Abstract:Machine learning (ML) has been increasingly used in a variety of domains, while solving ML programming tasks poses unique challenges due to the fundamental difference in the nature and the construct of general programming tasks, especially for developers who do not have ML backgrounds. Automatic code generation that produces a code snippet from a natural language description can be a promising technique to accelerate ML programming tasks. In recent years, although many deep learning-based neural code generation models have been proposed with high accuracy, the fact that most of them are mainly evaluated on general programming tasks calls into question their effectiveness and usefulness in ML programming tasks. In this paper, we set out to investigate the effectiveness of existing neural code generation models on ML programming tasks. For our analysis, we select six state-of-the-art neural code generation models, and evaluate their performance on four widely used ML libraries, with newly created 83K pairs of natural-language described ML programming tasks. Our empirical study reveals some good, bad, and missing aspects of neural code generation models on ML tasks, with a few major ones listed below. ( Good ) Neural code generation models perform significantly better on ML tasks than on non-ML tasks with an average difference of 10.6 points in BLEU-4 scores. ( Bad ) More than 80% of the generated code is semantically incorrect. ( Bad ) Code generation models do not have significance in improving developers’ completion time. ( Good ) The generated code can help developers write correct code by providing developers with clues for using correct APIs. ( Missing ) The observation from our user study reveals the missing aspects of code generation for ML tasks, e.g., decomposing code generation for divide-and-conquer into API sequence identification and API usage generation.

Neural Attribute Machines for Program Generation

A Hierarchical Neural Autoencoder for Paragraphs and Documents

Neural Machine Translation for Code Generation

The Neural Network Pushdown Automaton: Model, Stack and Learning Simulations

Neural-Symbolic Recursive Machine for Systematic Generalization

Neural Rule-Execution Tracking Machine For Transformer-Based Text Generation

$N$-gram Is Back: Residual Learning of Neural Text Generation with $n$-gram Language Model

ReGAN: RE[LAX|BAR|INFORCE] based Sequence Generation using GANs

Unlocking the Power of GANs in Non-Autoregressive Text Generation

Neurosymbolic AI for Enhancing Instructability in Generative AI

Grammar-based Neural Text-to-SQL Generation

Plug and Play Language Models: A Simple Approach to Controlled Text Generation

The Good, the Bad, and the Missing: Neural Code Generation for Machine Learning Tasks

Brain Inspired Sequences Production by Spiking Neural Networks With Reward-Modulated STDP

Generalizations across filler-gap dependencies in neural language models

An Artificial Neuron for Enhanced Problem Solving in Large Language Models

Unsupervised Inflection Generation Using Neural Language Modeling

Tree-Structured Neural Machine for Linguistics-Aware Sentence Generation

Neural Boltzmann Machines

A Grammar-Based Structural CNN Decoder for Code Generation