datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
cAI-Prism-B.5-K50
CompactAI-Prism B.5 K50
High-Density Distillation Dataset for Small Model English Language Acquisition
License: MITTop-K: 50 (Current release: K50)Source Model: Qwen3.5 2BPrimary Objective: Teach small-scale AI models to generate fluent, coherent English text through probability-aware distillation. Or at least help them sound less like they learned English from a fortune cookie.
Overview
CompactAI-Prism is a specialized training dataset designed… See the full description on the dataset page: https://huggingface.co/datasets/Glint-Research/cAI-Prism-B.5-K50.prism-bench
PRISM-Bench: Measuring Value, Evidence, and Source Hierarchies in Frontier AI Systems
Anonymous submission to NeurIPS 2026 Evaluations & Datasets Track.
License: CC BY 4.0 | Croissant: included with full RAI fields
Dataset Summary
PRISM-Bench is the first multi-model forced-choice benchmark measuring the upper three layers of the Authority Stack model:
Value (V) — which value priorities guided the decision (Schwartz 10-value framework)
Evidence (E) — which evidence type… See the full description on the dataset page: https://huggingface.co/datasets/enoch6101/prism-bench.PRISM-BENCH
PRISM-BENCH: Towards Benchmarking Multimodal Reasoning Beyond Accuracy
Dataset Description
PRISM-BENCH is a benchmark of 1,000 expert-curated questions spanning five complementary and underrepresented multimodal reasoning categories. Each instance is accompanied by a ground-truth answer, a human-authored idealized reasoning trace, and multi-label reasoning tags, enabling evaluation of both final-answer correctness and intermediate reasoning quality.
The benchmark is… See the full description on the dataset page: https://huggingface.co/datasets/vanshbhatia/PRISM-BENCH.