Breakpoint: Scalable evaluation of system-level reasoning in LLM code agents

• URL: http://arxiv.org/abs/2506.00172v1
• Date: Fri, 30 May 2025 19:23:51 GMT
• Title: Breakpoint: Scalable evaluation of system-level reasoning in LLM code agents
• Authors: Kaivalya Hariharan, Uzay Girit, Atticus Wang, Jacob Andreas,
• Abstract summary: We introduce Breakpoint, a benchmarking methodology that automatically generates code-repair tasks by adversarially corrupting functions.<n>We demonstrate that our methodology can scale to arbitrary difficulty, with state-of-the-art models' success rates ranging from 55% on the easiest tasks down to 0% on the hardest.
• Score: 40.37993572657772
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Benchmarks for large language models (LLMs) have predominantly assessed short-horizon, localized reasoning. Existing long-horizon suites (e.g. SWE-bench) rely on manually curated issues, so expanding or tuning difficulty demands expensive human effort and evaluations quickly saturate. However, many real-world tasks, such as software engineering or scientific research, require agents to rapidly comprehend and manipulate novel, complex structures dynamically; evaluating these capabilities requires the ability to construct large and varied sets of problems for agents to solve. We introduce Breakpoint, a benchmarking methodology that automatically generates code-repair tasks by adversarially corrupting functions within real-world software repositories. Breakpoint systematically controls task difficulty along two clear dimensions: local reasoning (characterized by code complexity metrics such as cyclomatic complexity) and system-level reasoning (characterized by call-graph centrality and the number of simultaneously corrupted interdependent functions). In experiments across more than 900 generated tasks we demonstrate that our methodology can scale to arbitrary difficulty, with state-of-the-art models' success rates ranging from 55% on the easiest tasks down to 0% on the hardest.

