argmaxinc

/

mlx-stable-diffusion-3.5-large-4bit-quantized

like 3

Follow

Argmax 82

I am using pinned Python 3.10 and Pytorch 2.4.0 versions to avoid compatibility issues.

Not using conda, but venv instead

python3.10 -m venv .venv
source .venv/bin/activate
pip install "torch==2.4.0" diffusionkit

diffusionkit-cli --prompt "detailed cinematic dof render of a \
detailed MacBook Pro on a wooden desk in a dim room with items \
around, messy dirty room. On the screen are the letters 'SD3 on \
DiffusionKit' glowing softly. High detail hard surface render" \
--model-version argmaxinc/mlx-stable-diffusion-3.5-large-4bit-quantized \
--height 768 \
--width 1360 \
--seed 1001 \
--step 50 \
--cfg 7 \
--t5 \
--output ~/Desktop/sd3_on_mac.png

INFO:diffusionkit.mlx.scripts.generate_images:Output image resolution will be 768x1360
INFO:diffusionkit.mlx:Pre text encoding peak memory: 0.0GB
INFO:diffusionkit.mlx:Pre text encoding active memory: 0.0GB
INFO:diffusionkit.mlx:Post text encoding peak memory: 2.628GB
INFO:diffusionkit.mlx:Post text encoding active memory: 1.786GB
INFO:diffusionkit.mlx:Text encoding time: 2.235s
INFO:diffusionkit.mlx:Pre denoise peak memory: 0.0GB
INFO:diffusionkit.mlx:Pre denoise active memory: 0.018GB
INFO:diffusionkit.mlx:Seed: 1001
  0%|                                                                                                                                                                                      | 0/50 [00:00<?, ?it/s]Traceback (most recent call last):
  File "/Users/wavecut/Downloads/diffusionkit/.venv/bin/diffusionkit-cli", line 8, in <module>
    sys.exit(cli())
  File "/Users/wavecut/Downloads/diffusionkit/.venv/lib/python3.10/site-packages/diffusionkit/mlx/scripts/generate_images.py", line 174, in cli
    image, _ = sd.generate_image(
  File "/Users/wavecut/Downloads/diffusionkit/.venv/lib/python3.10/site-packages/diffusionkit/mlx/__init__.py", line 421, in generate_image
    latents, iter_time = self.denoise_latents(
  File "/Users/wavecut/Downloads/diffusionkit/.venv/lib/python3.10/site-packages/diffusionkit/mlx/__init__.py", line 286, in denoise_latents
    latent, iter_time = sample_euler(
  File "/Users/wavecut/Downloads/diffusionkit/.venv/lib/python3.10/site-packages/diffusionkit/mlx/__init__.py", line 772, in sample_euler
    model.cache_modulation_params(extra_args.pop("pooled_conditioning"), timesteps)
  File "/Users/wavecut/Downloads/diffusionkit/.venv/lib/python3.10/site-packages/diffusionkit/mlx/__init__.py", line 682, in cache_modulation_params
    self.model.mmdit.cache_modulation_params(
  File "/Users/wavecut/Downloads/diffusionkit/.venv/lib/python3.10/site-packages/diffusionkit/mlx/mmdit.py", line 177, in cache_modulation_params
    logger.info(f"Cached modulation_params for timesteps={timesteps}")
TypeError: unsupported format string passed to mx.array.__format__
  0%|