datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
realistic_scen
Unreal MLLM Dataset - realistic_scen
Physics simulation dataset with unreal rules for multimodal language model evaluation.
Dataset Structure
Each row contains:
Video file with physics simulation
Plan and metadata as JSON strings
Multiple QA items (Rule Identification, Explanatory Reasoning, Predictive)
Optional prediction video
Features
features:
- name: difficulty
dtype: string
- name: file_name
dtype: video
- name: id
dtype: string
-… See the full description on the dataset page: https://huggingface.co/datasets/UnrealMLLM/realistic_scen.realistic-niah-count-empirical-law
Realistic NIAH Count Empirical Law V3.1
This private dataset is the frozen stimulus bank for the preregistered V3.1
realistic needle-in-a-haystack counting experiment.
Registered grid
Passage lengths: 1K, 2K, 3K, 5K, 8K, 10K, 15K, and 20K canonical tokens.
Needle counts: 1-10, 12, 15, 18, and 20.
Paired seeds: 1234-1263 (30 seeds per cell).
Total: 112 N x L cells and 3,360 unique stimuli.
The stimulus bank is shared across all registered models and prompt modes.… See the full description on the dataset page: https://huggingface.co/datasets/twistshan/realistic-niah-count-empirical-law.