SFT data that teaches a small open model the atomic sub-skills of word games - especially Wordle - as a curriculum to run before reinforcement learning. Each game isolates one skill (map a word to its letters, track used/absent letters, judge validity & meaning, reason over letter multisets, rhyme, fill a pattern, and read feedback and adjust), so RL can later compose skills the model already owns. Most data is generated programmatically (exact, zero API… See the full description on the dataset page:
https://huggingface.co/datasets/saketh-chervu/word-games-distillation.
