Challenge: Existing benchmarks focus on the repair generation capability of LLMs, lacking fine-grained evaluation of reflection.
Approach: They propose a benchmark with oracle reflections and a dual-task protocol to decouple evaluation of reflection from repair.
Outcome: The proposed benchmarks show that underperforming reflection capabilities remain a bottleneck for code repair.

Similar Papers

Turning the Tide: Repository-based Code Reflection (2025.findings-emnlp)

Copied to clipboard

Challenge: Code large language models (LLMs) enhance programming by understanding and generating code across languages.
Approach: a new benchmark evaluates code understanding and generation in repositories using code large language models.
Outcome: The proposed model improves code understanding and generation in repositories by evaluating 1,888 test cases across 6 programming languages.
SynFix: Dependency-Aware Program Repair via RelationGraph Analysis (2025.findings-acl)

Copied to clipboard

Challenge: Existing methods for resolving repository-level debugging are limited by their interdependencies.
Approach: They propose a RelationGraph-based approach that integrates large language models with structural search and synchronization techniques for coordinated program repair across codebases.
Outcome: SynFix resolves 52.33% of issues in SWE-bench-lite, 55.8% in Swe-bech-verified and 29.86% in S WE-beach-full.
Do LLMs Catch Their Own Mistakes? A Comprehensive Benchmark for Reflective Tool Use LLMs (2026.findings-acl)

Copied to clipboard

Challenge: Existing benchmarks primarily evaluate planning and execution success, overlooking the self-reflective dimension of tool use.
Approach: They propose a benchmark to assess LLMs’ self-reflective reasoning in tool-augmented multi-turn dialogues.
Outcome: The proposed benchmark covers 10 domains with 88 distinct APIs and 968 annotated dialogues, systematically injecting diverse error types arising from both user and assistant behavior.
PythonSaga: Redefining the Benchmark to Evaluate Code Generating LLMs (2024.findings-emnlp)

Copied to clipboard

Challenge: *HumanEval* and *MBPP* are two popular benchmarks for Python code generation.
Approach: They propose a large-scale human evaluation of two popular Python benchmarks . they propose 185 hand-crafted prompts in a balanced representation of 38 programming concepts across diverse difficulty levels.
Outcome: The proposed benchmarks show a critical bias towards a limited set of programming concepts, neglecting most of the other concepts entirely.
Forget What You Know about LLMs Evaluations - LLMs are Like a Chameleon (2025.emnlp-main)

Copied to clipboard

Challenge: Large language models (LLMs) excel on public benchmarks, but high scores may mask overreliance on dataset-specific surface cues rather than true language understanding.
Approach: They propose a meta-evaluation framework that systematically rephrases benchmark inputs to detect overfitting.
Outcome: The proposed framework detects performance degradation indicative of superficial pattern reliance on dataset-specific cues and distortion levels.
CodeScope: An Execution-based Multilingual Multitask Multidimensional Benchmark for Evaluating LLMs on Code Understanding and Generation (2024.acl-long)

Copied to clipboard

Challenge: Existing benchmarks for evaluating the code understanding and generation capacities of Large Language Models are insufficient . existing benchmarks focus on a narrow range of popular programming languages and specific tasks .
Approach: They propose an execution-based, multilingual, multitask evaluation benchmark for LLMs . they evaluate coding performance from three dimensions: length, difficulty, efficiency .
Outcome: The proposed benchmark covers 43 programming languages and eight coding tasks.
CodeReviewQA: The Code Review Comprehension Assessment for Large Language Models (2025.findings-acl)

Copied to clipboard

Challenge: State-of-the-art large language models (LLMs) have demonstrated impressive code generation capabilities but struggle with real-world software engineering tasks such as revising source code to address code reviews.
Approach: They propose a benchmark to evaluate large language models' ability to bridge both technical and conversational contexts by decomposing the generation task of code refinement into three essential reasoning steps.
Outcome: The proposed benchmark exposes specific model weaknesses in code review comprehension disentangled from their generative automated code refinement results.
Investigating the Transferability of Code Repair for Low-Resource Programming Languages (2025.findings-naacl)

Copied to clipboard

Challenge: Large language models (LLMs) have shown remarkable performance on code generation tasks.
Approach: They investigate the benefits of distilling the ability to repair code for both high and low resource languages to determine if the techniques are also applicable in low resource settings.
Outcome: The proposed techniques are effective in high- and low-resource languages, but weak in low-level languages.
CodeIF: Benchmarking the Instruction-Following Capabilities of Large Language Models for Code Generation (2025.acl-industry)

Copied to clipboard

Challenge: CodeIF assesses the ability of large language models to adhere to task-oriented instructions in code generation tasks.
Approach: They introduce a benchmark designed to assess LLMs' ability to adhere to task-oriented instructions within diverse code generation scenarios.
Outcome: The proposed benchmark assesses LLMs' ability to adhere to task-oriented instructions in code generation tasks across a wide range of complexity levels and programming domains.
SynthFix: Adaptive Neuro-Symbolic Code Vulnerability Repair (2026.findings-acl)

Copied to clipboard

Challenge: Large Language Models (LLMs) struggle with complex semantic and structural correctness required for automated code repair.
Approach: They propose a hybrid neural-symbolic framework that unifies code synthesis with compiler-informed symbolic feedback to improve LLM-based vulnerability repair.
Outcome: The proposed framework improves code repair accuracy and efficiency over strong SFT and RFT training strategies on the FixJS and CodeFlaws benchmarks.

What is GenGO?

GenGO is an NLP powered publication search system. It currenctly indexes 30k+ papers from ACL Anthology, and implements multi-aspect summarization, semantic search, and more!

Information

About
Limitations