Text-to-Image
Transformers
Safetensors
Hunyuan
text-generation
hunyuan
quantization
nf4
comfyui
custom nodes
autoregressive
Dit
HunyuanImage-3.0
instruct
image-editing
bitsandbytes
4bit
custom_code
4-bit precision
Instructions to use EricRollei/HunyuanImage-3.0-Instruct-NF4 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use EricRollei/HunyuanImage-3.0-Instruct-NF4 with Transformers:
# Load model directly from transformers import AutoModelForCausalLM model = AutoModelForCausalLM.from_pretrained("EricRollei/HunyuanImage-3.0-Instruct-NF4", trust_remote_code=True, dtype="auto") - Notebooks
- Google Colab
- Kaggle
Testing file assumes windows user
#1
by TobbeBobbe2346235425643 - opened
Just a tiny remark: If you want to use test file on Linux, make sure to change the path to "." for current directory, line 9 I believe.
On another side note: I can't get this to work on RTX 6000 Pro after trying lots of different transformers versions. Not really sure what else to test currently after a few hours of troubleshooting. It throws me different error messages, missing modules like _hunyuan_t5_compat_patched, _tkwrapper, transformers_modules.HunyuanImage-3, siglip2. Trying it on runpod, with a cuda 13.0 pod but in a venv. I have tried various loaders for the troubleshooting. Not really sure what to test now.