Dataset · nvidia
Nemotron Science SFT dataset
nvidia/Nemotron-SFT-Science-v2
A large dataset of scientific reasoning for physics, chemistry, and biology featuring synthetic and real tasks.
What's inside
Contains between 1M and 10M English examples. Covers physics, biology, and chemistry in multiple-choice and open-ended question formats. Solutions are generated by LLMs using Chain-of-Thought (CoT), Python, and search tools.
What you can do with it
- Training scientific LLMs with reasoning (CoT)
- Training models to use tools such as Python and search
- Tasks
- Text generation
- Languages
- English
- Size
- 1M–10M rows
- Downloads this month
- 22K
- Likes
- 21
Models trained on it
- mradermacher/LlamaScientific-8B-i1-GGUF 4.2K
- abhishekchohan/Qwen3.8-27B-AWQ-INT4-FP8KVImage text → text 4K
- mradermacher/FlashModel-Qwen3.5-9B-i1-GGUF 2.3K
- mradermacher/LlamaScientific-8B-GGUF 1.1K
- flashback2k/FlashModel-Qwen3.5-9B-GGUFText generation 732
- abhishekchohan/Qwen3.8-27B-AWQ-INT4Image text → text 660
- mradermacher/TinyBalls-110M-V1-GGUFText generation 445
- abhishekchohan/Qwen3.8-27B-GPTQ-INT4-FP8KVImage text → text 389
How to load it
from datasets import load_dataset
ds = load_dataset("nvidia/Nemotron-SFT-Science-v2")