# Learning to Retrieve: Dual-Level Long-Term Memory for Text-to-SQL Agents

arXiv: https://arxiv.org/abs/2606.00547 · HTML: https://mlsanity.com/paper/2606.00547/

## Key facts

- Cited by 4 papers in the map (data through 14 Oct 2026).
- Builds on 22 of its 43 references that are in the map.
- First posted on arXiv 30 May 2026.

## Builds on (most cited first, 22 of 22)

- [Generative Agents: Interactive Simulacra of Human Behavior](https://mlsanity.com/paper/2304.03442/) (2023)
- [Evaluating Very Long-Term Conversational Memory of LLM Agents](https://mlsanity.com/paper/2402.17753/) (2024)
- [Seq2SQL: Generating Structured Queries from Natural Language using Reinforcement Learning](https://mlsanity.com/paper/1709.00103/) (2017)
- [Spider 2.0: Evaluating Language Models on Real-World Enterprise Text-to-SQL Workflows](https://mlsanity.com/paper/2411.07763/) (2024)
- [MAC-SQL: A Multi-Agent Collaborative Framework for Text-to-SQL](https://mlsanity.com/paper/2312.11242/) (2023)
- [EASYTOOL: Enhancing LLM-based Agents with Concise Tool Instruction](https://mlsanity.com/paper/2401.06201/) (2024)
- [Evaluation and Benchmarking of LLM Agents: A Survey](https://mlsanity.com/paper/2507.21504/) (2025)
- [SimpleMem: Efficient Lifelong Memory for LLM Agents](https://mlsanity.com/paper/2601.02553/) (2026)
- [SQLNet: Generating Structured Queries From Natural Language Without Reinforcement Learning](https://mlsanity.com/paper/1711.04436/) (2017)
- [Bridging Textual and Tabular Data for Cross-Domain Text-to-SQL Semantic Parsing](https://mlsanity.com/paper/2012.12627/) (2020)
- [MemoChat: Tuning LLMs to Use Memos for Consistent Long-Range Open-Domain Conversation](https://mlsanity.com/paper/2308.08239/) (2023)
- [Process Reward Models for LLM Agents: Practical Framework and Directions](https://mlsanity.com/paper/2502.10325/) (2025)
- [MemInsight: Autonomous Memory Augmentation for LLM Agents](https://mlsanity.com/paper/2503.21760/) (2025)
- [TypeSQL: Knowledge-based Type-Aware Neural Text-to-SQL Generation](https://mlsanity.com/paper/1804.09769/) (2018)
- [RAP: Retrieval-Augmented Planning with Contextual Memory for Multimodal LLM Agents](https://mlsanity.com/paper/2402.03610/) (2024)
- [A Review of Prominent Paradigms for LLM-Based Agents: Tool Use (Including RAG), Planning, and Feedback Learning](https://mlsanity.com/paper/2406.05804/) (2024)
- [Hierarchical Memory for High-Efficiency Long-Term Reasoning in LLM Agents](https://mlsanity.com/paper/2507.22925/) (2025)
- [On the Structural Memory of LLM Agents](https://mlsanity.com/paper/2412.15266/) (2024)
- [Interactive Text-to-SQL Generation via Editable Step-by-Step Explanations](https://mlsanity.com/paper/2305.07372/) (2023)
- [BIRD-INTERACT: Re-imagining Text-to-SQL Evaluation for Large Language Models via Lens of Dynamic Interactions](https://mlsanity.com/paper/2510.05318/) (2025)
- [CoE-SQL: In-Context Learning for Multi-Turn Text-to-SQL with Chain-of-Editions](https://mlsanity.com/paper/2405.02712/) (2024)
- [Topic Coverage-based Demonstration Retrieval for In-Context Learning](https://mlsanity.com/paper/2509.12451/) (2025)

## Built on by (most cited first, 4 of 4)

- [Data Intelligence Agents: Interpreting, Modeling, and Querying Enterprise Data via Autonomous Coding Agents](https://mlsanity.com/paper/2606.19319/) (2026)
- [From Test-Time Scaling to Reusable Memory: Measuring Crystallization in Text-to-SQL](https://mlsanity.com/paper/2608.07213/) (2026)
- [Beyond Context Windows: Persistent Discovery Context for Data-Centric Agents](https://mlsanity.com/paper/2609.02129/) (2026)
- [Towards Reliable AI Data Scientists: Data Agents with Workflow Harnesses](https://mlsanity.com/paper/2609.35255/) (2026)

## Abstract (arXiv)

Interactive text-to-SQL agents solve database tasks through multi-turn interactions involving schema exploration, query execution, feedback interpretation, and decision revision. Long-term memory helps agents reuse past experiences, but existing retrieval methods remain limited. Static methods use fixed similarity heuristics, whereas dynamic methods often learn from sparse terminal outcomes and retrieve at a single decision horizon. This is insufficient when memory usefulness changes across interaction stages, since memories useful for initial planning may differ from those needed for local, state-conditioned execution. We propose MERIT, a dynamic multi-horizon memory retrieval framework. MERIT maintains episode-level memory for global strategic guidance and turn-level memory for local decision support. Both levels use learned retrieval policies optimized with reinforcement learning. To train turn-level retrieval despite limited intermediate supervision, MERIT uses a lightweight Process Reward Model to provide dense proxy rewards for local memory selection. On BIRD-Interact, MERIT achieves higher success rates and fewer interaction turns than no-memory, static-retrieval, and dynamic-retrieval baselines. Transfer results on Spider2-Snow further show that the learned policies transfer without degradation, staying above all memory baselines without benchmark-specific tuning. These results suggest that multi-horizon retrieval improves experience reuse in interactive text-to-SQL agents. Our code is available at https://github.com/YiboWANG214/MERIT.