/awesome-dense-text-retrieval

📖 A curated list of Dense Text Retrieval resources from all around the web.

MIT LicenseMIT

Awesome License

Dense Text Retrieval

Dense Text Retrieval (DTR) aims to overcome the limitations of sparse retrieval by matching in a continuous representation space learned via neural networks. [Learn more]

🗂 Datasets

Dataset Domain Language #Queries #Documents
📄 MS-MARCO General 🇬🇧 1,010,900+ 8,841,800+
📄 Natural Questions General 🇬🇧 91,500+
📄 COLIEE-2020 Law 🇬🇧, 🇯🇵 808 768
📄 BSARD Law 🇫🇷 1,108 22,633

🔥 Models

Model Language Size Year
📄 DPR 🇬🇧 2020
📄 ORQA 🇬🇧 2019
📄 DrQA 🇬🇧 2017

📚 Courses

🎙 Talks

  • 2021-09 Recent Developments in Neural Search, N. Reimers. [slides][video]
  • 2021-09 Training State-of-the-Art Neural Search Models, N. Reimers. [slides][video]

🗓 Conferences & Workshops

  • The Text Retrieval Conference (TREC) Deep Learning Track. [link]
  • The Special Interest Group on Information Retrieval (SIGIR). [link]

☁️ Tools

  • SentenceTransformers: a Python framework for state-of-the-art sentence, text and image embeddings. [link]
  • Jina AI: a cloud-native neural search framework for any kind of data. [link]
  • Haystack: an end-to-end neural search framework for textual data. [link]