Mastodawn

🚨 Introducing 𝗚𝗿𝗶𝘁𝗛𝗼𝗽𝗽𝗲𝗿, the new State-of-the-Art Multi-Hop Dense Retriever 🦗

Current approaches to multi-hop retrieval face critical trade-offs.

🔎 Decomposition-based methods break complex queries into simpler steps, but they are computationally expensive and difficult to train end-to-end.

⚡ Decomposition-free methods improve efficiency, but often struggle with long reasoning chains and generalization beyond their training data.

So how can we get the best of both worlds?

Show thread

UKP Lab Mar 10

💡 𝗚𝗿𝗶𝘁𝗛𝗼𝗽𝗽𝗲𝗿 is designed to deliver both efficiency and robustness, while enabling strong multi-hop reasoning without explicit decomposition.
🧩 𝗖𝗼𝗿𝗲 𝗶𝗱𝗲𝗮: unified learning for retrieval and reasoning

GritHopper’s joint training objective combines contrastive learning for embedding similarity with causal language modeling for next-token prediction.

Show thread

UKP Lab Mar 10

📌 𝗪𝗵𝗮𝘁 𝗺𝗮𝗸𝗲𝘀 𝗚𝗿𝗶𝘁𝗛𝗼𝗽𝗽𝗲𝗿 𝘀𝘁𝗮𝗻𝗱 𝗼𝘂𝘁?
✅ 𝗘𝗻𝗰𝗼𝗱𝗲𝗿-𝗼𝗻𝗹𝘆 𝗲𝗳𝗳𝗶𝗰𝗶𝗲𝗻𝗰𝘆: each hop requires just a single forward pass
🌍 𝗢𝘂𝘁-𝗼𝗳-𝗱𝗶𝘀𝘁𝗿𝗶𝗯𝘂𝘁𝗶𝗼𝗻 𝗿𝗼𝗯𝘂𝘀𝘁𝗻𝗲𝘀𝘀: stronger generalization than existing decomposition-free baselines
🔗 𝗨𝗻𝗶𝗳𝗶𝗲𝗱 𝘁𝗿𝗮𝗶𝗻𝗶𝗻𝗴: bridges dense retrieval and generative objectives, showing how post-retrieval generation loss improves retrieval
🧠 𝗦𝗲𝗹𝗳-𝘀𝘁𝗼𝗽𝗽𝗶𝗻𝗴 𝗿𝗲𝗮𝘀𝗼𝗻𝗶𝗻𝗴: uses generative capabilities inspired by ReAct to control its own state and determine when to stop

Show thread

UKP Lab

📄 𝗚𝗥𝗜𝗧𝗛𝗼𝗽𝗽𝗲𝗿: 𝗗𝗲𝗰𝗼𝗺𝗽𝗼𝘀𝗶𝘁𝗶𝗼𝗻-𝗙𝗿𝗲𝗲 𝗠𝘂𝗹𝘁𝗶-𝗛𝗼𝗽 𝗗𝗲𝗻𝘀𝗲 𝗥𝗲𝘁𝗿𝗶𝗲𝘃𝗮𝗹
🌐 Project Website: https://ukplab.github.io/eacl2026-GritHopper/
🔗 Paper (arXiv): https://arxiv.org/pdf/2503.07519
🔗 Code: https://github.com/UKPLab/GritHopper
🔗 Model: https://huggingface.co/UKPLab/GritHopper-7B

GRITHopper: Decomposition-Free Multi-Hop Dense Retrieval

State-of-the-art multi-hop dense retrieval without query decomposition. Accepted at EACL 2026. Code, model & paper available.

GRITHopper

Show thread

UKP Lab Mar 10

👥 Justus-Jonas Erker (UKP Lab/Technische Universität Darmstadt), Nils Reimers (Cohere), Iryna Gurevych (UKP Lab/Technische Universität Darmstadt)

See you at Hashtag#EACL2026 in Rabat 🕌!

#UKPLab #NLP #NLProc #InformationRetrieval #DenseRetrieval #MultiHop #FactChecking #QuestionAnswering #RAG