---
license: other
license_name: nvidia-license
license_link: LICENSE
---

# Model Overview

## Description

NL-Diffusion-Image introduces a new paradigm for high-resolution text-to-image generation via LLM based on masked discrete diffusion over tokenized image patches.
Each image is encoded into a sequence of discrete tokens (using a 128K codebook/vocabulary), and generation proceeds through iterative parallel unmasking - similar to Diffusion LLMs.
We finetune from [Nemotron-Labs-Diffusion](https://huggingface.co/nvidia/Nemotron-Labs-Diffusion-8B) and introduce 2 key components:

* A token-editing mechanism that allows the model to revise already-unmasked tokens during inference.
* Grouped Cross-Entropy (GCE) objective to handle large-vocabulary training efficiently.

This model is for research and development only.

![Teaser Image](assets/teaser.png)

### License/Terms of Use

GOVERNING TERMS: Use of this model is governed by the [NVIDIA Open Source License Agreement](https://developer.download.nvidia.com/licenses/NVIDIA-OneWay-Noncommercial-License-22Mar2022.pdf?t=eyJscyI6ImdzZW8iLCJsc2QiOiJodHRwczovL3d3dy5nb29nbGUuY29tLyJ).

## Deployment Geography

Global

## Use Case

This model is intended for individuals who would like to research and evaluate text-to-image generation.

## Release Date

Hugging Face: 07/01/2026 via [HuggingFace](https://huggingface.co/nvidia/NL-Diffusion-Image).

## References

* [NL-Diffusion-Image Paper](https://arxiv.org/abs/2606.29814): Shufan Li et al., "Nemotron-Labs-Diffusion-Image: Advancing Masked Discrete Diffusion for High-Resolution Image Synthesis,".
* [Nemotron-Labs-Diffusion Paper](https://d1qx31qr3h6wln.cloudfront.net/publications/Nemotron_Diffusion_Tech_Report.pdf?VersionId=1tm4XZATEzGV7cs51XAf.xmWupU20vYW): Yonggan Fu et al., "Nemotron-Labs-Diffusion: A Tri-Mode Language Model Unifying Autoregressive, Diffusion, and Self-Speculation Decoding".
* [Emu3.5 Paper](https://arxiv.org/pdf/2510.26583): Emu3.5 team, "Emu3.5: Native Multimodal Models are World Learners".

## Model Architecture

**Architecture Type:** Neural Network <br>
**Network Architecture:** Masked Diffusion Transformer, IBQ tokenizer for visual encoding/decoding <br>
**Number of model parameters:** ~8B parameters <br>

We encode 16x16 image patches using a pretrained discrete tokenizer from Emu3.5, with a codebook size of 128k token IDs.
We expand the Nemotron-Labs-Diffusion vocabulary with a corresponding number of randomly-initialized embeddings, and fine-tune the model on a dataset of image/caption pairs.

## Computational Load (Internal Only: For NVIDIA Models Only)

Estimated using `Compute ≈ 6ND`, with `N` the number of trained parameters (encoder + training-only decoder ≈ 1.45B) and `D` the number of patch tokens processed during training.
With 58B tokens for pre-training, 255B tokens for Supervised Fine Tuning, `D ≈ 3.1×10¹¹` tokens (image and text included).

**Cumulative Compute**: ≈ 1.5e22 FLOPS. <br>
**Estimated Energy**: ≈ 53000 kWh. <br>
**Estimated Emissions for Model Training**: ≈ 22.4 tCO2e. <br>

## Input

**Input Type(s):** Text <br>
**Input Format(s):** Character Strings <br>
**Input Parameters:** One-dimensional (1D) <br>
**Other Properties Related to Input:** Maximum prompt length is 900 tokens. <br>

## Output

**Output Type(s):** Images <br>
**Output Format:** Tensor (3xHxW) <br>
**Output Parameters:** Two-Dimensional (2D) <br>
**Other Properties Related to Output:** The output represents an RGB image. <br>

## Software Integration

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.

