Antoine Bosselut
Tenure Track Assistant Professor
Leads the EPFL NLP group, building language systems that represent and reason about knowledge, including commonsense reasoning and knowledge-grounded language models.
Antoine Bosselut's lab at EPFL works on knowledge, QA & retrieval, large language models and signal processing & information theory. 61 papers since 2023, 41 at top venues, mostly EMNLP and ACL. 10 PhD students. 19 alumni, most often to Alibaba and Google. Applications close .
At a glance
- papers, 2023–2026
- 16/ year
- at T1 venues
- 67%
- PhD students
- 10
- MS students
- 0
- alumni listed
- 19
Papers by yearfrom the lab homepage, 2023–2026
- 15 papers2026 · T1 10 · collecting
- 18 papers2025 · T1 13
- 17 papers2024 · T1 10
- 11 papers2023 · T1 8
T1 venuesOtherVenues: EMNLP, ACL, EACL, ICLR, NAACL
Research focuspaper titles that use each topic's words
- Large language models26 papers
- Knowledge, QA & retrieval11 papers
- Signal processing & information theory0 papers
- represent0 papers
- reason0 papers
Paper list
From the lab homepage.
202615 papers
- ICLR (T1)AbstRaL: Augmenting LLMs' Reasoning by Reinforcing Abstract Thinking (opens in a new tab)
- CHI (T1)AI meets Mathematics Education: Supporting Instructors in Large Mathematics Classes with Context-Aware AI (opens in a new tab)
- ACL (T1)Apertus: Democratizing Open and Compliant LLMs for Global Language Environments (opens in a new tab)
- EACLConLID: Supervised Contrastive Learning for Low-Resource Language Identification (opens in a new tab)
- COLMCresOWLve: Benchmarking Creative Problem-Solving Over Real-World Knowledge (opens in a new tab)
- ACL (T1)Crosscoding Through Time: Tracking Emergence & Consolidation Of Linguistic Representations Throughout LLM Pretraining (opens in a new tab)
- EACLDRIVINGVQA: A Dataset for Interleaved Visual Chain-of-Thought in Real-World Driving Scenarios (opens in a new tab)
- ICML (T1)DynaVieW: Schema-Guided World Modeling for Understanding Hierarchical Visual Dynamics (opens in a new tab)
- ICML (T1)Helpful to a Fault: Measuring Illicit Assistance in Multi-Turn, Multilingual LLM Agents (opens in a new tab)
- COLMLarge Language Models Align with the Human Brain during Creative Thinking (opens in a new tab)
- EMNLP (T1)Lingua Franca or Probing Artifact? Rethinking Latent Language in Multilingual LLMs (opens in a new tab)
- ICLR (T1)Mixture of Cognitive Reasoners: Modular Reasoning with Brain-Like Specialization (opens in a new tab)
202518 papers
- NAACL (T1)A Logical Fallacy-Informed Framework for Argument Generation (opens in a new tab)
- COLMCan Performant LLMs Be Ethical? Quantifying the Impact of Web Crawling Opt-Outs (opens in a new tab)
- EMNLP (T1)Context is Gold to find the Gold Passage: Evaluating and Training Contextual Document Embeddings (opens in a new tab)
- Proceedings of the Conference on Empirical Methods in Natural Language Processing (EMNLP) - FindingsCreative Preference Optimization (opens in a new tab)
- COLING (T1)Efficient Tool Use with Chain-of-Abstraction Reasoning (opens in a new tab)
- ICCCEvaluating Creative Short Story Generation in Humans and Large Language Models (opens in a new tab)
- NAACL (T1)Evaluating Morphological Compositional Generalization in Large Language Models (opens in a new tab)
- NeurIPS (T1)For Better or for Worse, Transformers Seek Patterns for Memorization (opens in a new tab)
- EMNLP (T1)From Language to Cognition: How LLMs Outgrow the Human Language Network (opens in a new tab)
- ICCV (T1)GeoExplorer: Active Geo-localization with Curiosity-Driven Exploration (opens in a new tab)
- ICLR (T1)INCLUDE: Evaluating Multilingual Language Understanding with Regional Knowledge (opens in a new tab)
- COLMLLMs Are In-Context Bandit Reinforcement Learners (opens in a new tab)
