search

hpcaitech / FastFold

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

Re-organize file and align results #99

Closed oahzxl closed 1 year ago

oahzxl commented 1 year ago
  1. Move template modification from nn to fastnn. Details can be seen in this pr https://github.com/hpcaitech/FastFold/pull/97.
  2. Fix some bugs to align results. Max error of alphafold's output is less than 5e-4 compared with openfold. And update some unit tests. Details can be seen in this pr https://github.com/hpcaitech/FastFold/pull/98.