From the 1 of 11 linked papers with an AI index.
11 papers
With Argus Eyes: Assessing Retrieval Gaps via Uncertainty Scoring to Detect and Remedy Retrieval Blind Spots
Zeinab Sadat Taghavi, Ali Modarressi, Hinrich Schutze +1
The paper identifies blind spots in neural retrievers used for retrieval‑augmented generation, where relevant entities are missed due to low embedding similarity, and proposes an u…
Expert-Aware Causal Tracing of Factual Recall in Sparse MoE Language Models
Yuetian Lu, Ali Modarressi, Yihong Liu +1
Causal tracing of factual recall has been studied predominantly in dense transformer language models, where interventions localize information flow to layers or feed-forward module…
Steering MoE LLMs via Expert (De)Activation
Mohsen Fayyaz, Ali Modarressi, Hanieh Deilamsalehy +5
Mixture-of-Experts (MoE) in Large Language Models (LLMs) routes each token through a subset of specialized Feed-Forward Networks (FFN), known as experts. We present SteerMoE, a fra…
Persistent Personas? Role-Playing, Instruction Following, and Safety in Extended Interactions
Pedro Henrique Luz de Araujo, Michael A. Hedderich, Ali Modarressi +2
Persona-assigned large language models (LLMs) are used in domains such as education, healthcare, and sociodemographic simulation. Yet, they are typically evaluated only in short, s…
ImpliRet: Benchmarking the Implicit Fact Retrieval Challenge
Zeinab Sadat Taghavi, Ali Modarressi, Yunpu Ma +1
Retrieval systems are central to many NLP pipelines, but often rely on surface-level cues such as keyword overlap and lexical semantic similarity. To evaluate retrieval beyond thes…
NoLiMa: Long-Context Evaluation Beyond Literal Matching
Ali Modarressi, Hanieh Deilamsalehy, Franck Dernoncourt +4
Recent large language models (LLMs) support long contexts ranging from 128K to 1M tokens. A popular method for evaluating these capabilities is the needle-in-a-haystack (NIAH) test…