What problem does it solve?
This Skill provides a structured way to handle all operations related to the OpenBench platform for the Coda chess engine, including benchmarking, submission, tuning, and result reading, saving users time and preventing common errors.
Core Features & Use Cases
- Benchmark Measurement: Automates the benchmarking process for Coda to ensure consistency with OpenBench.
- SPRT Submission: Simplifies the process of submitting new chess positions to SPRT for testing against Stockfish.
- SPSA Tune Submission: Allows users to submit SPSA tune requests to improve Coda's performance.
- Test Stopping and Status Checks: Offers functionality to stop running tests and check their status, aiding in resource management.
- Error Detection and Analysis: Detects and helps users diagnose common issues such as "Wrong Bench" errors.
- Use Case: Suppose you've developed a new tuning algorithm for the Coda chess engine. Use this Skill to bench it against Stockfish and submit an SPRT test, ensuring your work aligns with OpenBench standards.
Quick Start
Invoke the OB Skill with 'ob Submit <dev_branch> <dev_bench> --base-bench <base_bench>' to submit an SPRT test.