Related papers

• Analyzing Prominent LLMs: An Empirical Study of Performance and Complexity in Solving LeetCode Problems [0.0]
  Large Language Models (LLMs) like ChatGPT, Copilot, Gemini, and DeepSeek are transforming software engineering by automating key tasks.<n>This study benchmarks these four prominent LLMs on one hundred and fifty LeetCode problems across easy, medium, and hard difficulties.<n>We evaluate each model based on execution time, memory usage, and algorithmic complexity, revealing significant performance differences.
  arXiv  Detail & Related papers  (2025-08-05T21:50:52Z)
• CORE: Benchmarking LLMs Code Reasoning Capabilities through Static Analysis Tasks [12.465309397733249]
  Large language models (LLMs) have been widely adopted across diverse software engineering domains.<n>These applications require understanding beyond surface-level code patterns.<n>Existing benchmarks primarily evaluate end-to-end outcomes, such as whether code is correctly repaired or generated.
  arXiv  Detail & Related papers  (2025-07-03T01:35:58Z)
• Hierarchical Reasoning Model [16.223136644998203]
  HRM executes sequential reasoning tasks in a single forward pass without explicit supervision of the intermediate process.<n>With only 27 million parameters, HRM achieves exceptional performance on complex reasoning tasks using only 1000 training samples.
  arXiv  Detail & Related papers  (2025-06-26T19:39:54Z)
• Thinking Longer, Not Larger: Enhancing Software Engineering Agents via Scaling Test-Time Compute [61.00662702026523]
  We propose a unified Test-Time Compute scaling framework that leverages increased inference-time instead of larger models.<n>Our framework incorporates two complementary strategies: internal TTC and external TTC.<n>We demonstrate our textbf32B model achieves a 46% issue resolution rate, surpassing significantly larger models such as DeepSeek R1 671B and OpenAI o1.
  arXiv  Detail & Related papers  (2025-03-31T07:31:32Z)
• How to Get Your LLM to Generate Challenging Problems for Evaluation [33.625052642068624]
  CHASE is a unified framework to synthetically generate challenging problems using Large Language Models.<n>We implement CHASE to create evaluation benchmarks across three diverse domains.<n>The performance of state-of-the-art LLMs on these synthetic benchmarks lies in the range of 40-60% accuracy.
  arXiv  Detail & Related papers  (2025-02-20T16:09:55Z)
• BigCodeBench: Benchmarking Code Generation with Diverse Function Calls and Complex Instructions [72.56339136017759]
  We introduce BigCodeBench, a benchmark that challenges Large Language Models (LLMs) to invoke multiple function calls as tools from 139 libraries and 7 domains for 1,140 fine-grained tasks.<n>Our evaluation shows that LLMs are not yet capable of following complex instructions to use function calls precisely, with scores up to 60%, significantly lower than the human performance of 97%.<n>We propose a natural-language-oriented variant of BigCodeBench, BigCodeBench-Instruct, that automatically transforms the original docstrings into short instructions only with essential information.
  arXiv  Detail & Related papers  (2024-06-22T15:52:04Z)
• Can Long-Context Language Models Subsume Retrieval, RAG, SQL, and More? [54.667202878390526]
  Long-context language models (LCLMs) have the potential to revolutionize our approach to tasks traditionally reliant on external tools like retrieval systems or databases. We introduce LOFT, a benchmark of real-world tasks requiring context up to millions of tokens designed to evaluate LCLMs' performance on in-context retrieval and reasoning. Our findings reveal LCLMs' surprising ability to rival state-of-the-art retrieval and RAG systems, despite never having been explicitly trained for these tasks.
  arXiv  Detail & Related papers  (2024-06-19T00:28:58Z)
• PECC: Problem Extraction and Coding Challenges [3.287942619833188]
  We introduce PECC, a novel benchmark derived from Advent Of Code (AoC) challenges and Project Euler. Unlike conventional benchmarks, PECC requires LLMs to interpret narrative-embedded problems, extract requirements, and generate code. Results show varying model performance between narrative and neutral problems, with specific challenges in the Euler math-based subset.
  arXiv  Detail & Related papers  (2024-04-29T15:02:14Z)
• When Do Program-of-Thoughts Work for Reasoning? [51.2699797837818]
  We propose complexity-impacted reasoning score (CIRS) to measure correlation between code and reasoning abilities. Specifically, we use the abstract syntax tree to encode the structural information and calculate logical complexity. Code will be integrated into the EasyInstruct framework at https://github.com/zjunlp/EasyInstruct.
  arXiv  Detail & Related papers  (2023-08-29T17:22:39Z)
• TaskLAMA: Probing the Complex Task Understanding of Language Models [13.336015994186955]
  Structured Complex Task Decomposition (SCTD) is a problem of breaking down a complex real-world task into a directed acyclic graph over individual steps that contribute to achieving the task. We probe how accurately SCTD can be done with the knowledge extracted from Large Language Models (LLMs) Our experiments reveal that LLMs are able to decompose complex tasks into individual steps effectively, with a relative improvement of 15% to 280% over the best baseline.
  arXiv  Detail & Related papers  (2023-08-29T13:36:45Z)
• Tree-of-Mixed-Thought: Combining Fast and Slow Thinking for Multi-hop Visual Reasoning [16.495754104540605]
  Large language models (LLMs) can generate code-like plans for complex inference tasks such as visual reasoning. We propose a hierarchical plan-searching algorithm that integrates the one-stop reasoning (fast) and the Tree-of-thought (slow)
  arXiv  Detail & Related papers  (2023-08-18T16:21:40Z)
• Faith and Fate: Limits of Transformers on Compositionality [109.79516190693415]
  We investigate the limits of transformer large language models across three representative compositional tasks. These tasks require breaking problems down into sub-steps and synthesizing these steps into a precise answer. Our empirical findings suggest that transformer LLMs solve compositional tasks by reducing multi-step compositional reasoning into linearized subgraph matching.
  arXiv  Detail & Related papers  (2023-05-29T23:24:14Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.