Skip to content

Dataset · nvidia

Specialized synthetic data from NVIDIA

nvidia/Nemotron-Pretraining-Specialized-v1.1

Large corpus of synthetic texts on code, logic, and economics for LLM pre-training.

What's inside

The dataset is part of the pre-training data collection for the NVIDIA Nemotron 3 model family. It contains synthetic text data covering programming concepts and algorithms, formal logic, economics, and multiple-choice questions. The corpus size ranges from 10 million to 100 million records.

What you can do with it

  • Pre-training and fine-tuning language models to enhance logical reasoning
  • Improving LLM skills in algorithms, economics, and test-taking
Tasks
Text generation
Size
10M–100M rows
Downloads this month
4.7K
Likes
45

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("nvidia/Nemotron-Pretraining-Specialized-v1.1")