v0.1.7

What's Changed

• refactor: replace torch.einsum with rigids_apply by @chungongyu in #381
• fix: autocast removal by @chungongyu in #383
• feat: add lr_scheduler by @chungongyu in #384

Full Changelog: v0.1.6...v0.1.7

v0.1.6

What's Changed

• fix: pae.center_bins by @chungongyu in #377
• refactor: autocast default dtype from env by @chungongyu in #378
• fix: autocast -> x.float() by @chungongyu in #379

Full Changelog: v0.1.5...v0.1.6

v0.1.5

What's Changed

• fix: multi_chain_permutation re-make torsion angles by @chungongyu in #367
• feat: add violation metrics by @chungongyu in #370
• fix: torch.nn.functional.scaled_dot_product_attention by @chungongyu in #371
• fix: torch.cuda.amp is deprecated by @chungongyu in #372
• refactor: wrap torch.amp.GradScaler by @chungongyu in #373
• feat: multiply the final loss of each training example by the square root of the number of residues after cropping. by @chungongyu in #376

Full Changelog: v0.1.4...v0.1.5

v0.1.4

What's Changed

• feat: multimer-style relpos by @chungongyu in #361
• refactor: let headers know if it is recycling. by @chungongyu in #362
• fix: symmetric_ground_truth_renaming by @chungongyu in #366

Full Changelog: v0.1.3...v0.1.4

v0.1.3

What's Changed

• fix: multi_chain_permutation, torch.clone by @chungongyu in #313

Full Changelog: v0.1.2...v0.1.3

v0.1.2

What's Changed

• refactor: add InputEmbedder by @chungongyu in #199
• fix: multi_chain_permutation by @chungongyu in #310

Full Changelog: v0.1.1...v0.1.2

v0.1.1

• DNA/RNA all-atoms supported.

v0.1.0

Initial release of ProFOLD 2.