datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
jblimp
JBLiMP
This is the data from "JBLiMP: Japanese Benchmark of Linguistic Minimal Pairs" (Someya and Oseki, 2023). Only the validated pairs used for benchmarks are included, and only in JSONL format, since it's redundant with the TSV.
For details see the original git repo or the paper.
xwinograd-jaThis is the Japanese portion of the xwinograd dataset, formatted for easy use.
The original data can be found here. When using this data, please cite the original papers.
@misc{muennighoff2022crosslingual,
title={Crosslingual Generalization through Multitask Finetuning},
author={Niklas Muennighoff and Thomas Wang and Lintang Sutawika and Adam Roberts and Stella Biderman and Teven Le Scao and M Saiful Bari and Sheng Shen and Zheng-Xin Yong and Hailey Schoelkopf and Xiangru Tang… See the full description on the dataset page: https://huggingface.co/datasets/polm-stability/xwinograd-ja.stabilityai__stablelm-2-1_6b-details
Dataset Card for Evaluation run of stabilityai/stablelm-2-1_6b
Dataset automatically created during the evaluation run of model stabilityai/stablelm-2-1_6b
The dataset is composed of 44 configuration(s), each one corresponding to one of the evaluated task.
The dataset has been created from 2 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The "train" split is always pointing to the latest results.
An… See the full description on the dataset page: https://huggingface.co/datasets/open-llm-leaderboard/stabilityai__stablelm-2-1_6b-details.stabilityai__stablelm-2-12b-chat-details
Dataset Card for Evaluation run of stabilityai/stablelm-2-12b-chat
Dataset automatically created during the evaluation run of model stabilityai/stablelm-2-12b-chat
The dataset is composed of 44 configuration(s), each one corresponding to one of the evaluated task.
The dataset has been created from 2 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The "train" split is always pointing to the latest… See the full description on the dataset page: https://huggingface.co/datasets/open-llm-leaderboard/stabilityai__stablelm-2-12b-chat-details.stabilityai__stablelm-zephyr-3b-details
Dataset Card for Evaluation run of stabilityai/stablelm-zephyr-3b
Dataset automatically created during the evaluation run of model stabilityai/stablelm-zephyr-3b
The dataset is composed of 44 configuration(s), each one corresponding to one of the evaluated task.
The dataset has been created from 2 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The "train" split is always pointing to the latest results.… See the full description on the dataset page: https://huggingface.co/datasets/open-llm-leaderboard/stabilityai__stablelm-zephyr-3b-details.stabilityai__stablelm-2-12b-details
Dataset Card for Evaluation run of stabilityai/stablelm-2-12b
Dataset automatically created during the evaluation run of model stabilityai/stablelm-2-12b
The dataset is composed of 44 configuration(s), each one corresponding to one of the evaluated task.
The dataset has been created from 1 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The "train" split is always pointing to the latest results.
An… See the full description on the dataset page: https://huggingface.co/datasets/open-llm-leaderboard/stabilityai__stablelm-2-12b-details.exp-temporal-stability
Experiment H13: Temporal Stability Across Model Versions
Paper DOI: 10.5281/zenodo.19422427 — R15 (Zharnikov, 2026v)
Dataset DOI: 10.57967/hf/8455
Source Code: spectralbranding/sbt-papers/r15-ai-search-metamerism
Dataset Summary
450 LLM API calls testing whether successive model versions produce significantly different dimensional weight profiles for the same brands. Supplementary to the R15 study on dimensional collapse in AI-mediated brand perception (Zharnikov… See the full description on the dataset page: https://huggingface.co/datasets/spectralbranding/exp-temporal-stability.han-coordination-stability-stress-dataset-v1
Humanoid Coordination Stability Stress-Test Dataset
This dataset captures multi-agent coordination
under stress conditions including:
High task load
Delayed synchronization
Partial node failure
Conflicting objectives
Objective
To evaluate how decentralized humanoid agents
maintain coordination stability
under adverse conditions.
Data Fields
active_agent_count
task_load_index
sync_delay_ms
node_failure_flag
coordination_success_rate
recovery_time_ms… See the full description on the dataset page: https://huggingface.co/datasets/achiepatricia/han-coordination-stability-stress-dataset-v1.stabilityai__stablelm-2-1_6b-chat-details
Dataset Card for Evaluation run of stabilityai/stablelm-2-1_6b-chat
Dataset automatically created during the evaluation run of model stabilityai/stablelm-2-1_6b-chat
The dataset is composed of 44 configuration(s), each one corresponding to one of the evaluated task.
The dataset has been created from 2 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The "train" split is always pointing to the latest… See the full description on the dataset page: https://huggingface.co/datasets/open-llm-leaderboard/stabilityai__stablelm-2-1_6b-chat-details.stabilityai__StableBeluga2-details
Dataset Card for Evaluation run of stabilityai/StableBeluga2
Dataset automatically created during the evaluation run of model stabilityai/StableBeluga2
The dataset is composed of 38 configuration(s), each one corresponding to one of the evaluated task.
The dataset has been created from 1 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The "train" split is always pointing to the latest results.
