Which official extension?
lightningpixel/modly-hunyuan3d-mini-extension
Describe the bug
Extension installation fails because setup.py pins torch==2.7.0 with an --index-url of https://download.pytorch.org/whl/cu128, but that exact version is no longer served from that index. Only 2.9.0+cu128 and newer are currently available.
Error
Error: setup.py exited with code 1
ERROR: Could not find a version that satisfies the requirement torch==2.7.0
(from versions: 2.9.0+cu128, 2.9.1+cu128, 2.10.0+cu128, 2.11.0+cu128)
ERROR: No matching distribution found for torch==2.7.0
Full traceback:
Traceback (most recent call last):
File "", line 93, in
...
File "/home/dsusco/Modly/extensions/hunyuan3d-mini/setup.py", line 173, in setup
pip(venv, "install", *torch_pkgs, "--index-url", torch_index)
File "/home/dsusco/Modly/extensions/hunyuan3d-mini/setup.py", line 75, in pip
subprocess.run([str(pip_exe), *args], check=True)
subprocess.CalledProcessError: Command
'['/home/dsusco/Modly/extensions/hunyuan3d-mini/venv/bin/pip', 'install',
'torch==2.7.0', 'torchvision==0.22.0', '--index-url',
'https://download.pytorch.org/whl/cu128']' returned non-zero exit status 1.
Expected Behavior
setup.py's torch/CUDA version table should resolve to a torch build that
is currently hosted on the PyTorch index for the reported cuda_version.
Additional Context
Also seeing the same class of failure (hardcoded old torch pin against an aged-out CUDA index) on the trellis2-gguf extension.
Steps to reproduce
- Install Modly v0.4.2 AppImage
- Go to Models > Install from GitHub
- Enter https://github.com/lightningpixel/modly-hunyuan3d-mini-extension
- Installation runs setup.py, which attempts pip install torch==2.7.0
torchvision==0.22.0 --index-url https://download.pytorch.org/whl/cu128
Modly version
0.4.2
Operating system
Fedora Linux 43
GPU
GA107GL [RTX A400]
Screenshots & logs
No response
Confirmation
Which official extension?
lightningpixel/modly-hunyuan3d-mini-extension
Describe the bug
Extension installation fails because setup.py pins torch==2.7.0 with an --index-url of https://download.pytorch.org/whl/cu128, but that exact version is no longer served from that index. Only 2.9.0+cu128 and newer are currently available.
Error
Error: setup.py exited with code 1
ERROR: Could not find a version that satisfies the requirement torch==2.7.0
(from versions: 2.9.0+cu128, 2.9.1+cu128, 2.10.0+cu128, 2.11.0+cu128)
ERROR: No matching distribution found for torch==2.7.0
Full traceback:
Traceback (most recent call last):
File "", line 93, in
...
File "/home/dsusco/Modly/extensions/hunyuan3d-mini/setup.py", line 173, in setup
pip(venv, "install", *torch_pkgs, "--index-url", torch_index)
File "/home/dsusco/Modly/extensions/hunyuan3d-mini/setup.py", line 75, in pip
subprocess.run([str(pip_exe), *args], check=True)
subprocess.CalledProcessError: Command
'['/home/dsusco/Modly/extensions/hunyuan3d-mini/venv/bin/pip', 'install',
'torch==2.7.0', 'torchvision==0.22.0', '--index-url',
'https://download.pytorch.org/whl/cu128']' returned non-zero exit status 1.
Expected Behavior
setup.py's torch/CUDA version table should resolve to a torch build that
is currently hosted on the PyTorch index for the reported cuda_version.
Additional Context
Also seeing the same class of failure (hardcoded old torch pin against an aged-out CUDA index) on the trellis2-gguf extension.
Steps to reproduce
torchvision==0.22.0 --index-url https://download.pytorch.org/whl/cu128
Modly version
0.4.2
Operating system
Fedora Linux 43
GPU
GA107GL [RTX A400]
Screenshots & logs
No response
Confirmation