Latent Predictor Networks for Code Generation

Abstract

Many language generation tasks require the production of text conditioned on both structured and unstructured inputs. We present a novel neural network architecture which generates an output sequence conditioned on an arbitrary number of input functions. Crucially, our approach allows both the choice of conditioning context and the granularity of generation, for example characters or tokens, to be marginalised, thus permitting scalable and effective training. Using this framework, we address the problem of generating programming code from a mixed natural language and structured specification. We create two new data sets for this paradigm derived from the collectible trading card games Magic the Gathering and Hearthstone. On these, and a third preexisting corpus, we demonstrate that marginalising multiple predictors allows our model to outperform strong benchmarks.

References in corpus (6)

• Sequence to Sequence Learning with Neural Networks
• ADADELTA: An Adaptive Learning Rate Method
• Natural Language Processing (almost) from Scratch
• Exploring the Limits of Language Modeling
• Finding Function in Form: Compositional Character Models for Open Vocabulary Word Representation
• On End-to-End Program Generation from User Intention by Deep Neural Networks

Cited by in corpus (62)

• Evaluating Large Language Models Trained on Code
• A Deep Reinforced Model for Abstractive Summarization
• Pointer Sentinel Mixture Models
• A Network-based End-to-End Trainable Task-oriented Dialogue System
• On the Feasibility of Transfer-learning Code Smells using Deep Learning
• A Syntactic Neural Model for General-Purpose Code Generation
• Tree-to-tree Neural Networks for Program Translation
• StaQC: A Systematically Mined Question-Code Dataset from Stack Overflow
• Learning Python Code Suggestion with a Sparse Pointer Network
• A parallel corpus of Python functions and documentation strings for automated code documentation and code generation
• Deep Choice Model Using Pointer Networks for Airline Itinerary Prediction
• Language as a Latent Variable: Discrete Generative Models for Sentence Compression
• Abstract Syntax Networks for Code Generation and Semantic Parsing
• Learning to Generate Corrective Patches using Neural Machine Translation
• Latent Attention For If-Then Program Synthesis
• Program Induction by Rationale Generation : Learning to Solve and Explain Algebraic Word Problems
• pix2code: Generating Code from a Graphical User Interface Screenshot
• Program Synthesis with Large Language Models
• Insertion-based Decoding with automatically Inferred Generation Order
• EVIL: Exploiting Software via Natural Language
• Embedding API Dependency Graph for Neural Code Generation
• Neural Attribute Machines for Program Generation
• Relevance Transformer: Generating Concise Code Snippets with Relevance Feedback
• TRANX: A Transition-based Neural Abstract Syntax Parser for Semantic Parsing and Code Generation
• TreeGen: A Tree-Based Transformer Architecture for Code Generation
• Neural Program Search: Solving Programming Tasks from Description and Examples
• VirtualHome: Simulating Household Activities via Programs
• Synergy between Machine/Deep Learning and Software Engineering: How Far Are We?
• One-Shot Learning for Text-to-SQL Generation
• Neural Networks for Modeling Source Code Edits
• SPoC: Search-based Pseudocode to Code
• Can Neural Networks Understand Logical Entailment?
• Toward Code Generation: A Survey and Lessons from Semantic Parsing
• CoCoSum: Contextual Code Summarization with Multi-Relational Graph Neural Network
• Two-Stage Synthesis Networks for Transfer Learning in Machine Comprehension
• NAPS: Natural Program Synthesis Dataset
• Retrieval-Based Neural Code Generation
• TreeGAN: Syntax-Aware Sequence Generation with Generative Adversarial Networks
• Automatically Generating Commit Messages from Diffs using Neural Machine Translation
• Variable Name Recovery in Decompiled Binary Code using Constrained Masked Language Modeling
• JuICe: A Large Scale Distantly Supervised Dataset for Open Domain Context-based Code Generation
• Multi-Task Learning for Conversational Question Answering over a Large-Scale Knowledge Base
• Exploring Dynamic Selection of Branch Expansion Orders for Code Generation
• Jointly Learning Semantic Parser and Natural Language Generator via Dual Information Maximization
• CFGs-2-NLU: Sequence-to-Sequence Learning for Mapping Utterances to Semantics and Pragmatics
• Semantic Scaffolds for Pseudocode-to-Code Generation
• Sketch-Driven Regular Expression Generation from Natural Language and Examples
• Ain't Nobody Got Time For Coding: Structure-Aware Program Synthesis From Natural Language
• Natural Language Processing for Requirements Formalization: How to Derive New Approaches?
• Holistic Combination of Structural and Textual Code Information for Context based API Recommendation
• Multi-Source Pointer Network for Product Title Summarization
• ReAssert: Deep Learning for Assert Generation
• Effective Search of Logical Forms for Weakly Supervised Knowledge-Based Question Answering
• Exploring Neural Methods for Parsing Discourse Representation Structures
• A Neural-based Program Decompiler
• GANCoder: An Automatic Natural Language-to-Programming Language Translation Approach based on GAN
• Neural Networks Models for Analyzing Magic: the Gathering Cards
• Improving Automatic Source Code Summarization via Deep Reinforcement Learning
• Subword Language Model for Query Auto-Completion
• Sequence-to-Sequence Learning with Latent Neural Grammars
• Linguacodus: A Synergistic Framework for Transformative Code Generation in Machine Learning Pipelines
• Neural Programming by Example