**Runtime Engine(s):**
* PyTorch <br>


**Supported Hardware Microarchitecture Compatibility:** <br>
* NVIDIA Ampere <br>
* NVIDIA Blackwell <br>
* NVIDIA Jetson <br>
* NVIDIA Hopper <br>
* NVIDIA Lovelace <br>
* NVIDIA Pascal <br>
* NVIDIA Turing <br>
* NVIDIA Volta <br>

**Supported Operating System(s):** <br>
* Linux
* Linux 4 Tegra
* QNX
* Windows

The integration of foundation and fine-tuned models into AI systems requires additional testing using use-case-specific data to ensure safe and effective deployment.
Following the V-model methodology, iterative testing and validation at both unit and system levels are essential to mitigate risks, meet technical and functional requirements, and ensure compliance with safety and ethical standards before deployment.

This AI model can be embedded as an Application Programming Interface (API) call into the software environment described above.

## Model Version(s)

* NL-Diffusion-Image (8B).

**Links:**

* https://huggingface.co/nvidia/NL-Diffusion-Image

# Training and Evaluation Datasets

## Training Dataset

**LAION-115M-Clean Recaptioned**

A large, filtered collection of diverse real-world web images - including people, animals, objects, scenes, artwork, products, documents, and memes - paired with high-quality synthetic text captions generated from the images.

**Data Modality:**
* Multimodal (image, text)

**Image Training Data Size:**
* 115M samples

**Data Collection Method by dataset:**
* Automated

**Labeling Method by dataset:**
* Synthetic: Subset of 8M images recaptioned using Qwen3-VL

**MidJourney v6 520k Recaptioned**

High-quality synthetic Midjourney v6 images from user prompts.

**Data Modality:**
* Multimodal (image, text)

**Image Training Data Size:**
* 520k samples

**Data Collection Method by dataset:**
* Automated

**Labeling Method by dataset:**
* Synthetic: Images recaptioned using Qwen3-VL

**COYO700M Recaptioned**

Image–alt-text pairs from HTML pages.

**Data Modality:**
* Multimodal (image, text)

**Image Training Data Size:**
* 700M samples

**Data Collection Method by dataset:**
* Automated

**Labeling Method by dataset:**
* Synthetic: Subset of 24M images recaptioned using Qwen3-VL

**BLIP3o-60k Recaptioned**

**Data Modality:**
* Multimodal (image, text)

**Image Training Data Size:**
* 520k samples

**Data Collection Method by dataset:**
* Automated

**Labeling Method by dataset:**
* Synthetic: Images recaptioned using Qwen3-VL

## Evaluation Datasets

**ImageNet**

Labeled natural images of objects and scenes organized by WordNet categories.

**Link:**
* [ImageNet](https://www.image-net.org/)

**Data Modality:**
* Images

**Data Collection:**
* Automated

**Labeling Method:**
* Manually-Collected

**Training Images:**
* 1,281,167

**Validation Images:**
* 50,000

**GenEval**

Prompts for testing objects, attributes, counts, colors, and spatial relations.

**Link:**
* [GenEval](https://github.com/djghosh13/geneval)

**Data Modality:**
* Text

**Data Collection:**
* Manually-Collected

**Labeling Method:**
* Manually-Collected

**Captions/annotations:**
* 553 samples

**DPGBench**

Dense prompts with multiple objects, attributes, and relationships.

**Link:**
* [DPGBench](https://github.com/TencentQQGYLab/ELLA)

**Data Modality:**
* Text

**Data Collection:**
* Manually-Collected

**Labeling Method:**
* Manually-Collected

**Captions/annotations:**
* 1065 samples

**MJHQ-30K**

High-quality photorealistic Midjourney-style images used for aesthetic/FID evaluation.

**Link:**
* [MJHQ-30K](https://huggingface.co/datasets/playgroundai/MJHQ-30K/blob/main/README.md)

**Data Modality:**
* Multimodal (image, text)

**Data Collection:**
* Manually-Collected

**Labeling Method:**
* Automated

**CaptionsImages:**
* 30k samples

## GenEval Benchmark


| Model      | Params | Single Object | Two Objects | Counting | Colors | Position | Color Attri. | Overall |
|------------|---------|-------------|-----------|-------------|--------|----------|--------------|---------|
| Qwen-Image-2507 | 20B | 0.99 | 0.92 | 0.89 | 0.88 | 0.76 | 0.77 | 0.87 |
| Nemotron-Labs-Diffusion-Image | 8B | 0.98 | 0.93 | 0.83 | 0.94 | 0.88 | 0.82 | 0.90 |

