VDebugger: Harnessing Execution Feedback for Debugging Visual Programs

• URL: http://arxiv.org/abs/2406.13444v3
• Date: Fri, 04 Oct 2024 04:56:35 GMT
• Title: VDebugger: Harnessing Execution Feedback for Debugging Visual Programs
• Authors: Xueqing Wu, Zongyu Lin, Songyan Zhao, Te-Lin Wu, Pan Lu, Nanyun Peng, Kai-Wei Chang,
• Abstract summary: We introduce V Debugger, a critic-refiner framework trained to localize and debug visual programs by tracking execution step by step. V Debugger identifies and corrects program errors leveraging detailed execution feedback, improving interpretability and accuracy. Evaluations on six datasets demonstrate V Debugger's effectiveness, showing performance improvements of up to 3.2% in downstream task accuracy.
• Score: 103.61860743476933
• License:
• Abstract: Visual programs are executable code generated by large language models to address visual reasoning problems. They decompose complex questions into multiple reasoning steps and invoke specialized models for each step to solve the problems. However, these programs are prone to logic errors, with our preliminary evaluation showing that 58% of the total errors are caused by program logic errors. Debugging complex visual programs remains a major bottleneck for visual reasoning. To address this, we introduce VDebugger, a novel critic-refiner framework trained to localize and debug visual programs by tracking execution step by step. VDebugger identifies and corrects program errors leveraging detailed execution feedback, improving interpretability and accuracy. The training data is generated through an automated pipeline that injects errors into correct visual programs using a novel mask-best decoding technique. Evaluations on six datasets demonstrate VDebugger's effectiveness, showing performance improvements of up to 3.2% in downstream task accuracy. Further studies show VDebugger's ability to generalize to unseen tasks, bringing a notable improvement of 2.3% on the unseen COVR task. Code, data and models are made publicly available at https://github.com/shirley-wu/vdebugger/

Related papers

• NExT: Teaching Large Language Models to Reason about Code Execution [50.93581376646064]
  Large language models (LLMs) of code are typically trained on the surface textual form of programs. We propose NExT, a method to teach LLMs to inspect the execution traces of programs and reason about their run-time behavior.
  arXiv  Detail & Related papers  (2024-04-23T01:46:32Z)
• The Visual Debugger Tool [1.0624606551524207]
  Our tool visualizes program execution information graphically as an object diagram. Our tool is fully integrated into the popular Java development environment IntelliJ IDEA.
  arXiv  Detail & Related papers  (2024-04-19T15:02:29Z)
• CogCoM: Train Large Vision-Language Models Diving into Details through Chain of Manipulations [61.21923643289266]
  Chain of Manipulations is a mechanism that enables Vision-Language Models to solve problems step-by-step with evidence. After training, models can solve various visual problems by eliciting intrinsic manipulations (e.g., grounding, zoom in) actively without involving external tools. Our trained model, textbfCogCoM, achieves state-of-the-art performance across 9 benchmarks from 4 categories.
  arXiv  Detail & Related papers  (2024-02-06T18:43:48Z)
• Leveraging Print Debugging to Improve Code Generation in Large Language Models [63.63160583432348]
  Large language models (LLMs) have made significant progress in code generation tasks. But their performance in tackling programming problems with complex data structures and algorithms remains suboptimal. We propose an in-context learning approach that guides LLMs to debug by using a "print debug" method.
  arXiv  Detail & Related papers  (2024-01-10T18:37:59Z)
• Visual Program Distillation: Distilling Tools and Programmatic Reasoning into Vision-Language Models [17.540937747712082]
  We propose Visual Program Distillation (VPD), an instruction tuning framework that produces a vision-language model (VLM) VPD distills the reasoning ability of large language models by using them to sample multiple candidate programs. It translates each correct program into a language description of the reasoning steps, which are then distilled into a VLM.
  arXiv  Detail & Related papers  (2023-12-05T18:58:37Z)
• Teaching Large Language Models to Self-Debug [62.424077000154945]
  Large language models (LLMs) have achieved impressive performance on code generation. We propose Self- Debugging, which teaches a large language model to debug its predicted program via few-shot demonstrations.
  arXiv  Detail & Related papers  (2023-04-11T10:43:43Z)
• Generating Bug-Fixes Using Pretrained Transformers [11.012132897417592]
  We introduce a data-driven program repair approach which learns to detect and fix bugs in Java methods mined from real-world GitHub. We show that pretraining on source code programs improves the number of patches found by 33% as compared to supervised training from scratch. We refine the standard accuracy evaluation metric into non-deletion and deletion-only fixes, and show that our best model generates 75% more non-deletion fixes than the previous state of the art.
  arXiv  Detail & Related papers  (2021-04-16T05:27:04Z)
• Eye: Program Visualizer for CS2 [1.319058156672392]
  Eye is an interactive tool that visualizes a program's execution as it runs. It demonstrates properties and usage of data structures in a general environment. Eye opens up a gateway for CS2 students to more easily understand myriads of programs that are available on online programming websites.
  arXiv  Detail & Related papers  (2021-01-28T16:16:59Z)
• Graph-based, Self-Supervised Program Repair from Diagnostic Feedback [108.48853808418725]
  We introduce a program-feedback graph, which connects symbols relevant to program repair in source code and diagnostic feedback. We then apply a graph neural network on top to model the reasoning process. We present a self-supervised learning paradigm for program repair that leverages unlabeled programs available online.
  arXiv  Detail & Related papers  (2020-05-20T07:24:28Z)

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.