Dataset · nvidia
Mathematical proofs and formalization in Lean 4
nvidia/Nemotron-Math-Proofs-v1
About 580k proof tasks, 550k Lean 4 formalizations, and 900k model reasoning traces.
What's inside
The dataset contains about 580k natural language problems with proofs, about 550k Lean 4 theorem formalizations, and about 900k model-generated reasoning traces. Language is English, modality is text. Data is related to the Nemotron-Math and NeMo-Skills project; format is not specified in the card.
What you can do with it
- Training LLMs on mathematical proofs
- SFT for theorem formalization in Lean 4
- Evaluating long-context mathematical reasoning
- Tasks
- Text generation
- Languages
- English
- Size
- 100K–1M rows
- Downloads this month
- 1.6K
- Likes
- 132
Models trained on it
- nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16Text generation 1.1M
- nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-NVFP4Text generation 641.7K
- nvidia/NVIDIA-Nemotron-3-Nano-4B-BF16Text generation 216.1K
- nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-FP8Text generation 189.4K
- nvidia/NVIDIA-Nemotron-3-Nano-4B-FP8Text generation 55.8K
- unsloth/Nemotron-3-Nano-30B-A3B-GGUFText generation 31.7K
- lmstudio-community/NVIDIA-Nemotron-3-Nano-30B-A3B-MLX-4bitText generation 30.5K
- lmstudio-community/NVIDIA-Nemotron-3-Nano-30B-A3B-MLX-8bitText generation 27.6K
How to load it
from datasets import load_dataset
ds = load_dataset("nvidia/Nemotron-Math-Proofs-v1")