# sparse-embeddings

> Browse 3 items tagged with sparse-embeddings.

## Items

- [CSRv2](https://vectordb.ever.works/items/csrv2) — Contrastive Sparse Representation learning approach for ultra-sparse embeddings that achieves 7x speedup over Matryoshka Representation Learning with 300x improvements in compute and memory efficiency.
- [Progressive K-Annealing](https://vectordb.ever.works/items/progressive-k-annealing) — Training technique in CSRv2 that stabilizes sparsity learning by gradually increasing sparsity constraints, reducing dead neurons from >80% to ~20%.
- [Term Expansion](https://vectordb.ever.works/items/term-expansion) — A retrieval technique that expands queries or documents with related but not literally present terms. Key feature of learned sparse models like SPLADE, enabling identification of relevant documents even when exact terms don't match.

---

_Canonical page: https://vectordb.ever.works/tags/sparse-embeddings_