AxiomicLabs/LogicMark

LogicMark A procedurally generated benchmark for evaluating symbolic logic in language models. Each problem presents a set of variable equality/inequality premises and asks the model to identify which conclusion necessarily follows. Unlike knowledge-based benchmarks, LogicMark contains no facts a model could have memorised from pretraining. Every problem is generated fresh from abstract variable names (a, b, c, ...), so a model cannot pattern-match to training data - it must… See the full description on the dataset page: https://huggingface.co/datasets/AxiomicLabs/LogicMark.

<iframe src="https://silytics.si/embed/item/7d61ef00-e98c-4d4a-81a8-8cbdbd07843b" width="420" height="190" style="border:0" loading="lazy" title="Silytics entry"></iframe>

Silytics shows public metadata from Hugging Face and links to the original. Information may be incomplete or out of date. Last refreshed 17 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.