Nemotron H 47B Base 8K

🚀 Nemotron-H-47B-Base-8K

NVIDIA's large language model designed for text completion, supporting multiple languages and offering high performance on NVIDIA GPU-accelerated systems.

🚀 Quick Start

NVIDIA Nemotron-H-47B-Base-8K is a large language model (LLM) developed by NVIDIA, serving as a completion model for given text. It utilizes a hybrid model architecture mainly composed of Mamba - 2 and MLP layers, combined with just five Attention layers. Pruned and distilled from Nemotron-H-56B-Base-8K using 63B tokens, it features an 8K context length. Supported languages include English, German, Spanish, French, Italian, Korean, Portuguese, Russian, Japanese, and Chinese. For more detailed information on the model architecture, training, and evaluation, refer to the project page and the technical report.

To achieve the best performance on a given task, users are recommended to customize the model using the NeMo Framework suite of customization tools, including Parameter - Efficient Fine - Tuning (P - tuning, Adapters, LoRA, etc.), and Model Alignment (SFT, SteerLM, RLHF, etc.) with NeMo - Aligner.

This model is for research and development purposes only.

This model is part of the Nemotron - H Collection. You can find other models in this family here:

• Nemotron-H-56B-Base-8K
• Nemotron-H-47B-Base-8K
• Nemotron-H-8B-Base-8K

✨ Features

• Hybrid Architecture: Combines Mamba - 2 and MLP layers with five Attention layers.
• Multilingual Support: Supports German, Spanish, French, Italian, Korean, Portuguese, Russian, Japanese, Chinese, and English.
• 8K Context Length: Allows for longer input sequences.
• Customizable: Can be customized using the NeMo Framework.

📚 Documentation

License/Terms of Use

GOVERNING TERMS: Use of this model is governed by the NVIDIA Internal Scientific Research and Development Model License.

Model Developer: NVIDIA

Model Dates: October 2024 - March 2025

Data Freshness: September 2024. The pretraining data has a cutoff date of September 2024.

Use Case

This model is intended for developers and researchers building LLMs.

Release Date

4/12/2025

References

• [2504.03624] Nemotron-H: A Family of Accurate and Efficient Hybrid Mamba-Transformer Models

Model Architecture

Input

Output

Our AI models are designed and/or optimized to run on NVIDIA GPU - accelerated systems. By leveraging NVIDIA’s hardware (e.g., GPU cores) and software frameworks (e.g., CUDA libraries), the model achieves faster training and inference times compared to CPU - only solutions.

Software Integration

Model Version

• v1.0

Prompt Format

As this is a base model, no explicit prompt format is recommended or required.

💻 Usage Examples

Basic Usage

import torch
from transformers import AutoTokenizer, AutoModelForCausalLM

# Load the tokenizer and model
tokenizer  = AutoTokenizer.from_pretrained("nvidia/Nemotron-H-47B-Base-8K", trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained("nvidia/Nemotron-H-47B-Base-8K", torch_dtype=torch.bfloat16, trust_remote_code=True, device_map="auto")

prompt = "When was NVIDIA founded?"

outputs = model.generate(**tokenizer(prompt, return_tensors="pt", add_special_tokens=False).to(model.device))
print(tokenizer.decode(outputs[0]))

🔧 Technical Details

Training, Testing, and Evaluation Datasets

Training & Testing Datasets

The training corpus for Nemotron - H - 47B - Base - 8K consists of English and multilingual text (German, Spanish, French, Italian, Korean, Portuguese, Russian, Japanese, Chinese, and English), as well as code. Our sources cover a variety of document types such as webpages, dialogue, articles, and other written materials. This model was also improved using synthetic data from Qwen (Built with Qwen). The corpus spans domains including legal, math, science, finance, and more. We also include a small portion of question - answering, and alignment style data to improve model accuracies.

Data Collection for Training & Testing Datasets: Hybrid: Automated, Human, Synthetic

Data Labeling for Training & Testing Datasets: Hybrid: Automated, Human, Synthetic

Evaluation Datasets

We used the datasets listed in the next section to evaluate the model.

Data Collection for Training Datasets: Hybrid: Automated, Human, Synthetic

Data Labeling for Training Datasets: Hybrid: Automated, Human, Synthetic

Commonsense Understanding Evaluations

• ARC (Ai2 reasoning challenge) - Challenge: The challenge set of questions from a benchmark that contains grade - school level, multiple - choice science questions to assess question answering ability of language models. Dataset
• Hellaswag: Tests the ability of a language model to correctly finish the provided context from a choice of possible options. Dataset
• Winogrande: Tests the ability to choose the right option for a given sentence which requires commonsense reasoning. Dataset
• CommonsenseQA: A multiple - choice question answering dataset that requires different type of commonsense knowledge to predict the correct answers. Dataset

Coding Evaluations

• MBPP (Mostly Basic Python Programming Problems): Evaluates ability to generate solutions for Python programming tasks. [Dataset](https://github.com/google - research/google - research/tree/master/mbpp)
• MBPP+: Extended version of MBPP with additional validation. Dataset
• HumanEval: Tests code generation and completion abilities in Python. [Dataset](https://github.com/openai/human - eval)

Math Evaluations

• GSM8K (Grade School Math 8K): Evaluates grade school level mathematical word problem solving. [Dataset](https://github.com/openai/grade - school - math)
• MATH - 500: Tests advanced mathematical problem solving across algebra, geometry, and calculus. [Dataset](https://huggingface.co/datasets/HuggingFaceH4/MATH - 500)
• MATH Lvl 5: Only the most difficult questions from the MATH dataset. Dataset
• MATH - 500: Tests advanced mathematical problem solving across algebra, geometry, and calculus. [Dataset](https://huggingface.co/datasets/HuggingFaceH4/MATH - 500)

General Evaluations

• MMLU: Tests knowledge across 57 subjects including science, humanities, math and more. Dataset
• MMLU Pro: Evaluates language understanding models across a broad range of challenging, reasoning - focused questions across 14 diverse domains. [Dataset](https://huggingface.co/datasets/TIGER - Lab/MMLU - Pro)

Potential Known Risks for Usage

The model was trained on data that contains toxic language, unsafe content, and societal biases originally crawled from the internet. Therefore, the model may amplify those biases and return toxic responses especially when prompted with toxic prompts. The model may generate answers that may be inaccurate, omit key information, or include irrelevant or redundant text producing socially unacceptable or undesirable text, even if the prompt itself does not include anything explicitly offensive.

The model demonstrates weakness to indirect prompt injection via some encodings, including Base16, Hex/ASCII, and Braille, though is more resilient than other similar models to injections using the more common Base64 vector.

Inference

Ethical Considerations

NVIDIA believes Trustworthy AI is a shared responsibility and we have established policies and practices to enable development for a wide array of AI applications. When downloaded or used in accordance with our terms of service, developers should work with their internal model team to ensure this model meets requirements for the relevant industry and use case and addresses unforeseen product misuse.

For more detailed information on ethical considerations for this model, please see the Responsible Use Guide available at http://nvidia.com/nemotron - responsible - use.

Please report security vulnerabilities or NVIDIA AI Concerns [here](https://www.nvidia.com/en - us/support/submit - security - vulnerability/).

📄 License

Use of this model is governed by the NVIDIA Internal Scientific Research and Development Model License.