Multi-Billion VK Public Data Scrap
Regularly updated corpus of posts, comments, photos, videos, and profiles from VK in Russian and Ukrainian.
- Text classification
- Russian
- Ukrainian
Hugging Face datasets with plain-language descriptions: what's inside and what models to train on them. 4,788 in the catalog so far, with new ones added every hour.
Answered by Gemini 4 · the chat is kept only in this tab
Regularly updated corpus of posts, comments, photos, videos, and profiles from VK in Russian and Ukrainian.
A set of seven test tasks to check the quality and safety of language models in the low-resource Saraiki language.
A curated collection of 59 English datasets for training VLMs in logic, STEM, charts, and OCR.
A dynamic English-language benchmark for assessing how well multimodal LLMs understand scientific journals and illustrations.
Over 16 million historical and current exchange rate records from central banks and tax authorities.
Tabular dataset with authority scores for 10.3 million domains based on Common Crawl and Majestic for regression and classification tasks.
A set of under 1,000 complex multimodal questions with images for testing the limits of modern models.
Processed on-chain Polymarket trades in tables and text for market and user behavior analysis.
Sliding data sample over the last 7 days on crypto futures for time-series forecasting.
Formatted DocVQA version for one-click evaluation of large multimodal models.
Generated pages in Vietnamese and English with block annotations for training document parsing.
106 hours of free-camera flights across 17 games for 3D reconstruction and video generation.
Corpus of English medical and scientific texts for evaluating information retrieval models.
A quality-filtered subset of FineWeb2 across 20 languages for faster language model pretraining.
Coding tasks with reasoning traces, direct answers, and task synthesis in parquet.
Formatted version of the TextVQA benchmark for evaluating multimodal models in lmms-eval.
Set of queries and mixed corpora based on MSCOCO, Google_WIT, VisualNews, and OVEN for cross-modal search evaluation.
100 hours of meeting recordings with synchronized microphones, video, and text for ASR and meeting analysis.
Classic dataset for evaluating text embedding quality from the MTEB benchmark.
A large-scale multimodal benchmark for evaluating action understanding in video.
Between 1 and 10 million synthetic English question-answer pairs for evaluating question generation models.
.pcapng network packet captures for analyzing latency and network behavior during inference with various language models.
Multimodal dataset from Amazon Science to test VLM capability to find answers in bulky documents and images.
2,554 episodes in AI2-THOR and ProcTHOR simulators for evaluating the memory of embodied agents.