Dataset · nvidia
Nemotron post-training SFT and RL data in 6 languages
nvidia/Nemotron-Post-Training-Dataset-v2
NVIDIA post-training SFT and RL data for math, code, reasoning, and instructions across six languages.
What's inside
Nemotron-Post-Training-Dataset-v2 expands NVIDIA's post-training data: SFT and RL examples added for Spanish, French, German, Italian, and Japanese. The data supports improvements in math, code, general reasoning, and instruction following. Languages include English, German, Italian, French, Spanish, and Japanese; volume ranges from 1 to 10 million examples.
What you can do with it
- SFT/RL for a Nemotron-like reasoning model
- Evaluating multilingual reasoning
- Improving instruction following
- Languages
- English, German, Italian, French, Spanish, Japanese
- Size
- 1M–10M rows
- Downloads this month
- 29.9K
- Likes
- 162
- Access
- on request on Hugging Face
Models trained on it
- nvidia/NVIDIA-Nemotron-3-Nano-4B-BF16Text generation 216.1K
- nvidia/NVIDIA-Nemotron-Nano-9B-v2Text generation 204.4K
- nvidia/NVIDIA-Nemotron-Nano-9B-v2-FP8Text generation 149.7K
- nvidia/NVIDIA-Nemotron-3-Nano-4B-FP8Text generation 55.8K
- nvidia/NVIDIA-Nemotron-Nano-9B-v2-NVFP4Text generation 49.8K
- nvidia/NVIDIA-Nemotron-3-Nano-4B-GGUFText generation 23.9K
- nvidia/NVIDIA-Nemotron-Nano-12B-v2Text generation 3.1K
- mradermacher/Mercury-Acad-1.7B-i1-GGUF 3K
How to load it
from datasets import load_dataset
ds = load_dataset("nvidia/Nemotron-Post-Training-Dataset-v2")