Separius/awesome-sentence-embedding

Organizes embedding approaches around a modular pipeline—word embeddings → optional encoders (LSTMs) → pooling strategies → task-specific objectives—enabling direct comparison across 100+ models spanning static word vectors (Word2Vec, GloVe), contextualized representations (ELMo, BERT), and specialized architectures (InferSent, Skip-Thought). Includes annotated tables with publication dates, citation counts, training code repositories, and pretrained model links across multiple frameworks (TensorFlow, PyTorch, Theano), plus sections on OOV handling, vector mapping techniques, and evaluation benchmarks.

2,290 stars. No commits in the last 6 months.