This release supports LLVM 17.x . Following is the summary of changes:
- Modification of "BSD-3 Clause" license to "Apache 2.0 with LLVM Exceptions" license. (#114)
- Eigen is now optional. (#100, Thanks @PrasannaLanka)
- Vocabulary support with different embedding dimensions. (#126, Thanks @nishant-sachdeva)
- Vocabulary is in-memory, saved as
std::mapduring build. (#93, #102 , Thanks @PrasannaLanka, @RajivChitale) - Hyperparameter tuning support with Ray for training the Vocabulary (#119, #120, #123, Thanks @Charanyash, @Harikaraja)
- Addition of SQLite benchmarks as part of tests. (#113, Thanks @m-atalla)
What's Changed
- Patch to v1.0.0 by @svkeerthy in #22
- Merging main to llvm10 for release by @svkeerthy in #39
- Merging changes from main by @svkeerthy in #42
- Llvm17ray by @nishant-sachdeva in #85
- Update to cmake in order to support both standalone and llvm submodule builds. by @pranav-159 in #94
- Faster generation of vocabulary header by @RajivChitale in #102
- With pre-commit-checks by @RajivChitale in #104
- created file to generated vocabulary as map by @PrasannaLanka in #98
- LLVM17LIT by @nishant-sachdeva in #95
- updated vector initialisation by @yash3886 in #106
- new PR by @PrasannaLanka in #107
- Eigen package is optional (support for removed eigen) by @yash3886 in #108
- Benchmarks time evaluation script by @m-atalla in #111
- SQLite benchmark by @m-atalla in #113
- Changing BSD-4 license to Apache 2.0 with LLVM Exceptions by @svkeerthy in #114
- sqlite3 test errors fixed by @nishant-sachdeva in #115
- OpenKE Pytorch compatibility with link prediction scores by @Charanyash in #119
- Added Optuna Search and Configuration changes by @Harikaraja in #123
- Update README.md to reflect mutliple dims support by @svkeerthy in #124
- Seedemb opt by @nishant-sachdeva in #125
- Extending support for vocabularies of different dimensions by @nishant-sachdeva in #127
- Exposing vocab of different dims by @svkeerthy in #126
- Pulling latest changes from main by @svkeerthy in #128
- Exposing BB level representations by @svkeerthy in #129
- Training IR2Vec with the compile dataset by @nishant-sachdeva in #131
New Contributors
- @pranav-159 made their first contribution in #94
- @RajivChitale made their first contribution in #102
- @yash3886 made their first contribution in #106
- @Charanyash made their first contribution in #119
- @Harikaraja made their first contribution in #123
Full Changelog: v2.1.1...v2.1.2
Contributors
svkeerthy, nishant-sachdeva, and 7 other contributors