datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
SLING
SLING: Sino-Linguistic Evaluation of Large Language Models
This is the official SLING dataset, accompanying the EMNLP 2022 paper "SLING: Sino-Linguistic Evaluation of Large Language Models" by Yixiao Song♢ Kalpesh Krishna♠ Rajesh Bhatt♢ Mohit Iyyer♠.
You can find the paper on arxiv.
We use this dataset for evaluation of a small-scale Chinese Language Model for the BabyLM Challenge.
SLING Dataset
See SLING_Data and the readme file in it.
A complete list of all… See the full description on the dataset page: https://huggingface.co/datasets/BeetleLM/SLING.arabic-reasoning-dataset-logic
Arabic Logical Reasoning Tasks Dataset (Maximum 1000 Tasks)
Overview
This dataset comprises a series of logical reasoning tasks designed to evaluate and train artificial intelligence models on understanding and generating logical inferences in the Arabic language. Each task includes a unique identifier, the task type, the task text (a question and a proposed answer), and a detailed solution that outlines the thinking steps and the final answer.
Data Format
The… See the full description on the dataset page: https://huggingface.co/datasets/beetleware/arabic-reasoning-dataset-logic.dataforge-20260905T082520
dataforge-20260905T082520
Training dataset generated with DataForge: 22552 records (22550 tabular, 2 text).
Records: 59344
By type: {'text': 2, 'tabular': 22550, 'image': 11777, 'video': 25015}
Splits: {'train': 47277, 'val': 5909, 'test': 5911}
Generated locally with DataForge (text/tabular/image/video/document -> a
single deduplicated, unified-schema JSONL dataset).
ogiri-bokete-unsloth-vlm
Japanese Bokete Ogiri — Unsloth VLM format
YANS-official/ogiri-bokete を、UnslothのVision SFTで扱える会話形式に変換した非公開用データセットです。
各JSONLレコードは「1画像 + 1回答」です。
{
"messages": [
{"role": "user", "content": [
{"type": "image", "image": "images/124469.jpg"},
{"type": "text", "text": "この画像のお題に対して、面白い一言を1つ返してください。"}
]},
{"role": "assistant", "content": [
{"type": "text", "text": "..."}
]}
]
}
Files
train.jsonl: 1,678 records / 630 prompts… See the full description on the dataset page: https://huggingface.co/datasets/beezza/ogiri-bokete-unsloth-vlm.dataforge-20260904T124011
dataforge-20260904T124011
Training dataset generated with DataForge: 310 records (305 document, 5 image).
Records: 310
By type: {'image': 5, 'document': 305}
Splits: {'train': 248, 'val': 31, 'test': 31}
Generated locally with DataForge (text/tabular/image/video/document -> a
single deduplicated, unified-schema JSONL dataset).
spelling-bee-pangrams
Spelling Bee-style letter sets and their pangrams
Each row is one puzzle: 7 distinct letters (letters, alphabetical) and the list
of pangrams — words using all 7 letters. Nothing else.
Puzzles come from the dwyl/english-words words_alpha.txt word list (~370k
entries). Included: every 7-letter combination with at least one pangram and at
least 10 valid answers (words of 4+ letters using only the 7 letters).
Source: https://github.com/dwyl/english-words (words_alpha.txt)
sdf-data-bee_speedBeer-Tv-AdsBEE-spoke-data__Meta-Llama-3-8Bee-details
Dataset Card for Evaluation run of BEE-spoke-data/Meta-Llama-3-8Bee
Dataset automatically created during the evaluation run of model BEE-spoke-data/Meta-Llama-3-8Bee
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… See the full description on the dataset page: https://huggingface.co/datasets/open-llm-leaderboard/BEE-spoke-data__Meta-Llama-3-8Bee-details.beerdes-beep-corpus
DES Beep Corpus
A structured corpus of 195 beep-moment experience descriptions from Russell Hurlburt's Descriptive Experience Sampling (DES) research — the most rigorous method ever developed for investigating the contents of inner experience.
What is DES?
