Ellen Zhong
Ellen Zhong
Some of the cryodrgn scripts accept --device [int] to specify which GPU to run on, but we should instead support the official torch.device spec: https://pytorch.org/docs/stable/tensor_attributes.html#torch.device
Now that we have github pages set up (thanks @vineetbansal!), we should move the installation and quickstart in the main repo's `README.md` there. We should also convert the [cryodrgn2 notion...
The default GPU memory usage for `cryodrgn abinit_het` maxes out at 40.5 GB when running on `D=128` images with the default batch size of 8 (at the final fourier radius...
I've gotten the above feature request for figure-making purposes. I think it would be nice additional visualization to show the PC interpolation path in the `pc*/umap.png` plot produced by `cryodrgn...
The default domain for `cryodrgn abinit_homo` and `cryodrgn abinit_het` should be `hartley` (currently set to `fourier`). When I made this change, we ran into the following error since we never...
Frequency marching settings (`--l-start`, `--l-end`) and translation search grid extent (`--t-extent`) are currently specified in units of pixels for ab initio reconstruction. We should convert this to Angstroms.
We should report the CryoDRGN version at the beginning of training
The default behavior is to load the whole dataset in memory for training, however for particularly large datasets that don't fit in memory, an option is currently provided (`--lazy`) to...
Reporting an issue with dataset loading in cyrodrgn v0.3.3 which added multiprocessing to the initial image preprocessing steps. Submitted command: ``` sbatch -n 1 -N 1 --mem 999G -t 24:00:00...
I want to generate volumes from a cryoDRGN model locally on my MacBook. I installed the nightly build of pytorch. Also had to make a few changes to loading weights...