search

Graylab / IgFold

Fast, accurate antibody structure prediction from deep learning on massive set of natural antibodies
Other
325 stars 61 forks source link

No igfold_4.ckpt ? #23

Closed shubihu closed 1 year ago

shubihu commented 1 year ago

I downloaded the model weight, but there is no igfold_4.ckpt ?
The link is https://data.graylab.jhu.edu/IgFold.tar.gz

WenTao-tust commented 1 year ago

How do you install IgFold ? I keep having trouble installing it on my computer.

jeffreyruffolo commented 1 year ago

@shubihu, this is expected you should only have four models in the download (1, 2, 3, and 5). The numbers correspond to the random seeds used during training. The training run beginning with seed 4 was unstable, so we used 5 instead to get an ensemble of four models.

shubihu commented 1 year ago

How do you install IgFold ? I keep having trouble installing it on my computer. Just via pip like this

image
shubihu commented 1 year ago

@shubihu, this is expected you should only have four models in the download (1, 2, 3, and 5). The numbers correspond to the random seeds used during training. The training run beginning with seed 4 was unstable, so we used 5 instead to get an ensemble of four models.

OK, thanks.