DualSQL Trains Text-to-SQL Agents with Multi-Agent RL
DualSQL-8B hits 71.1% on BIRD dev with only 8B parameters, beating prior 32B single-model systems, so teams running SQL agents on modest hardware can get strong accuracy without large-model inference costs.
Google and collaborators propose DualSQL, which splits Text-to-SQL into two agents: one links questions to the right tables and columns, the other writes the SQL.
Both agents share the same model weights, so a single multi-agent RL run trains both roles, with three database-query tools available during reasoning.
To prevent multi-agent RL collapse, the authors add rollout guardrails and a reward called robust execution match.
Trained on only 3,755 examples, DualSQL-4B reaches 68.0% execution accuracy on BIRD dev, matching earlier 7B models, while DualSQL-8B reaches 71.
1%, ahead of previous single-model 32B systems.
Great paper from Google and colleagues. Trains Text-to-SQL agents using multi-agent RL. (bookmark it) This work proposes DualSQL, which splits Text-to-SQL into two agents, one that links the question to the right tables and columns and one that writes the SQL. Both agents run on the same model weights, so a single multi-agent RL run trains both roles together. The agents can query the database through three tools while they reason. Multi-agent RL tends to collapse during training, so the authors add guardrails on rollouts and a new reward, robust execution match, that judges SQL correctness more accurately. Trained on only 3,755 examples, DualSQL-4B reaches 68.0% execution accuracy on BIRD dev, matching earlier 7B models. DualSQL-8B reaches 71.1%, ahead of previous single-model systems with 32B parameters. Paper: arxiv.org/abs/2609.18135 Chat with Paper: academy.dair.ai/papers/dualsql…📊 1 ⚡
This report is categorized under AITOP's "Papers" desk, focusing on US frontier AI labs and Silicon Valley technology leaders. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows. Key entities and technical topics under continuous tracking include DualSQL, Google, Text-to-SQL, multi-agent RL.
Source attribution: curated from elvis (@omarsar0).