
return 2
# Pass CLOP's parameters to the engine by using
# cutechess-cli's initialization string feature
initstr = engine_param_cmd.format(name = argv[i], value = argv[i + 1])
fcp += ' initstr="%s"' % initstr
# Choose the engine's playing side (color) based on CLOP's seed
if clop_seed % 2 != 0:
fcp, scp = scp, fcp
cutechess_args = '-engine %s -engine %s %s' % (fcp, scp, options)
command = '%s %s' % (cutechess_cli_path, cutechess_args)
# Run cutechess-cli and wait for it to finish
process = Popen(command, shell = True, stdout = PIPE)
output = process.communicate()[0]
if process.returncode != 0:
sys.stderr.write('failed to execute command: %s\n' % command)
return 2
# Convert Cutechess-cli's result into W/L/D
# Note that only one game should be played
result = -1
for line in output.decode("utf-8").splitlines():
if line.startswith('Finished game'):
if line.find(": 1-0") != -1:
result = clop_seed % 2
elif line.find(": 0-1") != -1:
result = (clop_seed % 2) ^ 1
elif line.find(": 1/2-1/2") != -1:
result = 2
else:
sys.stderr.write('the game did not terminate properly\n')
return 2
break
if result == 0:
sys.stdout.write('W\n')
elif result == 1:
sys.stdout.write('L\n')
elif result == 2:
sys.stdout.write('D\n')
if __name__ == "__main__":
sys.exit(main())Code: Select all
#
# DummyExperiment.clop
#
# Example of experiment definition
#
# Name (used for .log and .dat files)
Name DummyExperiment
# Script for running a game. See DummyScript.py for details.
Script ./home/quantum/Downloads/Testing/cutechess/tools/clop-cutechess-cli.py
# Parameter(s) to be optimized
# <parameter_type> <name> <min> <max>
# <parameter_type> may be:
# LinearParameter
# IntegerParameter
# GammaParameter
# IntegerGammaParameter
# For GammaParameter, quadratic regression is performed on log(x)
# Warning: 123 and not 123.0 should be used for IntegerParameter
IntegerParameter Space 50 150
IntegerParameter Aggressiveness 50 150
# This could be the list of machine names in a distributed experiment.
# In order to run 4 games in parallel, 2 on machine1, 2 on machine2:
Processor machine1
# Call the script "Replications" times in a row with identical parameter values
# Replications may be used to alternate colors, for instance.
# Seed % Replications would indicate color.
Replications 2
# Parameters of statistical model of outcome
# For binary outcome (Win/Loss, no draws), use "DrawElo 0"
# For chess, use "DrawElo 100"
DrawElo 100
# Regression parameters
# H 3 is recommended (it is the default value)
# Correlations may be "all" (default) or "none"
# Even if variables are not correlated "all" should work well. The problem is
# that the regression might become very costly if the number of variables is
# high. So use "Correlations none" only if you are certain parameters are
# independent or you have so many variables that "all" is too costly.
H 3
Correlations all
I had to download the 0.6 version from github because i could not find it into their current git repository