Synthetic corpus for reasoning models
79.6 million Wikipedia-based text examples for end-to-end training of small reasoning models.
- Text generation
- Zero shot classification
- Summarization
- English
- French
- Italian
- +5
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
79.6 million Wikipedia-based text examples for end-to-end training of small reasoning models.
Large geophysical benchmark of tens of thousands of 3D velocity volumes, 2D slices, and wavefields for seismic exploration.
Agent rollouts across various frameworks for software engineering tasks, intended for distillation and training SWE agents.
31k hours of image-prompted spontaneous speech across 105 Indian languages, with transcripts for ASR and multimodal tasks.
Large grounded vision-language dataset for chest CT analysis; metadata lists text, 100K–1M.
13,083 English customer banking queries labeled with 77 fine-grained intents for classification.
Instructional text dataset for the second SFT stage of the open OpenCoder model.
A multimodal dataset with schematic images and questions to test VLM engineering reasoning.
6.3M documents and 14.7B tokens of math web text for LLM pre-training and fine-tuning.
English pairs of images, questions, and answers gathered from arXiv research publications.
A small audio and text dataset of English sentences containing Yiddish words for ASR tasks.
Grid-aligned ocean data (temperature, salinity, surface height) across 50 depth levels for ML models.
A dictionary mapping English words to their phonetic transcription for speech synthesis.
A corpus of 10 to 100 thousand text records associated with the UniREditBench reasoning-based image editing benchmark.
Over 33 thousand open legal articles and publications in Traditional Chinese for fine-tuning language models.
Math competition problems with step-by-step solutions for benchmarking in lighteval.
A large-scale benchmark featuring optical and synthetic aperture radar (SAR) remote sensing images and text descriptions for visual grounding.
Between 100k and 1M audio-text pairs in Moore with verified alignment for ASR and TTS.
925K English instructions emphasizing multi-image and dense text tasks.
A curated selection of 8,000 English website screenshots for training models to understand web interface elements.
About 200k English school math word problems with answers generated by GPT-4-Turbo.
Collection of 48,284 English books from Project Gutenberg, prepared for generative model training.
Large multi-domain image-text corpus for 4D world modeling and robotics.
Q&A based on historical US Treasury bulletins to evaluate AI agents' ability to analyze tables, charts, and text.