# Beyond "What to Retrieve": Uncertainty in Retrieval-Augmented Code Generation

arXiv: https://arxiv.org/abs/2607.24884 · HTML: https://mlsanity.com/paper/2607.24884/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Placed in RAG for code, one of 15 directions in Retrieval-augmented generation.
- Builds on 20 of its 30 references that are in the map.
- First posted on arXiv 27 Jul 2026.

## Placed in

- Retrieval-augmented generation / [RAG for code](https://mlsanity.com/rag/rag-for-code/)

## Builds on (most cited first, 20 of 20)

- [Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks](https://mlsanity.com/paper/2005.11401/) (2020)
- [Evaluating Large Language Models Trained on Code](https://mlsanity.com/paper/2107.03374/) (2021)
- [On Calibration of Modern Neural Networks](https://mlsanity.com/paper/1706.04599/) (2017)
- [Code Llama: Open Foundation Models for Code](https://mlsanity.com/paper/2308.12950/) (2023)
- [Self-RAG: Learning to Retrieve, Generate, and Critique through Self-Reflection](https://mlsanity.com/paper/2310.11511/) (2023)
- [Teaching Large Language Models to Self-Debug](https://mlsanity.com/paper/2304.05128/) (2023)
- [CodeT: Code Generation with Generated Tests](https://mlsanity.com/paper/2207.10397/) (2022)
- [RepoCoder: Repository-Level Code Completion Through Iterative Retrieval and Generation](https://mlsanity.com/paper/2303.12570/) (2023)
- [UniXcoder: Unified Cross-Modal Pre-training for Code Representation](https://mlsanity.com/paper/2203.03850/) (2022)
- [LEVER: Learning to Verify Language-to-Code Generation with Execution](https://mlsanity.com/paper/2302.08468/) (2023)
- [DocPrompting: Generating Code by Retrieving the Docs](https://mlsanity.com/paper/2207.05987/) (2022)
- [Retrieval Augmented Code Generation and Summarization](https://mlsanity.com/paper/2108.11601/) (2021)
- [ReACC: A Retrieval-Augmented Code Completion Framework](https://mlsanity.com/paper/2203.07722/) (2022)
- [Can ChatGPT replace StackOverflow? A Study on Robustness and Reliability of Large Language Model Code Generation](https://mlsanity.com/paper/2308.10335/) (2023)
- [Repoformer: Selective Retrieval for Repository-Level Code Completion](https://mlsanity.com/paper/2403.10059/) (2024)
- [On the Impacts of Contexts on Repository-Level Code Generation](https://mlsanity.com/paper/2406.11927/) (2024)
- [What to Retrieve for Effective Retrieval-Augmented Code Generation? An Empirical Study and Beyond](https://mlsanity.com/paper/2503.20589/) (2025)
- [ExecRepoBench: Multi-level Executable Code Completion Evaluation](https://mlsanity.com/paper/2412.11990/) (2024)
- [Retrieval-Augmented Generation with Estimation of Source Reliability](https://mlsanity.com/paper/2410.22954/) (2024)
- [Incoherence as Oracle-less Measure of Error in LLM-Based Code Generation](https://mlsanity.com/paper/2507.00057/) (2025)

## Built on by (most cited first, 1 of 1)

- [RCL: A Retrieval-Confidence Layer for Detecting Insufficient Context in Enterprise Retrieval-Augmented Code Generation](https://mlsanity.com/paper/2609.11023/) (2026)

## Abstract (arXiv)

Repository-level code generation relies on heterogeneous evidence whose relevance, compatibility, and completeness are inherently uncertain. Similar-code examples, repository context, and project-specific APIs may provide complementary information, but can also introduce noisy, redundant, or conflicting signals. Existing retrieval-augmented approaches primarily optimize retrieval relevance without explicitly modeling how uncertainty in retrieved evidence affects downstream generation. We introduce OpenCoder, an uncertainty-aware framework that estimates source-specific uncertainty, uses it to filter and rank heterogeneous evidence, and guides generation, verification, and repair. A factorial analysis over API knowledge, repository context, and similar-code evidence reveals no universal additive source ranking; instead, significant cross-source interactions depend on the accompanying evidence and LLM backend. On an expanded 32-task RepoExec-inline evaluation, OpenCoder improves GPT selected-output correctness over Baseline RAG from 56.25% to 78.13%. However, it matches a verification-and-repair control, and the corresponding Gemini improvement is not statistically supported, indicating backend-dependent benefits. Target-aware API refinement also substantially improves API-set retrieval. These findings support treating uncertainty as an actionable control signal for repository-level retrieval, verification, and repair.