## Text-to-Image Generation Performance on DPG Benchmark and MJHQ-30k Dataset

| Model      | Params | Codebook | DPG | MJHQ FID | MJHQ  HPSv3 |
|------------|--------|-----------|----|----------|-------------|
| MMaDa      | 8B     | 8,192 | 53.4 | 32.85 | 5.43 |
| LaViDa-O   | 10B    | 8,192 | 81.8 | 6.68 | 8.81 |
| Nemotron-Labs-Diffusion-Image | 8B | 131,072 | 85.2 | 6.46 | 9.61|
| Nemotron-Labs-Diffusion-Image*</sup>  | 8B | 131,072 | 86.9 | 12.23 | 10.76 |

<sup>*</sup> Finetuned on 6M synthetic data for better image quality


## Key Considerations

This model can generate synthetic images and may produce content that is inaccurate, offensive, or otherwise inappropriate. Users should implement robust safety guardrails — including content filtering, abuse monitoring, and access controls— to reduce the risk of harmful outputs. Users are responsible for ensuring that their use of the model complies with all applicable laws and regulations, and for regularly reviewing and updating their guardrails as risks evolve. For more information about the implementation of guardrails, please see the [Nemotron-3.5-Content-Safety Guardrail Model](https://huggingface.co/nvidia/Nemotron-3.5-Content-Safety).

## Inference

**Acceleration Engine:** TensorRT, TensorRT-LLM <br>
**Engine:** PyTorch <br>
**Test Hardware:** NVIDIA Hopper (H100) <br>

## 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. 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.

Please make sure you have proper rights and permissions for all input image and video content; if image or video includes people, personal health information, or intellectual property, the image or video generated will not blur or maintain proportions of image subjects included.

Users are responsible for model inputs and outputs. Users are responsible for ensuring safe integration of this model, including implementing guardrails as well as other safety mechanisms, prior to deployment.

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


## Quick Start 

```
import torch
from transformers import AutoModelForCausalLM, PreTrainedTokenizerFast

model_path = "shufanlNvidia/NLD-Diffusion-Image-8B-Internal"
tokenizer = PreTrainedTokenizerFast.from_pretrained(model_path)
model = AutoModelForCausalLM.from_pretrained(
    model_path,
    trust_remote_code=True,
    torch_dtype=torch.bfloat16,
)
model = model.to("cuda").eval()
model.config.dlm_paradigm = "bidirectional"
PROMPT = (
    "A full-body shot of hyper-realistic female cyborg, human facial skin seamlessly integrated with a glossy white mechanical head shell. "
    "Features a realistic human ear, blue eyes. bright, outdoor, background with blue sky, illuminated by striking bright white studio lighting, "
    "casting soft shadows. Cyberpunk aesthetic, high-tech minimalism, shot on 85mm lens, photorealistic, Unreal Engine 5 render, intricately detailed, "
    "8k resolution, high dynamic range, chest with whit armor plate, cute, beautiful, sexy, glossy surface, reflective, Artstation, pixiv, no hair, "
    "3D render, stylized eyesz"
)
torch.manual_seed(42)

image = model.text_to_image(
    PROMPT,
    tokenizer=tokenizer,
    image_resolution=1024,
    n_tokens=(1024 // 16) * (1024 // 16),
    guidance_scale=5.0,
    temperature=0.86,
    n_steps=64,
    schedule="shift",
    shift=5,
    confidence_policy="mmada",
    schedule_temp="linear",
    alg_temp=1.0,
    dynamic_temperature=False,
    min_temperature=0.01,
    edit_threshold=0.6,
    micro_cond="ORIGINAL WIDTH : 1024; ORIGINAL HEIGHT : 1024; TOP : 0; LEFT : 0; SCORE : 6.520; HPS: 3.220",
    block_policy=2,
    is_legacy=False,
    use_cache=False,
)
image.save("output.webp")
```

See [demo_inference_release.py](demo_inference_release.py) for more details of inference arguments.