search

hpcaitech / FastFold

Optimizing AlphaFold Training and Inference on GPU Clusters
Apache License 2.0
562 stars 86 forks source link

inference with lower memory usage #149

Closed double-vin closed 1 year ago

double-vin commented 1 year ago
  1. Add drop_unused parameter for scatter and split function, and use empty cache() to clear memory_reserved.
  2. Use the chunk operation on the _gather function. This method can be used for monomer.
Gy-Lu commented 1 year ago

Hi, there was something wrong with our CI. And I have just fixed it. Sorry to say that, but it is necessary to rebase your code to the lastest main branch, to use the fixed CI. Sorry again :(