An… See the full description on the dataset page: https://huggingface.co/datasets/open-llm-leaderboard/stabilityai__StableBeluga2-details.schemaforge-stability-ai-news-16
stability.ai
Auto-refined by SchemaForge
Metadata
Topic: Stability AI News
Quality Score: 0.92
Source: Autonomous web scraper
Extracted Facts
Stable Audio 3.0 model family released
Brand Studio creative production platform introduced
Stability AI joins Tech Coalition
Warner Music Group and Stability AI collaborate on AI tools
Universal Music Group and Stability AI form strategic alliance
Stability AI and EA partner for game development
Stability… See the full description on the dataset page: https://huggingface.co/datasets/GudduButt/schemaforge-stability-ai-news-16.stabilityai__stablelm-2-zephyr-1_6b-details
Dataset Card for Evaluation run of stabilityai/stablelm-2-zephyr-1_6b
Dataset automatically created during the evaluation run of model stabilityai/stablelm-2-zephyr-1_6b
The dataset is composed of 44 configuration(s), each one corresponding to one of the evaluated task.
The dataset has been created from 2 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The "train" split is always pointing to the latest… See the full description on the dataset page: https://huggingface.co/datasets/open-llm-leaderboard/stabilityai__stablelm-2-zephyr-1_6b-details.stabilityai__stablelm-3b-4e1t-details
Dataset Card for Evaluation run of stabilityai/stablelm-3b-4e1t
Dataset automatically created during the evaluation run of model stabilityai/stablelm-3b-4e1t
The dataset is composed of 44 configuration(s), each one corresponding to one of the evaluated task.
The dataset has been created from 1 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The "train" split is always pointing to the latest results.
An… See the full description on the dataset page: https://huggingface.co/datasets/open-llm-leaderboard/stabilityai__stablelm-3b-4e1t-details.han-humanoid-locomotion-stability-v1
Humanoid Locomotion Stability Dataset
Overview
Dataset ini berisi parameter pergerakan humanoid saat berjalan
dan label stabilitasnya.
Features
step_length_cm
stride_frequency_hz
center_of_mass_shift_cm
ground_reaction_force_n
terrain_type_index
imu_balance_variance
Target
stability_status (stable / unstable)
Task
Binary Classification
pg19-stability-bench
PG-19 Stability Evaluation Prompts
Long-context prompts in chat-message format at five bucket sizes (8K, 16K, 32K, 64K, 128K user-message tokens), designed for output-stability evaluation of LLMs under stress: as context grows (and RoPE scaling extends the effective window), do generations stay coherent — or do they degrade into mojibake, token soup, phrase loops, or script drift?
Each prompt is a 2-turn conversation (system + user) ready to send to any OpenAI-compatible… See the full description on the dataset page: https://huggingface.co/datasets/nnilayy/pg19-stability-bench.project-aio
Project AIO (AI王) Data
This contains the data distributed for the AI王 quiz competition.
License
See the original license information.
The data for the second and third competitions, used here, is distributed under a CC-BY-SA 4.0 license.
han-humanoid-gait-stability-dataset-v1
Humanoid Gait Stability Dataset
Description
Dataset berisi metrik kestabilan langkah humanoid saat berjalan
di berbagai kondisi permukaan.
Features
sample_id (string)
surface_type (categorical)
center_of_mass_shift (float)
step_width_variation (float)
torso_tilt_variance (float)
foot_slip_ratio (float)
Label
stability_class (stable / moderate / unstable)
Task
Multiclass Classification
Metrics
Accuracy
F1 Score
schemaforge-stability-ai-news-17
stability.ai
Auto-refined by SchemaForge
Metadata
Topic: Stability AI News
Quality Score: 0.92
Source: Autonomous web scraper
Extracted Facts
Stable Audio 3.0 is a model family for artistic experimentation
Stable Audio 3.0 is trained on fully licensed data
Brand Studio is a creative production platform powered by Stability AI
Stability AI has joined the Tech Coalition to combat online child sexual exploitation
Warner Music Group and Stability AI… See the full description on the dataset page: https://huggingface.co/datasets/GudduButt/schemaforge-stability-ai-news-17.han-swarm-coordination-stability-dataset-v1
Humanoid Swarm Coordination Stability Dataset
This dataset models coordination stability
across multiple humanoid agents
operating as a distributed swarm.
It captures synchronization behavior,
task alignment, and coordination drift
under dynamic mission conditions.
Objective
To measure and optimize swarm stability
in decentralized humanoid environments.
Data Fields
coordination_event_id
swarm_id
agent_ids
mission_type
synchronization_score
task_alignment_index… See the full description on the dataset page: https://huggingface.co/datasets/achiepatricia/han-swarm-coordination-stability-dataset-v1.han-humanoid-interaction-stability-events-v1
Humanoid Interaction Stability Events (HISE)
Problem Definition
In human-robot interaction, stability degradation
can occur due to environmental shifts, ambiguous commands,
or actuator inconsistencies.
This dataset captures structured interaction events
where stability levels fluctuate during task execution.
Dataset Description
Each entry represents a single interaction event
within a controlled humanoid task session.
Features
interaction_id… See the full description on the dataset page: https://huggingface.co/datasets/ariefansclub/han-humanoid-interaction-stability-events-v1.stabilityDownloaded from: https://github.com/songlab-cal/tape
Routine-Stability-Data