Releases: steineggerlab/alphafold
Releases · steineggerlab/alphafold
v2.3.6
- Fix issues with Jax >= 0.4 creating NaNs
- Fix broken SCOPData import in BioPython 1.82
- Move tensorflow-cpu dependency to ColabFold
v2.3.5
- Update OpenMM imports to work with new OpenMM API (>7.7)
v2.3.4
- move tolerance compute inside module.py
- cast outputs to np.asarray(?,np.float16) to avoid memory leaks
v2.3.3
- fix memory leaks
- use bfloat16 for representations
- move confidence compute inside module.py
2.3.2
- (bugfix) making sure different random seed is used for each recycle, is important for dropouts.
- moving random seed generation from modules_multimer.py to model.py
- raw pAE features (asym_id, logits, breaks) are now returned
2.3.1
Merge pull request #3 from sokrypton/main
v2.3.1
2.1.12
Fix jax NaN issue for version 3.8
2.1.11
Update AlphFold to release 2.2.2
2.1.10
Improve compile speed by ~5 times