Datasetpublic

Stormtrooperaim/Ultra-Thinker-30k

Ultra-Thinker 🧠 A comprehensive collection of high-quality reasoning and conversational datasets designed to enhance the thinking capabilities of large language models. Overview Ultra-Thinker aggregates diverse sources focused on: Chain-of-thought reasoning - Step-by-step logical deduction Mathematical problem solving - Complex computational challenges Advanced logical reasoning - Multi-step inference and analysis This curated compilation combines… See the full description on the dataset page: https://huggingface.co/datasets/Stormtrooperaim/Ultra-Thinker-30k.

sourceHugging Facemitupdated 8mo agoView on Hugging Face

4likes46downloads

Dataset Card

Ultra-Thinker 🧠

A comprehensive collection of high-quality reasoning and conversational datasets designed to enhance the thinking capabilities of large language models.

Overview

Ultra-Thinker aggregates diverse sources focused on:

—Chain-of-thought reasoning - Step-by-step logical deduction
—Mathematical problem solving - Complex computational challenges
—Advanced logical reasoning - Multi-step inference and analysis

This curated compilation combines cutting-edge reasoning datasets to push the boundaries of AI cognitive capabilities.

📊 Included Datasets

All datasets are from TeichAI on Hugging Face:

High-Reasoning Collections

Dataset	Size	Link
Gemini 3 Pro Preview (High Reasoning)	250x	🔗 View Dataset
Gemini 3 Pro Preview (High Reasoning)	1000x	🔗 View Dataset
Gemini 3 Flash Preview	1200x	🔗 View Dataset
GPT-5.2 (High Reasoning)	250x	🔗 View Dataset
Claude 4.5 Opus (High Reasoning)	250x	🔗 View Dataset
Claude Sonnet 4.5 (High Reasoning)	250x	🔗 View Dataset
Claude Haiku 4.5 (High Reasoning)	1700x	🔗 View Dataset

Specialized Model Collections

Dataset	Size	Link
MiniMax M2.1	8800x	🔗 View Dataset
MiniMax M2.1	1000x	🔗 View Dataset
MiMo V2 Flash	2300x	🔗 View Dataset
GLM-4.7	350x	🔗 View Dataset
GLM-4.7	2000x	🔗 View Dataset
Kimi K2 Thinking	1000x	🔗 View Dataset

DeepSeek V3.2 Special Collections

Dataset	Focus	Link
DeepSeek V3.2 Speciale	OpenCodeReasoning (3k)	🔗 View Dataset
DeepSeek V3.2 Speciale	OpenR1 Math (3k)	🔗 View Dataset

🚀 Use Cases

—Fine-tuning language models for enhanced reasoning
—Training AI systems on complex problem-solving
—Benchmarking model performance on multi-step tasks
—Research in cognitive AI capabilities

📝 Citation

If you use Ultra-Thinker in your research, please cite the original datasets from TeichAI.

🔗 Resources

—Source Organization: TeichAI on Hugging Face
—All Datasets: Browse the complete collection on the TeichAI profile

Built with datasets from the TeichAI community