ianncity/

GLM-5.2-Logic-Puzzles

6000x traces distilled from GLM-5.2 on High reasoning

	Token Count: 5M~?







	


	Distribution:


Puzzles:

•Tokenization blindless ex: counting the r's in strawberry •Goal reasoning ex: the car wash test (theres no car wash question exactly just prompts like it so its not just benchmaxxing) •Reading comprehension traps •Temporal reasoning •Many other categories not worth mentioning

Prompts… See the full description on the dataset page: https://huggingface.co/datasets/ianncity/GLM-5.2-Logic-Puzzles.