Views
No views yet
HiDream-I1 quantized to 4 bits, allowing the full model to run in less than 16GB of VRAM.HiDream-I1is a new open-source image generative foundation model with 17B parameters that achieves state-of-the-art image generation quality within seconds.
| Name | Min VRAM | Steps | HuggingFace |
|---|---|---|---|
| HiDream-I1-Full | 16 GB | 50 | 🤗 Original / NF4 |
| HiDream-I1-Dev | 16 GB | 28 | 🤗 Original / NF4 |
| HiDream-I1-Fast | 16 GB | 16 | 🤗 Original / NF4 |
>= Ampere (e.g. A100, H100, A40, RTX 3090, RTX 4090)>= 16 GB>= 16 GBpip install hdi1 --no-build-isolation[!NOTE] It's recommended that you start a new python environment for this package to avoid dependency conflicts.
To do that, you can useconda create -n hdi1 python=3.12and thenconda activate hdi1.
Or you can usepython3 -m venv venvand thensource venv/bin/activateon Linux orvenv\Scripts\activateon Windows.
1python -m hdi1 "A cat holding a sign that says 'hello world'"
2
3# or you can specify the model
4python -m hdi1 "A cat holding a sign that says 'hello world'" -m fast[!NOTE] The inference script will try to automatically downloadmeta-llama/Llama-3.1-8B-Instructmodel files. You need to agree to the license of the Llama model on your HuggingFace account and login usinghuggingface-cli loginin order to use the automatic downloader.
python -m hdi1.web