# Improving Code Generation by Training with Natural Language Feedback

arXiv: https://arxiv.org/abs/2303.16749 · HTML: https://mlsanity.com/paper/2303.16749/

## Key facts

- Cited by 76 papers in the map (data through 14 Oct 2026).
- Builds on 28 of its 36 references that are in the map.
- First posted on arXiv 28 Mar 2023.

## Builds on (most cited first, 28 of 28)

- [PyTorch: An Imperative Style, High-Performance Deep Learning Library](https://mlsanity.com/paper/1912.01703/) (2019)
- [Training language models to follow instructions with human feedback](https://mlsanity.com/paper/2203.02155/) (2022)
- [Chain-of-Thought Prompting Elicits Reasoning in Large Language Models](https://mlsanity.com/paper/2201.11903/) (2022)
- [Evaluating Large Language Models Trained on Code](https://mlsanity.com/paper/2107.03374/) (2021)
- [Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedback](https://mlsanity.com/paper/2204.05862/) (2022)
- [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)
- [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)
- [Solving math word problems with process- and outcome-based feedback](https://mlsanity.com/paper/2211.14275/) (2022)
- [CodeGen: An Open Large Language Model for Code with Multi-Turn Program Synthesis](https://mlsanity.com/paper/2203.13474/) (2022)
- [GPT-NeoX-20B: An Open-Source Autoregressive Language Model](https://mlsanity.com/paper/2204.06745/) (2022)
- [e-SNLI: Natural Language Inference with Natural Language Explanations](https://mlsanity.com/paper/1812.01193/) (2018)
- [A Systematic Evaluation of Large Language Models of Code](https://mlsanity.com/paper/2202.13169/) (2022)
- [Can language models learn from explanations in context?](https://mlsanity.com/paper/2204.02329/) (2022)
- [Measuring Progress on Scalable Oversight for Large Language Models](https://mlsanity.com/paper/2211.03540/) (2022)
- [SPoC: Search-based Pseudocode to Code](https://mlsanity.com/paper/1906.04908/) (2019)
- [A Survey of Reinforcement Learning Informed by Natural Language](https://mlsanity.com/paper/1906.03926/) (2019)
- [Large Language Models are Few-shot Testers: Exploring LLM-based General Bug Reproduction](https://mlsanity.com/paper/2209.11515/) (2022)
- [Productivity Assessment of Neural Code Completion](https://mlsanity.com/paper/2205.06537/) (2022)
- [Evaluating Explanations: How much do explanations from the teacher aid students?](https://mlsanity.com/paper/2012.00893/) (2020)
- [CERT: Continual Pre-Training on Sketches for Library-Oriented Code Generation](https://mlsanity.com/paper/2206.06888/) (2022)
- [When Can Models Learn From Explanations? A Formal Framework for Understanding the Roles of Explanation Data](https://mlsanity.com/paper/2102.02201/) (2021)
- [Training Language Models with Language Feedback](https://mlsanity.com/paper/2204.14146/) (2022)
- [Speak to your Parser: Interactive Text-to-SQL with Natural Language Feedback](https://mlsanity.com/paper/2005.02539/) (2020)

