Harry Scells
@hscells.bsky.social
Assistant Professor @unituebingen.bsky.social @health-nlp.com, information retrieval researcher
https://scells.me #IR #CS
https://scells.me #IR #CS
Reposted by Harry Scells
Honored to win the ICTIR Best Paper Honorable Mention Award for "Axioms for Retrieval-Augmented Generation"!
Our new axioms are integrated with ir_axioms: github.com/webis-de/ir_...
Nice to see axiomatic IR gaining momentum.
Our new axioms are integrated with ir_axioms: github.com/webis-de/ir_...
Nice to see axiomatic IR gaining momentum.
July 18, 2025 at 2:18 PM
Honored to win the ICTIR Best Paper Honorable Mention Award for "Axioms for Retrieval-Augmented Generation"!
Our new axioms are integrated with ir_axioms: github.com/webis-de/ir_...
Nice to see axiomatic IR gaining momentum.
Our new axioms are integrated with ir_axioms: github.com/webis-de/ir_...
Nice to see axiomatic IR gaining momentum.
Reposted by Harry Scells
Congrats to the authors @lgnp.bsky.social @timhagen.bsky.social @maik-froebe.bsky.social @matthias-hagen.bsky.social @benno-stein.de @martin-potthast.com @hscells.bsky.social from @unikassel.bsky.social @hessianai.bsky.social @scadsai.bsky.social @unituebingen.bsky.social @uni-jena.de & Uni Weimar
July 16, 2025 at 9:04 PM
Reposted by Harry Scells
Reposted by Harry Scells
Congratulations to the authors @lgnp.bsky.social @deckersniklas.bsky.social @martin-potthast.com @hscells.bsky.social !
📄 Preprint: arxiv.org/abs/2407.21515
💻 Code: github.com/webis-de/ada...
📄 Preprint: arxiv.org/abs/2407.21515
💻 Code: github.com/webis-de/ada...
Learning Effective Representations for Retrieval Using Self-Distillation with Adaptive Relevance Margins
Representation-based retrieval models, so-called biencoders, estimate the relevance of a document to a query by calculating the similarity of their respective embeddings. Current state-of-the-art bien...
arxiv.org
June 22, 2025 at 12:33 PM
Congratulations to the authors @lgnp.bsky.social @deckersniklas.bsky.social @martin-potthast.com @hscells.bsky.social !
📄 Preprint: arxiv.org/abs/2407.21515
💻 Code: github.com/webis-de/ada...
📄 Preprint: arxiv.org/abs/2407.21515
💻 Code: github.com/webis-de/ada...
Reposted by Harry Scells
Short Paper: Rank-DistiLLM: Closing the Effectiveness Gap Between Cross-Encoders and LLMs for Passage Re-ranking webis.de/publications...
Full Paper: Set-Encoder: Permutation-Invariant Inter-Passage Attention for Listwise Passage Re-Ranking with Cross-Encoders webis.de/publications...
Full Paper: Set-Encoder: Permutation-Invariant Inter-Passage Attention for Listwise Passage Re-Ranking with Cross-Encoders webis.de/publications...
Webis Publications
Publications by the Webis group
webis.de
April 9, 2025 at 12:37 PM
Short Paper: Rank-DistiLLM: Closing the Effectiveness Gap Between Cross-Encoders and LLMs for Passage Re-ranking webis.de/publications...
Full Paper: Set-Encoder: Permutation-Invariant Inter-Passage Attention for Listwise Passage Re-Ranking with Cross-Encoders webis.de/publications...
Full Paper: Set-Encoder: Permutation-Invariant Inter-Passage Attention for Listwise Passage Re-Ranking with Cross-Encoders webis.de/publications...
Reposted by Harry Scells
🧵 2/4 Key findings:
1️⃣ Humans write best? No! LLM responses are rated better than human.
2️⃣ Essay answers? No! Bullet lists are often preferred.
3️⃣ Evaluate with BLEU? No! Reference-based metrics don't align with human preferences.
4️⃣ LLMs as judges? No! Prompted models produce inconsistent labels.
1️⃣ Humans write best? No! LLM responses are rated better than human.
2️⃣ Essay answers? No! Bullet lists are often preferred.
3️⃣ Evaluate with BLEU? No! Reference-based metrics don't align with human preferences.
4️⃣ LLMs as judges? No! Prompted models produce inconsistent labels.
April 7, 2025 at 3:34 PM
🧵 2/4 Key findings:
1️⃣ Humans write best? No! LLM responses are rated better than human.
2️⃣ Essay answers? No! Bullet lists are often preferred.
3️⃣ Evaluate with BLEU? No! Reference-based metrics don't align with human preferences.
4️⃣ LLMs as judges? No! Prompted models produce inconsistent labels.
1️⃣ Humans write best? No! LLM responses are rated better than human.
2️⃣ Essay answers? No! Bullet lists are often preferred.
3️⃣ Evaluate with BLEU? No! Reference-based metrics don't align with human preferences.
4️⃣ LLMs as judges? No! Prompted models produce inconsistent labels.