What Makes Large Language Models Reason in (Multi-Turn) Code Generation?

• URL: http://arxiv.org/abs/2410.08105v2
• Date: Sat, 12 Oct 2024 12:42:19 GMT
• Title: What Makes Large Language Models Reason in (Multi-Turn) Code Generation?
• Authors: Kunhao Zheng, Juliette Decugis, Jonas Gehring, Taco Cohen, Benjamin Negrevergne, Gabriel Synnaeve,
• Abstract summary: Chain-of-thought has established itself as a popular vehicle for improving the outputs of large language models (LLMs) We investigate the effects of a wide range of prompting strategies with a focus on automatic re-prompting over multiple turns and computational requirements. Our study reveals strategies that consistently improve performance across all models with small and large sampling budgets.
• Score: 28.614888506962988
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Prompting techniques such as chain-of-thought have established themselves as a popular vehicle for improving the outputs of large language models (LLMs). For code generation, however, their exact mechanics and efficacy are under-explored. We thus investigate the effects of a wide range of prompting strategies with a focus on automatic re-prompting over multiple turns and computational requirements. After systematically decomposing reasoning, instruction, and execution feedback prompts, we conduct an extensive grid search on the competitive programming benchmarks CodeContests and TACO for multiple LLM families and sizes (Llama 3.0 and 3.1, 8B, 70B, 405B, and GPT-4o). Our study reveals strategies that consistently improve performance across all models with small and large sampling budgets. We then show how finetuning with such an optimal configuration allows models to internalize the induced reasoning process and obtain improvements in performance and scalability for multi-turn code generation.

Related papers

• Reference Trustable Decoding: A Training-Free Augmentation Paradigm for Large Language Models [79.41139393080736]
  Large language models (LLMs) have rapidly advanced and demonstrated impressive capabilities. We propose Reference Trustable Decoding (RTD), a paradigm that allows models to quickly adapt to new tasks without fine-tuning.
  arXiv  Detail & Related papers  (2024-09-30T10:48:20Z)
• Inference Optimization of Foundation Models on AI Accelerators [68.24450520773688]
  Powerful foundation models, including large language models (LLMs), with Transformer architectures have ushered in a new era of Generative AI. As the number of model parameters reaches to hundreds of billions, their deployment incurs prohibitive inference costs and high latency in real-world scenarios. This tutorial offers a comprehensive discussion on complementary inference optimization techniques using AI accelerators.
  arXiv  Detail & Related papers  (2024-07-12T09:24:34Z)
• What's Wrong with Your Code Generated by Large Language Models? An Extensive Study [80.18342600996601]
  Large language models (LLMs) produce code that is shorter yet more complicated as compared to canonical solutions. We develop a taxonomy of bugs for incorrect codes that includes three categories and 12 sub-categories, and analyze the root cause for common bug types. We propose a novel training-free iterative method that introduces self-critique, enabling LLMs to critique and correct their generated code based on bug types and compiler feedback.
  arXiv  Detail & Related papers  (2024-07-08T17:27:17Z)
• Should AI Optimize Your Code? A Comparative Study of Current Large Language Models Versus Classical Optimizing Compilers [0.0]
  Large Language Models (LLMs) raise intriguing questions about the potential for AI-driven approaches to revolutionize code optimization methodologies. This paper presents a comparative analysis between two state-of-the-art Large Language Models, GPT-4.0 and CodeLlama-70B, and traditional optimizing compilers.
  arXiv  Detail & Related papers  (2024-06-17T23:26:41Z)
• Mixture-of-Instructions: Comprehensive Alignment of a Large Language Model through the Mixture of Diverse System Prompting Instructions [7.103987978402038]
  We introduce a novel technique termed Mixture-of-Instructions (MoI) MoI employs a strategy of instruction concatenation combined with diverse system prompts to boost the alignment efficiency of language models. Our methodology was applied to the open-source Qwen-7B-chat model, culminating in the development of Qwen-SFT-MoI.
  arXiv  Detail & Related papers  (2024-04-29T03:58:12Z)
• Enhancing Code Generation Performance of Smaller Models by Distilling the Reasoning Ability of LLMs [36.409470894115074]
  We propose the CodePLAN framework, which aims to transfer LLMs' code generation reasoning capabilities to smaller models. Our approach improves the smaller model's code generation performance by over 130% on the challenging APPS benchmark.
  arXiv  Detail & Related papers  (2024-03-20T03:09:54Z)
• Leveraging Reinforcement Learning and Large Language Models for Code Optimization [14.602997316032706]
  This paper introduces a new framework to decrease the complexity of code optimization. The proposed framework builds on large language models (LLMs) and reinforcement learning (RL) We run several experiments on the PIE dataset using a CodeT5 language model and RRHF, a new reinforcement learning algorithm.
  arXiv  Detail & Related papers  (2023-12-09T19:50:23Z)
• MAgIC: Investigation of Large Language Model Powered Multi-Agent in Cognition, Adaptability, Rationality and Collaboration [102.41118020705876]
  Large Language Models (LLMs) have marked a significant advancement in the field of natural language processing. As their applications extend into multi-agent environments, a need has arisen for a comprehensive evaluation framework. This work introduces a novel benchmarking framework specifically tailored to assess LLMs within multi-agent settings.
  arXiv  Detail & Related papers  (2023-11-14T21:46:27Z)
• CodeRL: Mastering Code Generation through Pretrained Models and Deep Reinforcement Learning [92.36705236706678]
  "CodeRL" is a new framework for program synthesis tasks through pretrained LMs and deep reinforcement learning. During inference, we introduce a new generation procedure with a critical sampling strategy. For the model backbones, we extended the encoder-decoder architecture of CodeT5 with enhanced learning objectives.
  arXiv  Detail & Related papers  (2022-07-05T02:42:15Z)
• Optimization-Inspired Learning with Architecture Augmentations and Control Mechanisms for Low-Level Vision [74.9260745577362]
  This paper proposes a unified optimization-inspired learning framework to aggregate Generative, Discriminative, and Corrective (GDC) principles. We construct three propagative modules to effectively solve the optimization models with flexible combinations. Experiments across varied low-level vision tasks validate the efficacy and adaptability of GDC.
  arXiv  Detail & Related papers  (2020-12-10T03:24:53Z)
• A Learned Performance Model for Tensor Processing Units [5.733911161090224]
  We demonstrate a method of learning performance models from a corpus of graph programs for Processing Unit (TPU) instances. We show that our learned model outperforms a heavily-optimized analytical performance model on two tasks. It helps an autotuner discover faster programs in a setting where access to TPUs is limited or expensive.
  arXiv  Detail & Related papers  (2020-08-03T17:24:52Z)

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.