## Built on by (most cited first, 30 of 76)

- [Teaching Large Language Models to Self-Debug](https://mlsanity.com/paper/2304.05128/) (2023)
- [Large Language Models as Optimizers](https://mlsanity.com/paper/2309.03409/) (2023)
- [Open Problems and Fundamental Limitations of Reinforcement Learning from Human Feedback](https://mlsanity.com/paper/2307.15217/) (2023)
- [AlpacaFarm: A Simulation Framework for Methods that Learn from Human Feedback](https://mlsanity.com/paper/2305.14387/) (2023)
- [Fine-Grained Human Feedback Gives Better Rewards for Language Model Training](https://mlsanity.com/paper/2306.01693/) (2023)
- [Large Language Models for Software Engineering: A Systematic Literature Review](https://mlsanity.com/paper/2308.10620/) (2023)
- [HuatuoGPT-o1, Towards Medical Complex Reasoning with LLMs](https://mlsanity.com/paper/2412.18925/) (2024)
- [InterCode: Standardizing and Benchmarking Interactive Coding with Execution Feedback](https://mlsanity.com/paper/2306.14898/) (2023)
- [Automatically Correcting Large Language Models: Surveying the landscape of diverse self-correction strategies](https://mlsanity.com/paper/2308.03188/) (2023)
- [Vision-Language Models are Zero-Shot Reward Models for Reinforcement Learning](https://mlsanity.com/paper/2310.12921/) (2023)
- [RLEF: Grounding Code LLMs in Execution Feedback with Reinforcement Learning](https://mlsanity.com/paper/2410.02089/) (2024)
- [Generalized Planning in PDDL Domains with Pretrained Large Language Models](https://mlsanity.com/paper/2305.11014/) (2023)
- [Debug like a Human: A Large Language Model Debugger via Verifying Runtime Execution Step-by-step](https://mlsanity.com/paper/2402.16906/) (2024)
- [GLoRe: When, Where, and How to Improve LLM Reasoning via Global and Local Refinements](https://mlsanity.com/paper/2402.10963/) (2024)
- [RLHF Deciphered: A Critical Analysis of Reinforcement Learning from Human Feedback for LLMs](https://mlsanity.com/paper/2404.08555/) (2024)
- [If LLM Is the Wizard, Then Code Is the Wand: A Survey on How Code Empowers Large Language Models to Serve as Intelligent Agents](https://mlsanity.com/paper/2401.00812/) (2024)
- [Large Language Models for Supply Chain Optimization](https://mlsanity.com/paper/2307.03875/) (2023)
- [GPT4AIGChip: Towards Next-Generation AI Accelerator Design Automation via Large Language Models](https://mlsanity.com/paper/2309.10730/) (2023)
- [CodeChain: Towards Modular Code Generation Through Chain of Self-revisions with Representative Sub-modules](https://mlsanity.com/paper/2310.08992/) (2023)
- [CodeTransOcean: A Comprehensive Multilingual Benchmark for Code Translation](https://mlsanity.com/paper/2310.04951/) (2023)
- [Learning to Generate Better Than Your LLM](https://mlsanity.com/paper/2306.11816/) (2023)
- [Reasoning in Token Economies: Budget-Aware Evaluation of LLM Reasoning Strategies](https://mlsanity.com/paper/2406.06461/) (2024)
- [A Closer Look at the Self-Verification Abilities of Large Language Models in Logical Reasoning](https://mlsanity.com/paper/2311.07954/) (2023)
- [LeDex: Training LLMs to Better Self-Debug and Explain Code](https://mlsanity.com/paper/2405.18649/) (2024)
- [CodeTree: Agent-guided Tree Search for Code Generation with Large Language Models](https://mlsanity.com/paper/2411.04329/) (2024)
- [CodeCoT: Tackling Code Syntax Errors in CoT Reasoning for Code Generation](https://mlsanity.com/paper/2308.08784/) (2023)
- [LeTI: Learning to Generate from Textual Interactions](https://mlsanity.com/paper/2305.10314/) (2023)
- [Can We Rely on LLM Agents to Draft Long-Horizon Plans? Let's Take TravelPlanner as an Example](https://mlsanity.com/paper/2408.06318/) (2024)
- [Sample, Don't Search: Rethinking Test-Time Alignment for Language Models](https://mlsanity.com/paper/2504.03790/) (2025)
- [Exploring and Characterizing Large Language Models For Embedded System Development and Debugging](https://mlsanity.com/paper/2307.03817/) (2023)

## Abstract (arXiv)

The potential for pre-trained large language models (LLMs) to use natural language feedback at inference time has been an exciting recent development. We build upon this observation by formalizing an algorithm for learning from natural language feedback at training time instead, which we call Imitation learning from Language Feedback (ILF). ILF requires only a small amount of human-written feedback during training and does not require the same feedback at test time, making it both user-friendly and sample-efficient. We further show that ILF can be seen as a form of minimizing the KL divergence to the ground truth distribution and demonstrate a proof-of-concept on a neural program synthesis task. We use ILF to improve a Codegen-Mono 6.1B model's pass@1 rate by 38% relative (and 10% absolute) on the Mostly Basic Python Problems (MBPP) benchmark, outperforming both fine-tuning on MBPP and fine-tuning on repaired programs written by humans. Overall, our results suggest that learning from human-written natural language feedback is both more effective and sample-efficient than training exclusively on demonstrations for improving an LLM's performance on code generation tasks.