MASPrism attributes failures in multi-agent systems by ranking candidates from prefill-stage NLL and attention signals of a 0.6B SLM, beating baselines by up to 33.41% Top-1 accuracy and proprietary LLMs by up to 89.5% relative improvement while processing traces in 2.66 seconds.
Gomez, Lukasz Kaiser, and Illia Polosukhin
4 Pith papers cite this work. Polarity classification is still indexing.
4
Pith papers citing it
citation-role summary
background 1
method 1
citation-polarity summary
verdicts
UNVERDICTED 4representative citing papers
MoRER builds an ER model repository via feature distribution clustering of tasks, achieving competitive results with limited labels versus active learning, transfer learning, and self-supervised methods on three multi-source datasets.
K-CARE uses behavior-derived anchoring and expert prototype analogies to ground LLMs and improve relevance on knowledge-intensive e-commerce cases.
Data-influence-score filtering using validation-set loss on downstream coding tasks improves Code-LLM performance, with the most beneficial training data varying significantly across different programming tasks.
citing papers explorer
-
MASPrism: Lightweight Failure Attribution for Multi-Agent Systems Using Prefill-Stage Signals
MASPrism attributes failures in multi-agent systems by ranking candidates from prefill-stage NLL and attention signals of a 0.6B SLM, beating baselines by up to 33.41% Top-1 accuracy and proprietary LLMs by up to 89.5% relative improvement while processing traces in 2.66 seconds.
-
Efficient Model Repository for Entity Resolution: Construction, Search, and Integration
MoRER builds an ER model repository via feature distribution clustering of tasks, achieving competitive results with limited labels versus active learning, transfer learning, and self-supervised methods on three multi-source datasets.
-
K-CARE: Knowledge-driven Symmetrical Contextual Anchoring and Analogical Prototype Reasoning for E-commerce Relevance
K-CARE uses behavior-derived anchoring and expert prototype analogies to ground LLMs and improve relevance on knowledge-intensive e-commerce cases.
-
An Empirical Study on Influence-Based Pretraining Data Selection for Code Large Language Models
Data-influence-score filtering using validation-set loss on downstream coding tasks improves Code-LLM performance, with the most beneficial training data varying significantly across different programming tasks.