datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
MiroFlow-BenchmarksThese are the benchmarking datasets used for MiroFlow Framework. More information: https://github.com/MiroMindAI/MiroThinker
MiroEval-data
MiroEval: Benchmarking Multimodal Deep Research Agents in Process and Outcome
MiroEval is a comprehensive evaluation framework for Deep Research systems, providing automated task generation and assessment across three complementary dimensions: Factual correctness, Point-wise quality, and Process quality.
Quick Start
1. Setup
All three evaluation modules share a single Python environment managed by uv at the repo root:
uv sync
If you use… See the full description on the dataset page: https://huggingface.co/datasets/miromind-ai/MiroEval-data.MiroMind-M1-SFT-719K
MiroMind-M1
🧾 Overview
Training performance of MiroMind-M1-RL-7B on AIME24 and AIME25.
MiroMind-M1 is a fully open-source series of reasoning language models built on Qwen-2.5, focused on advancing mathematical reasoning. It is trained through supervised fine-tuning (SFT) on 719K curated problems and reinforcement learning with verifiable rewards (RLVR) on 62K challenging examples, using a context-aware multi-stage policy optimization method… See the full description on the dataset page: https://huggingface.co/datasets/miromind-ai/MiroMind-M1-SFT-719K.MiroVerse-v0.1
MiroVerse: A Reproducible, Full-Trajectory, Ever-Growing Deep Research Dataset
🔥 News & Updates
MiroVerse v0.1 has been released. This dataset can be used with our training framework, MiroTrain. In MiroVerse v0.1, we provide both SFT and DPO data, making it easy to reproduce MiroThinker-v0.1’s benchmark performance on Qwen3. Give it a try!
The initial release of MiroVerse (v0.1) is coming this Friday—stay tuned!
🔥 First Batch of MiroVerse… See the full description on the dataset page: https://huggingface.co/datasets/miromind-ai/MiroVerse-v0.1.MiroMind-M1-RL-62K
MiroMind-M1
🧾 Overview
Training performance of MiroMind-M1-RL-7B on AIME24 and AIME25.
MiroMind-M1 is a fully open-source series of reasoning language models built on Qwen-2.5, focused on advancing mathematical reasoning. It is trained through supervised fine-tuning (SFT) on 719K curated problems and reinforcement learning with verifiable rewards (RLVR) on 62K challenging examples, using a context-aware multi-stage policy optimization method… See the full description on the dataset page: https://huggingface.co/datasets/miromind-ai/MiroMind-M1-RL-62K.MiroMind-SFTMiroRL-GenQA
MiroRL-GenQA
A curated dataset for reinforcement learning (RL) training within the MiroRL framework.
Overview
Source: Provided by MiroMind AI as part of the MiroRL project.
Format & Size: Contains ~13.1k examples in Parquet format for efficient loading and processing.
License: Released under CC-BY-NC-4.0 for non-commercial use.
Purpose: Designed to serve as high-quality input for RL fine-tuning in the MiroRL pipeline.
Dataset Structure
Each record… See the full description on the dataset page: https://huggingface.co/datasets/miromind-ai/MiroRL-GenQA.MiroMind-M1-SFT-719K-transformedformatted_miromind-1000MiroMind-SFT-mini