Xiangqi Gameplay Trajectories from Pikafish Engine Training

1,000 high-quality Chinese Chess matches extracted from the training pipeline of the Pikafish engine. The dataset provides structured, step-by-step action sequences in JSONL format. It was created by ysong18 and last updated on 2026-06-07.

Use Cases

• Train reinforcement learning agents based on the structured, step-by-step action sequences.
• Benchmark game-playing AI performance against trajectories from a leading neural-network-backed engine.
• Analyze Xiangqi game strategy and move patterns from the high-quality match data.

Strengths

• Contains 1,000 high-quality Chinese Chess matches.
• Data originates from the training pipeline of Pikafish, a leading Xiangqi engine.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.

Provenance

Source

Processed from the open-source training pipeline of Pikafish; original source credited to the px0data dataset on Kaggle.

Collection Method

Extracted and processed from engine training trajectories.

Freshness

Last updated 2026-06-07 05:55:08; freshness should be verified.