Hironsan/awesome-embedding-models

Organizes foundational embedding research across word-level (Word2Vec, GloVe, FastText), contextual (ELMo, BERT), and sentence/document models, with curated links to landmark papers, evaluation benchmarks, and pre-trained model repositories. Covers the evolution from count-based to neural prediction methods, including critical evaluation frameworks that distinguish intrinsic versus extrinsic performance metrics. Aggregates training datasets (Wikipedia), benchmark suites (SemEval, WordSimilarity-353), and downloadable model implementations across TensorFlow Hub and AllenNLP.

1,827 stars. No commits in the last 6 months.