Changelog
All notable changes to this project will be documented in this file.
The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.
[Unreleased]
Added
- Glue dataset.
- Support for embedding multiple columns in the storage.
- Split language models abstractions into generative and classifiers.
Changed
- Storage now supports retriving slices and list of indices.
v0.0.3 - 2024-01-24
Added
- Website added.
- LanguageModels support prediction via logits.
Changed
- Optimizers uses a generator like interface.
v0.0.2 - 2024-01-20
Changed
- Decoupling of state into different components.
- Evaluators renamed to Adaptors.
v0.0.1 - 2024-01-18
Added
- Auto release to PyPI.
- A few implementations for baselines.
Changed
- Index now support batch size != 1 and k != 1.
- Batch size mandatary for Index, Embedders, LanguageModel, and Optimizer.
- Visualization change.
v0.0.0 - 2024-01-10
Added
- Index - Faiss, Hnswlib, Polar, Whitening.
- Storage - Pandas, Datasets
- Embedders - SentenceTransformers
- Corpus - Composed
- LanguageModel - Huggingface
- Scores - BLEU (NLTK, SacreBLEU), Rouge (Rouge, RougeScore), Exact
- Adaptor - BigBench
- Optimizer - Ax / BoTorch, KMeans