Kakao Buffalo Versions Save

WARNING: THIS IS BROKEN VERSION

Bug Fixes

• Fix failure with creating a large database
• Fix bug that get_most_similar_item returns Nan values.

Misc

• Clean Code(Linting, Refactoring)
• Add new Jupiter example explaining large scale recommendation with Buffalo

Update

• Add CUDA accelerated BPR-MF sample usage

  • When tested against ml-20m and Kakao Brunch Dataset, CUDA version of BPR-MF was about 5x times faster than CPU implementation of BPR-MF in training time.

• Add metric truncated AUC

• Add notebook examples

• Update documents

Bug fixes

• Fix SGD update of BPR
• Fix to use bias when calculating top-k recommendation in BPR-MF

Misc

• Clean code(linting, refactoring)
• Add internal data type check and state check
• Change validation format to be CSR matrix
  • Validation took about 40% faster than the previous version.
  • but this might be harmful because it requires larger memory during validation.

Bug Fixes

• hotfix, backward compatibility for pbar, iter_pbar

Bug Fixes

• Fix numerical error when calculating training loss in conjugate gradient method(CPU implementation only)
• Fix broken print out for progress bar.

Misc

• No longer need root permission for installation.

Bug Fixes

• Correct illegal Python Index Package(PIP). (Added missing files to MANIFEST)

Misc

• Apply https://github.com/kakao/buffalo/issues/10
• Update unittest for travis-ci environments.

Misc

• Added travis-ci (WIP)
• Added examples/Dockerfile #7
• Correct stream default option(disable sppmi)

Misc

• Fix memory corruption error when using optimize feature of ALS
• Correct typos

Misc

• Now buffalo can be installed with Python Package Index. pip install buffalo
• Added ld.conf.d/buffalo.pc.