# Coarse-Tuning Models of Code with Reinforcement Learning Feedback

arXiv: https://arxiv.org/abs/2305.18341 · HTML: https://mlsanity.com/paper/2305.18341/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Builds on 34 of its 46 references that are in the map.
- First posted on arXiv 25 May 2023.

## Builds on (most cited first, 30 of 34)

- [Attention Is All You Need](https://mlsanity.com/paper/1706.03762/) (2017)
- [BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding](https://mlsanity.com/paper/1810.04805/) (2018)
- [Language Models are Few-Shot Learners](https://mlsanity.com/paper/2005.14165/) (2020)
- [Training language models to follow instructions with human feedback](https://mlsanity.com/paper/2203.02155/) (2022)
- [Proximal Policy Optimization Algorithms](https://mlsanity.com/paper/1707.06347/) (2017)
- [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)
- [Learning to summarize from human feedback](https://mlsanity.com/paper/2009.01325/) (2020)
- [The Pile: An 800GB Dataset of Diverse Text for Language Modeling](https://mlsanity.com/paper/2101.00027/) (2020)
- [Fine-Tuning Language Models from Human Preferences](https://mlsanity.com/paper/1909.08593/) (2019)
- [High-Dimensional Continuous Control Using Generalized Advantage Estimation](https://mlsanity.com/paper/1506.02438/) (2015)
- [Generative Adversarial Imitation Learning](https://mlsanity.com/paper/1606.03476/) (2016)
- [Competition-Level Code Generation with AlphaCode](https://mlsanity.com/paper/2203.07814/) (2022)
- [CodeBERT: A Pre-Trained Model for Programming and Natural Languages](https://mlsanity.com/paper/2002.08155/) (2020)
- [CodeT5: Identifier-aware Unified Pre-trained Encoder-Decoder Models for Code Understanding and Generation](https://mlsanity.com/paper/2109.00859/) (2021)
- [CodeSearchNet Challenge: Evaluating the State of Semantic Code Search](https://mlsanity.com/paper/1909.09436/) (2019)
- [GraphCodeBERT: Pre-training Code Representations with Data Flow](https://mlsanity.com/paper/2009.08366/) (2020)
- [InCoder: A Generative Model for Code Infilling and Synthesis](https://mlsanity.com/paper/2204.05999/) (2022)
- [Unified Pre-training for Program Understanding and Generation](https://mlsanity.com/paper/2103.06333/) (2021)
- [CodeRL: Mastering Code Generation through Pretrained Models and Deep Reinforcement Learning](https://mlsanity.com/paper/2207.01780/) (2022)
- [Efficient Training of Language Models to Fill in the Middle](https://mlsanity.com/paper/2207.14255/) (2022)
- [CodeNet: A Large-Scale AI for Code Dataset for Learning a Diversity of Coding Tasks](https://mlsanity.com/paper/2105.12655/) (2021)
- [Multi-lingual Evaluation of Code Generation Models](https://mlsanity.com/paper/2210.14868/) (2022)
- [Write, Execute, Assess: Program Synthesis with a REPL](https://mlsanity.com/paper/1906.04604/) (2019)
- [Contrastive Code Representation Learning](https://mlsanity.com/paper/2007.04973/) (2020)
- [DOBF: A Deobfuscation Pre-Training Objective for Programming Languages](https://mlsanity.com/paper/2102.07492/) (2021)
- [Adversarial Examples for Models of Code](https://mlsanity.com/paper/1910.07517/) (2019)
- [CoditT5: Pretraining for Source Code and Natural Language Editing](https://mlsanity.com/paper/2208.05446/) (2022)
- [Latent Execution for Neural Program Synthesis](https://mlsanity.com/paper/2107.00101/) (2021)
- [Fault-Aware Neural Code Rankers](https://mlsanity.com/paper/2206.03865/) (2022)

## Built on by (most cited first, 2 of 2)

- [Agents in Software Engineering: Survey, Landscape, and Vision](https://mlsanity.com/paper/2409.09030/) (2024)
- [Tests as Prompt: A Test-Driven-Development Benchmark for LLM Code Generation](https://mlsanity.com/paper/2505.09027/) (2025)

## Abstract (arXiv)

Large Language Models (LLMs) pre-trained on code have recently emerged as the dominant approach to program synthesis. However, these models are trained using next-token prediction, which ignores the syntax and semantics of code. We propose RLCF, that further trains a pre-trained LLM via reinforcement learning, using feedback from a grounding function that scores the quality of the code. The grounding function uses (i) compiler-derived feedback on whether the code it generates passes a set of correctness checks; and (ii) feedback from a different LLM that compares the generated code to a reference code. RLCF is model- and language-agnostic. We empirically evaluate it on the MBJP and MathQA tasks for Java. Our experiments show that RLCF raises the odds that an LLM-generated program compiles, is executable, and produces the right output on tests, often allowing LLMs to match the performance of 2x-8x larger LLMs.