Compile by Training: Turning Natural-Language Specifications into Reusable Neural Functions

Forum topic · 小凯 · 2026-09-05

Summary

Compile by training is a new technique presented by Yuntian Deng, Pengyu Nie, and Stuart Shieber (arXiv:2609.04199) that compiles a natural-language specification into a reusable neural function. At compile time, large teacher models generate task-specific examples used to train a small adapter attached to a compact interpreter. The resulting function runs independently of the teacher models, avoiding repeated costs, latency, and provider dependence of calling remote LLMs at inference time, and can be stored, versioned, and composed like ordinary software. On FuzzyBench-Hard—a benchmark subset where the Program-as-Weights fast compiler produced no exact matches—compile by training achieves 83.6% semantic accuracy. This higher accuracy trades off against compile-time cost: roughly one minute per compilation versus seconds for fast compilers. The authors deploy the compiler as a public interactive service and demonstrate compiled functions in three applications: a multi-site website assistant, a language-controlled 3D avatar, and a bidirectional English-Claudish translator.

Paper Overview

Field: NLP
Authors: Yuntian Deng, Pengyu Nie, Stuart Shieber
Published: 2026-09-03
arXiv: 2609.04199

Abstract

Many recurring text functions are easy to describe in natural language but hard to implement with rules, while calling large remote models for every input incurs repeated cost, latency, and provider dependence. The authors propose compile by training, which turns a natural-language specification into a reusable neural function.

At compile time, teacher models generate task-specific examples that are used to train a small adapter for a compact interpreter. The resulting function runs without the teachers and can be stored, versioned, and composed like ordinary software.

Results

On FuzzyBench-Hard—a subset on which the Program-as-Weights fast compiler failed to produce any exact matches—compile by training reaches 83.6% semantic accuracy.

Higher accuracy comes with higher compile-time cost: about one minute per compilation, compared to seconds for fast compilers.

Deployment

The compiler is deployed as a public interactive service, with compiled functions demonstrated in:

A multi-site website assistant
A language-controlled 3D avatar
A bidirectional English–Claudish translator

---

*Source: forum post, collected 2026-09-05.*