search

gelles-brandeis / tapqir

Bayesian analysis of co-localization single-molecule microscopy image data.
Apache License 2.0
4 stars 0 forks source link

CUDA out of memory #359

Closed jc-brandeis closed 1 year ago

jc-brandeis commented 2 years ago

I get this error with 113 AOIs, 2 channels, and 1169 frames COSMOS-HMM fitting. The models converges but will not write the matlab parameter file.

87% 87000/100000 [44:50:51<6:33:18, 1.82s/it] /home/jchung/anaconda3/envs/tapqir-env/lib/python3.8/site-packages/torch/distributions/gamma.py:71: UserWarning: Specified kernel cache directory could not be created! This disables kernel caching. Specified directory is /home/jchung/.cache/torch/kernels. This warning will appear only once per process. (Triggered internally at ../aten/src/ATen/native/cuda/jit_utils.cpp:860.) self.rate * value - torch.lgamma(self.concentration)) Iteration #87000 model converged. Fitting the data: Done Computing stats ...

ordabayevy commented 2 years ago

Hi Johnson, can you try using smaller AOI batch size? And since fitting is already done you can set the number of iterations to 1 so that it immediately computes stats.

gelles-brandeis commented 1 year ago

I think this one is corrected in more recent versions of tapqir.