Descriptive Experience Sampling asks participants to wear a beeper that sounds at random intervals throughout the day. At each beep, the participant freezes their experience and jots notes about whatever was in their… See the full description on the dataset page: https://huggingface.co/datasets/david-ar/des-beep-corpus.big_beetle_datasetBEE-spoke-data__smol_llama-220M-openhermes-details
Dataset Card for Evaluation run of BEE-spoke-data/smol_llama-220M-openhermes
Dataset automatically created during the evaluation run of model BEE-spoke-data/smol_llama-220M-openhermes
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… See the full description on the dataset page: https://huggingface.co/datasets/open-llm-leaderboard/BEE-spoke-data__smol_llama-220M-openhermes-details.BEE-spoke-data__smol_llama-220M-GQA-details
Dataset Card for Evaluation run of BEE-spoke-data/smol_llama-220M-GQA
Dataset automatically created during the evaluation run of model BEE-spoke-data/smol_llama-220M-GQA
The dataset is composed of 43 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… See the full description on the dataset page: https://huggingface.co/datasets/open-llm-leaderboard/BEE-spoke-data__smol_llama-220M-GQA-details.BEE-spoke-data__smol_llama-101M-GQA-details
Dataset Card for Evaluation run of BEE-spoke-data/smol_llama-101M-GQA
Dataset automatically created during the evaluation run of model BEE-spoke-data/smol_llama-101M-GQA
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… See the full description on the dataset page: https://huggingface.co/datasets/open-llm-leaderboard/BEE-spoke-data__smol_llama-101M-GQA-details.beecare-text-rich-qa-bilingual-large
BeeCare Text Rich QA Bilingual Large
Unsloth-friendly large bilingual text dataset. Default split is train. Columns include question, answer, instruction, output, text, labels, severity, and safety tags.
Use text for simple text SFT, or map instruction -> prompt and output -> response if the UI offers Alpaca-style mapping.
sweep-statusbeerbeer_3BEE-spoke-data__smol_llama-220M-GQA-fineweb_edu-details
Dataset Card for Evaluation run of BEE-spoke-data/smol_llama-220M-GQA-fineweb_edu
Dataset automatically created during the evaluation run of model BEE-spoke-data/smol_llama-220M-GQA-fineweb_edu
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… See the full description on the dataset page: https://huggingface.co/datasets/open-llm-leaderboard/BEE-spoke-data__smol_llama-220M-GQA-fineweb_edu-details.BEE-spoke-data__tFINE-900m-e16-d32-flan-infinity-instruct-7m-T2T_en-1024-details
Dataset Card for Evaluation run of BEE-spoke-data/tFINE-900m-e16-d32-flan-infinity-instruct-7m-T2T_en-1024
Dataset automatically created during the evaluation run of model BEE-spoke-data/tFINE-900m-e16-d32-flan-infinity-instruct-7m-T2T_en-1024
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… See the full description on the dataset page: https://huggingface.co/datasets/open-llm-leaderboard/BEE-spoke-data__tFINE-900m-e16-d32-flan-infinity-instruct-7m-T2T_en-1024-details.BEE-spoke-data__tFINE-900m-instruct-orpo-details
Dataset Card for Evaluation run of BEE-spoke-data/tFINE-900m-instruct-orpo
Dataset automatically created during the evaluation run of model BEE-spoke-data/tFINE-900m-instruct-orpo
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… See the full description on the dataset page: https://huggingface.co/datasets/open-llm-leaderboard/BEE-spoke-data__tFINE-900m-instruct-orpo-details.small_beetle_datasetThis dataset was produced by the Snakemake workflow in: https://github.com/songlab-cal/gpn/tree/main/workflow/make_dataset
The following accessions are included in this dataset:
Assembly Accession
Assembly Name
Organism Name
GCF_031307605.1
icTriCast1.1
Tribolium castaneum
GCF_963966145.1
icTenMoli1.1
Tenebrio molitor
GCF_036711695.1
CSIRO_AGI_Zmor_V1
Zophobas morio
GCF_015345945.1
Tmad_KSU_1.1
Tribolium madens
The only adapted config is this:
# this chroms are forced to… See the full description on the dataset page: https://huggingface.co/datasets/sbuedenb/small_beetle_dataset.BEE-spoke-data__tFINE-900m-e16-d32-flan-details
Dataset Card for Evaluation run of BEE-spoke-data/tFINE-900m-e16-d32-flan
Dataset automatically created during the evaluation run of model BEE-spoke-data/tFINE-900m-e16-d32-flan
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… See the full description on the dataset page: https://huggingface.co/datasets/open-llm-leaderboard/BEE-spoke-data__tFINE-900m-e16-d32-flan-details.BEE-spoke-data__tFINE-900m-e16-d32-instruct_2e-details
Dataset Card for Evaluation run of BEE-spoke-data/tFINE-900m-e16-d32-instruct_2e
Dataset automatically created during the evaluation run of model BEE-spoke-data/tFINE-900m-e16-d32-instruct_2e
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… See the full description on the dataset page: https://huggingface.co/datasets/open-llm-leaderboard/BEE-spoke-data__tFINE-900m-e16-d32-instruct_2e-details.big_beetle_dataset-1024Beebig_beetle_dataset-8192beecare-text-public-bilingual-large-train
BeeCare Text Public Bilingual Large Train
Default train split for Unsloth text fine-tuning. Each row has messages and metadata.
Use in Unsloth Studio as: yahelr1/beecare-text-public-bilingual-large-train.
big_beetle_dataset-2048