Training LLMs to Better Self-Debug and Explain Code

• URL: http://arxiv.org/abs/2405.18649v1
• Date: Tue, 28 May 2024 23:20:24 GMT
• Title: Training LLMs to Better Self-Debug and Explain Code
• Authors: Nan Jiang, Xiaopeng Li, Shiqi Wang, Qiang Zhou, Soneya Binta Hossain, Baishakhi Ray, Varun Kumar, Xiaofei Ma, Anoop Deoras,
• Abstract summary: We propose a training framework that significantly improves self-sourcedging capability of LLMs. We propose an automated pipeline to collect a high-quality dataset for code explanation and refinement. We perform supervised fine-tuning (SFT) and further reinforcement learning (RL) on both success and failure trajectories with a novel reward design.
• Score: 36.604898865514365
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: In the domain of code generation, self-debugging is crucial. It allows LLMs to refine their generated code based on execution feedback. This is particularly important because generating correct solutions in one attempt proves challenging for complex tasks. Prior works on self-debugging mostly focus on prompting methods by providing LLMs with few-shot examples, which work poorly on small open-sourced LLMs. In this work, we propose a training framework that significantly improves self-debugging capability of LLMs. Intuitively, we observe that a chain of explanations on the wrong code followed by code refinement helps LLMs better analyze the wrong code and do refinement. We thus propose an automated pipeline to collect a high-quality dataset for code explanation and refinement by generating a number of explanations and refinement trajectories and filtering via execution verification. We perform supervised fine-tuning (SFT) and further reinforcement learning (RL) on both success and failure trajectories with a novel reward design considering code explanation and refinement quality. SFT improves the pass@1 by up to 15.92% and pass@10 by 9.30% over four benchmarks. RL training brings additional up to 3.54% improvement on pass@1 and 2.55% improvement on pass@10. The trained LLMs show iterative refinement ability, and can keep refining code continuously. Lastly, our human evaluation shows that the LLMs trained with our framework generate more useful code explanations and help developers better understand bugs in source code.

Related papers

• Improving LLM Reasoning through Scaling Inference Computation with Collaborative Verification [52.095460362197336]
  Large language models (LLMs) struggle with consistent and accurate reasoning. LLMs are trained primarily on correct solutions, reducing their ability to detect and learn from errors. We propose a novel collaborative method integrating Chain-of-Thought (CoT) and Program-of-Thought (PoT) solutions for verification.
  arXiv  Detail & Related papers  (2024-10-05T05:21:48Z)
• Applying RLAIF for Code Generation with API-usage in Lightweight LLMs [15.366324461797582]
  Reinforcement Learning from AI Feedback (RLAIF) has demonstrated significant potential across various domains. This paper introduces an RLAIF framework for improving the code generation abilities of lightweight (1B parameters) LLMs.
  arXiv  Detail & Related papers  (2024-06-28T17:16:03Z)
• Self-play with Execution Feedback: Improving Instruction-following Capabilities of Large Language Models [54.14602121129874]
  We introduce AutoIF, the first scalable and reliable method for automatically generating instruction-following training data. AutoIF transforms the validation of instruction-following data quality into code verification.
  arXiv  Detail & Related papers  (2024-06-19T13:29:53Z)
• Aligning LLMs for FL-free Program Repair [14.935596175148586]
  This paper investigates a new approach to adapt large language models (LLMs) to program repair. Our core insight is that LLM's APR capability can be greatly improved by simply aligning the output to their training objective. Based on this insight, we designed D4C, a straightforward prompting framework for APR.
  arXiv  Detail & Related papers  (2024-04-13T02:36:40Z)
• Unsupervised Information Refinement Training of Large Language Models for Retrieval-Augmented Generation [128.01050030936028]
  We propose an information refinement training method named InFO-RAG. InFO-RAG is low-cost and general across various tasks. It improves the performance of LLaMA2 by an average of 9.39% relative points.
  arXiv  Detail & Related papers  (2024-02-28T08:24:38Z)
• StepCoder: Improve Code Generation with Reinforcement Learning from Compiler Feedback [58.20547418182074]
  We introduce StepCoder, a novel framework for code generation, consisting of two main components. CCCS addresses the exploration challenge by breaking the long sequences code generation task into a Curriculum of Code Completion Subtasks. FGO only optimize the model by masking the unexecuted code segments to provide Fine-Grained Optimization. Our method improves the ability to explore the output space and outperforms state-of-the-art approaches in corresponding benchmarks.
  arXiv  Detail & Related papers  (2024-02-02T13:14:31Z)
• Code Prompting Elicits Conditional Reasoning Abilities in Text+Code LLMs [65.2379940117181]
  We introduce code prompting, a chain of prompts that transforms a natural language problem into code. We find that code prompting exhibits a high-performance boost for multiple LLMs. Our analysis of GPT 3.5 reveals that the code formatting of the input problem is essential for performance improvement.
  arXiv  Detail & Related papers  (2024-01-18T15:32:24Z)
• At Which Training Stage Does Code Data Help LLMs Reasoning? [21.74241875923737]
  This paper explores the impact of code data on Large Language Models (LLMs) at different stages. Pre-training LLMs with the mixture of code and text can significantly enhance LLMs' general reasoning capability. At the instruction-tuning stage, code data endows LLMs the task-specific reasoning capability.
  arXiv  Detail & Related papers  (2023-09-28T09:50:27Z)
• CodeApex: A Bilingual Programming Evaluation Benchmark for Large Language Models [43.655927559990616]
  We propose CodeApex, a benchmark dataset focusing on the programming comprehension, code generation, and code correction abilities of LLMs. We evaluate 12 widely used LLMs, including both general-purpose and specialized models. GPT-4 exhibits the best programming capabilities, achieving approximate accuracy of 69%, 54%, and 66% on the three tasks, respectively.
  arXiv  Detail & Related papers  (2023-09-05T04:12:01Z)
• Coarse-Tuning Models of Code with Reinforcement Learning Feedback [0.0]
  Large Language Models (LLMs) pre-trained on code have emerged as the dominant approach to program synthesis. We propose RLCF, that further trains a pre-trained LLM via reinforcement learning, using feedback from a grounding function that scores the quality of the code.
  arXiv  Detail & Related papers  (2023-05-25T22:09:08Z)

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.