# Type Prediction With Program Decomposition and Fill-in-the-Type Training

arXiv: https://arxiv.org/abs/2305.17145 · HTML: https://mlsanity.com/paper/2305.17145/

## Key facts

- Cited by 6 papers in the map (data through 14 Oct 2026).
- Builds on 29 of its 66 references that are in the map.
- First posted on arXiv 25 May 2023.

## Builds on (most cited first, 29 of 29)

- [Evaluating Large Language Models Trained on Code](https://mlsanity.com/paper/2107.03374/) (2021)
- [Program Synthesis with Large Language Models](https://mlsanity.com/paper/2108.07732/) (2021)
- [CodeGen: An Open Large Language Model for Code with Multi-Turn Program Synthesis](https://mlsanity.com/paper/2203.13474/) (2022)
- [Large Language Models Can Be Easily Distracted by Irrelevant Context](https://mlsanity.com/paper/2302.00093/) (2023)
- [StarCoder: may the source be with you!](https://mlsanity.com/paper/2305.06161/) (2023)
- [CodeSearchNet Challenge: Evaluating the State of Semantic Code Search](https://mlsanity.com/paper/1909.09436/) (2019)
- [InCoder: A Generative Model for Code Infilling and Synthesis](https://mlsanity.com/paper/2204.05999/) (2022)
- [A Systematic Evaluation of Large Language Models of Code](https://mlsanity.com/paper/2202.13169/) (2022)
- [The Stack: 3 TB of permissively licensed source code](https://mlsanity.com/paper/2211.15533/) (2022)
- [Efficient Training of Language Models to Fill in the Middle](https://mlsanity.com/paper/2207.14255/) (2022)
- [SantaCoder: don't reach for the stars!](https://mlsanity.com/paper/2301.03988/) (2023)
- [Multi-lingual Evaluation of Code Generation Models](https://mlsanity.com/paper/2210.14868/) (2022)
- [Do Long-Range Language Models Actually Use Long-Range Context?](https://mlsanity.com/paper/2109.09115/) (2021)
- [CodeFill: Multi-token Code Completion by Jointly Learning from Structure and Naming Sequences](https://mlsanity.com/paper/2202.06689/) (2022)
- [PanGu-Coder: Program Synthesis with Function-Level Language Modeling](https://mlsanity.com/paper/2207.11280/) (2022)
- [LambdaNet: Probabilistic Type Inference using Graph Neural Networks](https://mlsanity.com/paper/2005.02161/) (2020)
- [TypeWriter: Neural Type Prediction with Search-based Validation](https://mlsanity.com/paper/1912.03768/) (2019)
- [Typilus: Neural Type Hints](https://mlsanity.com/paper/2004.10657/) (2020)
- [Type4Py: Practical Deep Similarity Learning-Based Type Inference for Python](https://mlsanity.com/paper/2101.04470/) (2021)
- [OptTyper: Probabilistic Type Inference by Optimising Logical and Natural Constraints](https://mlsanity.com/paper/2004.00348/) (2020)
- [ManyTypes4Py: A Benchmark Python Dataset for Machine Learning-based Type Inference](https://mlsanity.com/paper/2104.04706/) (2021)
- [Do Machine Learning Models Produce TypeScript Types That Type Check?](https://mlsanity.com/paper/2302.12163/) (2023)
- [PYInfer: Deep Learning Semantic Type Inference for Python Variables](https://mlsanity.com/paper/2106.14316/) (2021)
- [Solver-based Gradual Type Migration](https://mlsanity.com/paper/2109.05049/) (2021)
- [Type Inference for Static Compilation of JavaScript (Extended Version)](https://mlsanity.com/paper/1608.07261/) (2016)
- [Dynamic Type Inference for Gradual Hindley--Milner Typing](https://mlsanity.com/paper/1810.12619/) (2018)
- [Practical Optional Types for Clojure](https://mlsanity.com/paper/1812.03571/) (2018)
- [A Gradual Type System for Elixir](https://mlsanity.com/paper/2104.08366/) (2021)
- [Gradual Soundness: Lessons from Static Python](https://mlsanity.com/paper/2206.13831/) (2022)

## Built on by (most cited first, 6 of 6)

- [Unifying the Perspectives of NLP and Software Engineering: A Survey on Language Models for Code](https://mlsanity.com/paper/2311.07989/) (2023)
- [Knowledge Transfer from High-Resource to Low-Resource Programming Languages for Code LLMs](https://mlsanity.com/paper/2308.09895/) (2023)
- [VerMCTS: Synthesizing Multi-Step Programs using a Verifier, a Large Language Model, and Tree Search](https://mlsanity.com/paper/2402.08147/) (2024)
- [AgenticTyper: Automated Typing of Legacy Software Projects Using Agentic AI](https://mlsanity.com/paper/2602.21251/) (2026)
- [The Power of Types: Exploring the Impact of Type Checking on Neural Bug Detection in Dynamically Typed Languages](https://mlsanity.com/paper/2411.15368/) (2024)
- [TypyBench: Evaluating LLM Type Inference for Untyped Python Repositories](https://mlsanity.com/paper/2507.22086/) (2025)

## Abstract (arXiv)

TypeScript and Python are two programming languages that support optional type annotations, which are useful but tedious to introduce and maintain. This has motivated automated type prediction: given an untyped program, produce a well-typed output program. Large language models (LLMs) are promising for type prediction, but there are challenges: fill-in-the-middle performs poorly, programs may not fit into the context window, generated types may not type check, and it is difficult to measure how well-typed the output program is. We address these challenges by building OpenTau, a search-based approach for type prediction that leverages large language models. We propose a new metric for type prediction quality, give a tree-based program decomposition that searches a space of generated types, and present fill-in-the-type fine-tuning for LLMs. We evaluate our work with a new dataset for TypeScript type prediction, and show that 47.4% of files type check (14.5% absolute improvement) with an overall rate of 3.3 type errors per file. All code, data, and models are available at: https://github.com/GammaTauAI/opentau.