A unified bandit framework for general open multi-agent systems with global-UCB algorithms and regret bounds linear in entry uncertainty and dependent on system stability and agent patterns.
Llm-based multi-agent systems for software engineering: Literature review, vision, and the road ahead
3 Pith papers cite this work. Polarity classification is still indexing.
3
Pith papers citing it
representative citing papers
SWE-EVO shows GPT-5.4 with OpenHands reaching only 25% success on complex multi-file evolution tasks versus 72.8% on SWE-Bench Verified, and introduces Fix Rate as a partial-progress metric.
citing papers explorer
-
Bandit Learning in General Open Multi-agent Systems
A unified bandit framework for general open multi-agent systems with global-UCB algorithms and regret bounds linear in entry uncertainty and dependent on system stability and agent patterns.
-
SWE-EVO: Benchmarking Coding Agents in Long-Horizon Software Evolution Scenarios
SWE-EVO shows GPT-5.4 with OpenHands reaching only 25% success on complex multi-file evolution tasks versus 72.8% on SWE-Bench Verified, and introduces Fix Rate as a partial-progress metric.
- Learning to Interrupt in Language-based Multi-agent Communication