Refine model card overview
#16
by qianlim - opened
README.md
CHANGED
|
@@ -22,6 +22,16 @@ tags:
|
|
| 22 |
|
| 23 |
## Description
|
| 24 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 25 |
Cosmos3 is a collection of Omnimodal world models capable of generating dynamic, high-quality video, image, audio, and action commands from combinations of text, image, video, and action trajectory inputs. It serves as a foundational building block for a broad range of Physical AI applications and research spanning world understanding, world generation, simulation, and embodied policy learning.
|
| 26 |
|
| 27 |
This model is ready for commercial and non-commercial use.
|
|
|
|
| 22 |
|
| 23 |
## Description
|
| 24 |
|
| 25 |
+
Cosmos3-Super-Text2Image-4Step is a 4-step distilled version of the base [Cosmos3-Super-Text2Image](https://huggingface.co/nvidia/Cosmos3-Super-Text2Image) model. Given a text prompt, it generates a high-fidelity image.
|
| 26 |
+
|
| 27 |
+
- **What's different from the base model:** Distilled into a fixed 4-step sampling schedule without the need for classifier-free guidance. At the recommended 50-step + CFG setting for the Text2Image model, this replaces 50 × 2 = 100 model evaluations with 4, leading up to 25× fewer diffusion-model evaluations per image (see [PBR](https://huggingface.co/nvidia/Cosmos3-Super-Text2Image-4Step#pbr-performance-benchmark-reporting) for latency estimates).
|
| 28 |
+
- **Quality:** At just 4 steps, ranks #2 among open-source models on the Artificial Analysis Text-to-Image leaderboard, within one place of the full-step Cosmos3-Super-Text2Image (as of 2026/07/17; see [Benchmarks section](https://huggingface.co/nvidia/Cosmos3-Super-Text2Image-4Step#benchmarks)).
|
| 29 |
+
- **Input / Output:** Text prompt in → PNG image out. Supports various aspect ratios and resolutions.
|
| 30 |
+
- **Example usage and output:** See [Usage](https://huggingface.co/nvidia/Cosmos3-Super-Text2Image-4Step#usage-run-inference-with-vllm-omni) section.
|
| 31 |
+
- **Hardware:** The 64B model needs a multi-GPU H100/H200 node (4–8 GPUs) or a single B200; it does not fit on a single smaller GPU. See [Usage](https://huggingface.co/nvidia/Cosmos3-Super-Text2Image-4Step#usage-run-inference-with-vllm-omni). Full model details below.
|
| 32 |
+
|
| 33 |
+
Cosmos3-Super-Text2Image-4Step was developed by NVIDIA as a part of Cosmos3.
|
| 34 |
+
|
| 35 |
Cosmos3 is a collection of Omnimodal world models capable of generating dynamic, high-quality video, image, audio, and action commands from combinations of text, image, video, and action trajectory inputs. It serves as a foundational building block for a broad range of Physical AI applications and research spanning world understanding, world generation, simulation, and embodied policy learning.
|
| 36 |
|
| 37 |
This model is ready for commercial and non-commercial use.
|