# TF-Coder: Program Synthesis for Tensor Manipulations

arXiv: https://arxiv.org/abs/2003.09040 · HTML: https://mlsanity.com/paper/2003.09040/

## Key facts

- Cited by 27 papers in the map (data through 7 Oct 2026).
- First posted on arXiv 19 Mar 2020.

## Built on by (most cited first, 27 of 27)

- [Fuzz4All: Universal Fuzzing with Large Language Models](https://mlsanity.com/paper/2308.04748/) (2023)
- [Natural Language to Code Generation in Interactive Data Science Notebooks](https://mlsanity.com/paper/2212.09248/) (2022)
- [Self-Improving Language Models for Evolutionary Program Synthesis: A Case Study on ARC-AGI](https://mlsanity.com/paper/2507.14172/) (2025)
- [CrossBeam: Learning to Search in Bottom-Up Program Synthesis](https://mlsanity.com/paper/2203.10452/) (2022)
- [ExeDec: Execution Decomposition for Compositional Generalization in Neural Program Synthesis](https://mlsanity.com/paper/2307.13883/) (2023)
- [Shellcode_IA32: A Dataset for Automatic Shellcode Generation](https://mlsanity.com/paper/2104.13100/) (2021)
- [HYSYNTH: Context-Free LLM Approximation for Guiding Program Synthesis](https://mlsanity.com/paper/2405.15880/) (2024)
- [Non-Programmers Can Label Programs Indirectly via Active Examples: A Case Study with Text-to-SQL](https://mlsanity.com/paper/2205.12422/) (2022)
- [Program Synthesis with Best-First Bottom-Up Search](https://mlsanity.com/paper/2310.04327/) (2023)
- [Grounding Data Science Code Generation with Input-Output Specifications](https://mlsanity.com/paper/2402.08073/) (2024)
- [Scaling Neural Program Synthesis with Distribution-based Search](https://mlsanity.com/paper/2110.12485/) (2021)
- [Compositional Generalization and Decomposition in Neural Program Synthesis](https://mlsanity.com/paper/2204.03758/) (2022)
- [NeuRI: Diversifying DNN Generation via Inductive Rule Inference](https://mlsanity.com/paper/2302.02261/) (2023)
- [LambdaBeam: Neural Program Search with Higher-Order Functions and Lambdas](https://mlsanity.com/paper/2306.02049/) (2023)
- [A Survey on Artificial Intelligence for Source Code: A Dialogue Systems Perspective](https://mlsanity.com/paper/2202.04847/) (2022)
- [Contextualized Data-Wrangling Code Generation in Computational Notebooks](https://mlsanity.com/paper/2409.13551/) (2024)
- [Towards A Measure Of General Machine Intelligence](https://mlsanity.com/paper/2109.12075/) (2021)
- [Can NMT Understand Me? Towards Perturbation-based Evaluation of NMT Models for Code Generation](https://mlsanity.com/paper/2203.15319/) (2022)
- [AbstractBeam: Enhancing Bottom-Up Program Synthesis using Library Learning](https://mlsanity.com/paper/2405.17514/) (2024)
- [Cardiverse: Harnessing LLMs for Novel Card Game Prototyping](https://mlsanity.com/paper/2502.07128/) (2025)
- [Can We Generate Shellcodes via Natural Language? An Empirical Study](https://mlsanity.com/paper/2202.03755/) (2022)
- [WikiCoder: Learning to Write Knowledge-Powered Code](https://mlsanity.com/paper/2303.08574/) (2023)
- [mlirSynth: Automatic, Retargetable Program Raising in Multi-Level IR using Program Synthesis](https://mlsanity.com/paper/2310.04196/) (2023)
- [Syno: Structured Synthesis for Neural Operators](https://mlsanity.com/paper/2410.23745/) (2024)
- [APRIL: API Synthesis with Automatic Prompt Optimization and Reinforcement Learning](https://mlsanity.com/paper/2509.25196/) (2025)
- [Lossless Tensor Compression as Program Synthesis](https://mlsanity.com/paper/2608.02162/) (2026)
- [Vectorizer: Vectorizing NumPy Programs with Shape-Guided Rewrite](https://mlsanity.com/paper/2609.08088/) (2026)

## Abstract (arXiv)

The success and popularity of deep learning is on the rise, partially due to powerful deep learning frameworks such as TensorFlow and PyTorch that make it easier to develop deep learning models. However, these libraries also come with steep learning curves, since programming in these frameworks is quite different from traditional imperative programming with explicit loops and conditionals. In this work, we present a tool called TF-Coder for programming by example in TensorFlow. TF-Coder uses a bottom-up weighted enumerative search, with value-based pruning of equivalent expressions and flexible type- and value-based filtering to ensure that expressions adhere to various requirements imposed by the TensorFlow library. We train models to predict TensorFlow operations from features of the input and output tensors and natural language descriptions of tasks, to prioritize relevant operations during search. TF-Coder solves 63 of 70 real-world tasks within 5 minutes, sometimes finding simpler solutions in less time compared to experienced human programmers.