kpms.fit_pca error

[CoderYangQi]

##Hi, thanks for your code! When I run the code(docs\keypoint_moseq_colab.ipynb) with dlc_example_project, I got the error in kpms.fit_pca. Could you help me?:

File "D:\USERS\yq\code\MotionTracking\keypoint-moseq\YQScripts\colabTest.py", line 63, in
pca = kpms.fit_pca(**data, **config())
File "d:\Tools\Anaconda_envs\envs\keypoint_moseq\lib\site-packages\jax_moseq\models\keypoint_slds\alignment.py", line 312, in fit_pca
Y_flat = preprocess_for_pca(
File "d:\Tools\Anaconda_envs\envs\keypoint_moseq\lib\site-packages\jax_moseq\models\keypoint_slds\alignment.py", line 388, in preprocess_for_pca
Y_aligned, v, h = align_egocentric(Y, anterior_idxs, posterior_idxs)
File "d:\Tools\Anaconda_envs\envs\keypoint_moseq\lib\site-packages\jax_moseq\models\keypoint_slds\alignment.py", line 429, in align_egocentric
Y_aligned = inverse_rigid_transform(Y, v, h)
File "d:\Tools\Anaconda_envs\envs\keypoint_moseq\lib\site-packages\jax_moseq\models\keypoint_slds\alignment.py", line 172, in inverse_rigid_transform
return apply_rotation(Y - v[..., na, :], -h)
File "d:\Tools\Anaconda_envs\envs\keypoint_moseq\lib\site-packages\jax_moseq\models\keypoint_slds\alignment.py", line 214, in apply_rotation
return jnp.einsum("...kj,...ij->...ki", Y, rot_matrix)
File "d:\Tools\Anaconda_envs\envs\keypoint_moseq\lib\site-packages\jax_src\numpy\lax_numpy.py", line 2923, in einsum
operands, contractions = contract_path(
File "d:\Tools\Anaconda_envs\envs\keypoint_moseq\lib\site-packages\opt_einsum\contract.py", line 324, in contract_path
raise ValueError(
ValueError: Size of label 'd' for operand 1 (70) does not match previous terms (10030).

Your Environment

• Version/GitHub branch used:
• Device: GPU 3090
• OS: Windows Python3.9 (conda install, window-gpu)
• Link to your project:

[calebweinreb]

Hi,

This is the repo for depth MoSeq... Issues related to keypoint moseq should be raised here https://github.com/dattalab/keypoint-moseq

Also, the colab notebook is mostly meant for google colab. I would follow the tutorial if working on your own computer. To debug this, I would recommend running the colab notebook on google colab. The error shouldn't occur there, and then you check if there was some issue with ho you loaded the data which produced the error above.