media Hugging Face Forums · 1h ago · open_models

Young25 releases Humor_Vote, a Chinese humor dataset with human vote labels

from English

User Young25 has released the Humor_Vote dataset on Hugging Face, providing a resource for evaluating and training models on Chinese humor. The dataset consists of native Chinese humorous questions paired with four candidate answers and corresponding human vote counts.

Each row contains a question_cn field, four answer fields (answer_cn_0 through answer_cn_3), the chosen_cn label, and vote counts per answer.
All content is in native Chinese rather than translated text to preserve natural wording and punchlines.
The data structure supports direct use for SFT, DPO-style preference pairs, reward modeling, or as a probe for model humor evaluation.
It is also suitable for entertainment applications such as party icebreakers, quiz games, and humor content generation.

The release aims to assist researchers working on humor evaluation and preference data, with the author welcoming feedback on format and label quality.

Importance 1/3 Trust 1/3 Hugging Face Forums Evaluation & benchmarks Training data

Read original

media Hugging Face Forums · 32d ago · 23 views

louidev releases GlassBallAI, a time-locked dataset of Gemini forecasts for studying LLM behavior

louidev has released the GlassBallAI dataset on Hugging Face, capturing live forecasting behavior from Google's Gemini models before outcomes are known. The dataset contains over 3,655 rows of data collected between February 17 and May 19, 2026, covering Gemini 2.5 Flash, 2.5 Pro, 2.5 Flash Lite, and 3.0 Flash Preview.

arxiv arXiv cs.AI · 34d ago · 19 views

ViHoRec: A Quality-Controlled Vietnamese Hotel Recommendation Dataset and Cold-Start Benchmark

Researchers introduce ViHoRec, a public dataset of 18,267 interactions between 6,832 users and 560 hotels for Vietnamese hotel recommendation. The resource addresses challenges in cross-platform entity resolution and privacy-preserving release using HMAC pseudonyms.

arxiv arXiv cs.LG · 36d ago · 8 views

SYNRARE generates synthetic rare disease EHRs for ML benchmarking

SYNRARE is a graphical user interface based on the Synthea framework designed to generate synthetic Electronic Health Records (EHR) for rare disease patients. It allows researchers to create patient data that differs from common disease cases by a definable degree, addressing privacy and legal barriers associated with real patient data.

arxiv arXiv cs.LG · 36d ago · 7 views

TSAI releases MetaFraud benchmark for multimodal financial fraud detection in the metaverse

Researchers have introduced TSAI-MetaFraud, a new multimodal benchmark dataset designed to address the lack of comprehensive resources for detecting fraud and illicit behavior in virtual economies. Unlike existing datasets that analyze user behavior, authentication, or financial transactions in isolation, this resource integrates behavioral, transactional, and graph-structured information.

arxiv arXiv cs.LG · 39d ago · 21 views

ImputeViz visual analytics dashboard for diagnosing missing data and comparing imputation methods

Researchers have introduced ImputeViz, an integrated visual analytics dashboard designed to support the diagnosis of missingness patterns, configuration of imputation models, and evaluation of results. The system integrates widely used methods such as MICE, Random Forest, XGBoost, and kNN within an interactive environment that makes missingness structures explicit.