A 1,400 row science reasoning dataset. Each row contains a GLM 5.1 chain of thought paired with three structured auxiliary fields produced by a separate annotator pass.
Question str science problem statement
Answer str GLM 5.1 final answer
Reasoning str GLM 5.1 full chain of thought
Difficulty int 3 medium, 4 hard, 5 extreme
Summary str condensed chronological retelling of Reasoning… See the full description on the dataset page:
https://huggingface.co/datasets/WilhelmH/Defined-Reasoning.
