Word vectors and the sentences they build

What it does This repository is a curated reference list that tracks research papers offering pretrained word and sentence embedding models. It organizes them into tables showing publication dates, citation counts, training code repositories, and direct links to pretrained weights. The list also breaks down the standard pipeline—word embeddings, optional encoders, pooling methods, and evaluation benchmarks—to show how different papers fit together.

The interesting bit Rather than dumping links, the author structures the field like an anatomy lesson: almost every sentence embedding is just word embeddings plus pooling, so the tables map each paper to its components and whether you can actually download the weights today. It is an explicit admission that the list itself will always be incomplete, which is refreshingly honest for an awesome-list.

Key highlights

• Covers word embeddings, contextualized embeddings, encoders, pooling strategies, OOV handling, and evaluation tasks in one place
• Tables include citation counts and direct links to both training code and pretrained model downloads
• Explicitly flags broken or missing download links where known
• Organizes papers by architectural component rather than chronology alone
• Acknowledges its own incompleteness while inviting pull requests to fill gaps

Caveats

• The provided source is truncated, so the full scope of recent models is not visible in the excerpt
• Some listed pretrained model links are explicitly marked as broken
• The author concedes the list will always be incomplete, so coverage gaps are expected

Verdict Worth bookmarking if you are picking a sentence embedding model and want to compare options by citation count and weight availability. Skip it if you need a hands-on tutorial or a framework rather than a literature index.