This dataset contains natural language game scenarios and optimal decisions computed by solvers in No Limit Texas Hold’em. It is divided into pre-flop and post-flop datasets, each with training and test splits. The data is stored in both .json and .csv formats:
JSON files: Contain the natural language prompts (instruction) and optimal decisions (output) derived from the game scenarios.
CSV files: Contain structured game information from which the JSON files… See the full description on the dataset page:
https://huggingface.co/datasets/WhiteWatson/PokerBench.
