Open math problems for reasoning benchmark
17.6K published math problems of varying difficulty for testing reasoning models.
- Question answering
- Text generation
- English
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
17.6K published math problems of varying difficulty for testing reasoning models.
A collection of verifiable tasks for training computer agents using reinforcement learning.
Short text snippets from English video tutorials with timestamps for search and RAG.
57,937 filtered trajectories from three models for SFT in math, code, reasoning, and multilingual tasks.
Parallel texts for evaluating translation from English into multiple languages, including low-resource ones.
A small collection of JSONL traces of AI agent operation with saved tool schemas for SFT.
Raw anonymized traces of the Claude Fable 5 agent for training and analyzing agent scenarios.
Over 142 thousand live chats with ChatGPT, Claude, Gemini, Grok, and Perplexity for chatbot training and prompt analysis.
Large Chinese dataset (1M to 10M) for search, recommendations, and RAG systems.
Large English and Japanese code corpus for training and fine-tuning language models.
A large open speech dataset across 25 European languages for ASR and speech translation.
Clean sectioned corpus in Arabic and English for pre-training a small LLM.
110k Chinese examples with R1 answers gathered in the output field for SFT training.
A large-scale collection of diverse Arabic texts from various sources for a wide range of NLP tasks.
A large code change history corpus across 20 programming languages for generating and supplementing commit messages.
Approximately 900K question-answer pairs (40% image-based, 60% text-based) for boosting multimodal model reasoning logic.
Questions and answers based on real Thai court rulings with reasoning and legal references.
A large collection of French public domain monographs for training generative language models.
15,140 jailbreak prompts collected from December 2022 to December 2023.
Test tasks for evaluating autonomous agents in healthcare scenarios: from treatment authorization to patient management.
From 10,000 to 100,000 test questions from Arabic school exams for evaluating language models.
A large corpus of openly licensed German texts from 41 sources for training language models.
Synthetic financial documents in six languages with personal data markup for NER training.
1,225 queries for evaluating language agents in multi-step trip planning with constraints.