chess-tuning-tools
chess-tuning-tools copied to clipboard
A collection of scripts aimed at efficiently tuning chess engine parameters.
The tuning server and the tuning workers currently communicate directly with the database. This requires trust in the tuning workers, because they need to know the access credentials. To be...
The tuning server reads the minimum samplesize from the configuration file and creates jobs with that samplesize. In practice it can be very useful to create additional jobs with a...
The tuning server of the distributed framework computes the current optimum regularly. It is often interesting to look at the convergence of the parameters over time. That is why these...
Being able to tune against multiple engines would be a nice to have feature, but would require considerable changes to the current script: - Setup of the engine matches -...
* Chess Tuning Tools version: [My fork](https://github.com/Claes1981/chess-tuning-tools) (no opening book, possibility to set normalize_y through option) * Python version: 3.8.7 * Operating System: Linux (OpenSUSE Tumbleweed) ### Description I have...
This is an experiment to play with the idea of validation matches without actually adding "proper" code, and to reason about what it might look like before actually implementing it....
To understand how well the tuning process is proceeding, it might be very useful to have periodic validation matches after N iterations (with a typical N of maybe 100). These...
## Description This pull request changes the data format from compressed NumPy arrays to [HDF5](https://www.hdfgroup.org/solutions/hdf5/). The old format is still supported in that it will be read and converted. ##...
* Chess Tuning Tools version: 0.6.0b2 * Python version: 3.8.3 * Operating System: OpenSUSE Tumbleweed ### Description Experiment with categorical UCI parameter. The tuner tries to set it to a...