202417 papers
- EMNLP (T1)"Flex Tape Can't Fix That": Bias and Misinformation in Edited Language Models (opens in a new tab)
- CHI (T1)A Design Space for Intelligent and Interactive Writing Assistants (opens in a new tab)
- ACL (T1)Complex Reasoning over Logical Queries on Commonsense Knowledge Graphs (opens in a new tab)
- ECCV (T1)ConGeo: Robust Cross-view Geo-localization across Ground View Variations (opens in a new tab)
- AAAI (T1)ConVQG: Contrastive Visual Question Generation with Multimodal Guidance (opens in a new tab)
- PNAS (T1)Could ChatGPT get an Engineering Degree? Evaluating Higher Education Vulnerability to AI Assistants (opens in a new tab)
- SIGIR (T1)Course Recommender Systems Need to Consider the Job Market (opens in a new tab)
- ACL (T1)DiffuCOMET: Contextual Commonsense Knowledge Diffusion (opens in a new tab)
- EMNLP (T1)Discovering Knowledge-Critical Subnetworks in Pretrained Language Models (opens in a new tab)
- International Journal of Artificial Intelligence in EducationEnhancing Procedural Writing Through Personalized Example Retrieval: A Case Study on Cooking Recipes (opens in a new tab)
- COLMInstruction-tuning Aligns LLMs to the Human Brain (opens in a new tab)
- NLP4HRJOBSKAPE: A Framework for Generating Synthetic Job Postings to Enhance Skill Matching (opens in a new tab)
202311 papers
- Proceedings of the Conference on Empirical Methods in Natural Language Processing (EMNLP) - FindingsBreaking the Language Barrier: Improving Cross-Lingual Reasoning with Structured Self-Attention (opens in a new tab)
- EMNLP (T1)Can language models learn analogical reasoning? Investigating training objectives and comparisons to human performance (opens in a new tab)
- Proceedings of the Conference on Empirical Methods in Natural Language Processing (EMNLP) - FindingsCAR: Conceptualization-Augmented Reasoner for Zero-Shot Commonsense Question Answering (opens in a new tab)
- EMNLP (T1)CRAB: Assessing the Strength of Causal Relationships Between Real-world Events (opens in a new tab)
- EMNLP (T1)CRoW: Benchmarking Commonsense Reasoning in Real-World Tasks (opens in a new tab)
- ACL (T1)DISCO: Distilling Counterfactuals with Large Language Models (opens in a new tab)
- Proceedings of the Conference of the European Chapter of the Association for Computational Linguistics System Demonstrationskogito: A Commonsense Knowledge Inference Toolkit (opens in a new tab)
- ACL (T1)Mitigating Label Biases for In-context Learning (opens in a new tab)
- ACL (T1)PeaCoK: Persona Commonsense Knowledge for Consistent and Engaging Narratives (opens in a new tab)
- NeurIPS (T1)RECKONING: Reasoning through Dynamic Knowledge Encoding (opens in a new tab)
- EMNLP (T1)Towards a Mechanistic Interpretation of Multi-Step Reasoning Capabilities of Language Models (opens in a new tab)
Where alumni went19 listed, 60% to industry
Industry 3Academia 2
Most often to
- Alibaba
- Meta
Also to
- ETH Zurich
Apply to EPFLFall 2027
EDIC Computer and Communication Sciences PhD
- Round
- Round 1
- Opens
- —
- GRE
- optional
- English
- TOEFL 100 / IELTS 7.0
- Fee
- none
Confirmed on the program page.
Similar labsnearest by research summary
- USCMU
Louis-Philippe Morency
- Signal processing & information theory
- GBImperial
Francesca Toni
- Trustworthy & safe ML
- GBEdinburgh
Pasquale Minervini
- Knowledge, QA & retrieval
- USCMU
Yiming Yang
- Accelerators, chips & architecture
- CNTsinghua
Jie Tang
- Large language models
- USNYU
Ernest Davis
- Reasoning & neuro-symbolic AI
Your notes
Save this lab to keep notes.
Data
Checked against the lab homepage.
- Collected
- Verified
- Papers
- The lab homepage
- Members
- Lab homepage
- Alumni
- Lab homepage
Tiers are venue classifications, not lab rankings.
EPFLcloses