Value Error
Final model path: C:\KnG\miniconda\envs\Gradio\HYM\downloaded_models\HY-Motion-1.0
Local environment detected. Loading model at startup.
Initializing T2MRuntime...
Loading model from C:\KnG\miniconda\envs\Gradio\HYM\downloaded_models\HY-Motion-1.0\config.yml...
Traceback (most recent call last):
File "C:\KnG\miniconda\envs\Gradio\HYM\gradio_app.py", line 912, in
demo = create_demo(final_model_path)
File "C:\KnG\miniconda\envs\Gradio\HYM\gradio_app.py", line 892, in create_demo
runtime = T2MRuntime(
File "C:\KnG\miniconda\envs\Gradio\HYM\hymotion\utils\t2m_runtime.py", line 87, in init
self.load()
File "C:\KnG\miniconda\envs\Gradio\HYM\hymotion\utils\t2m_runtime.py", line 122, in load
pipeline = load_object(
File "C:\KnG\miniconda\envs\Gradio\HYM\hymotion\utils\loaders.py", line 14, in load_object
obj = getattr(module, name)(**module_args)
File "C:\KnG\miniconda\envs\Gradio\HYM\hymotion\pipeline\motion_diffusion.py", line 461, in init
super().init(
File "C:\KnG\miniconda\envs\Gradio\HYM\hymotion\pipeline\motion_diffusion.py", line 132, in init
self._parse_buffer(self.motion_type)
File "C:\KnG\miniconda\envs\Gradio\HYM\hymotion\pipeline\motion_diffusion.py", line 142, in _parse_buffer
self.body_model = WoodenMesh()
File "C:\KnG\miniconda\envs\Gradio\HYM\hymotion\pipeline\body_model.py", line 229, in init
model = load_model_data(model_path)
File "C:\KnG\miniconda\envs\Gradio\HYM\hymotion\pipeline\body_model.py", line 98, in load_model_data
v_template_flat = np.frombuffer(f.read(), dtype=np.float32)
ValueError: buffer size must be a multiple of element size
Gradio
Windows
Python 3.10
RTX 2000 series
+1 同样的问题
To avoid compatibility issues, please make sure you’re using numpy==1.26.4(the same version we use).
我这边按照requirements又试了一遍，以及numpy==1.26.4依旧呈现上述问题@QingShuai
执行命令如下：python3 local_infer.py --model_path ckpts/tencent/HY-Motion-1.0-Lite --disable_duration_est --disable_rewrite
报错如下：
Traceback (most recent call last):
File "/cx8k/fs101/cy/avatar/HY-Motion-1.0/local_infer.py", line 395, in
main()
File "/cx8k/fs101/cy/avatar/HY-Motion-1.0/local_infer.py", line 341, in main
runtime = T2MRuntime(
File "/cx8k/fs101/cy/avatar/HY-Motion-1.0/hymotion/utils/t2m_runtime.py", line 87, in init
self.load()
File "/cx8k/fs101/cy/avatar/HY-Motion-1.0/hymotion/utils/t2m_runtime.py", line 140, in load
p = load_object(
File "/cx8k/fs101/cy/avatar/HY-Motion-1.0/hymotion/utils/loaders.py", line 14, in load_object
obj = getattr(module, name)(**module_args)
File "/cx8k/fs101/cy/avatar/HY-Motion-1.0/hymotion/pipeline/motion_diffusion.py", line 461, in init
super().init(
File "/cx8k/fs101/cy/avatar/HY-Motion-1.0/hymotion/pipeline/motion_diffusion.py", line 132, in init
self._parse_buffer(self.motion_type)
File "/cx8k/fs101/cy/avatar/HY-Motion-1.0/hymotion/pipeline/motion_diffusion.py", line 142, in _parse_buffer
self.body_model = WoodenMesh()
File "/cx8k/fs101/cy/avatar/HY-Motion-1.0/hymotion/pipeline/body_model.py", line 229, in init
model = load_model_data(model_path)
File "/cx8k/fs101/cy/avatar/HY-Motion-1.0/hymotion/pipeline/body_model.py", line 98, in load_model_data
v_template_flat = np.frombuffer(f.read(), dtype=np.float32)
ValueError: buffer size must be a multiple of element size
This error could be caused by missing Git LFS assets. Some files in the repository are managed by Git LFS, and it seems they might not have been fully downloaded to your local environment.
We realized that we missed including the LFS instructions in our README, and we appreciate you bringing this to our attention.
Could you please try running the following commands to see if it resolves the issue?
'''
git lfs install
git lfs pull
'''
This error could be caused by missing Git LFS assets. Some files in the repository are managed by Git LFS, and it seems they might not have been fully downloaded to your local environment.
We realized that we missed including the LFS instructions in our README, and we appreciate you bringing this to our attention.
Could you please try running the following commands to see if it resolves the issue?
'''
git lfs install
git lfs pull
'''
Yes, confirmed that after executing the above commands it is working properly now. Great work for the team. Thanks.