# Improving the Validity of Automatically Generated Feedback via Reinforcement Learning

arXiv: https://arxiv.org/abs/2403.01304 · HTML: https://mlsanity.com/paper/2403.01304/

## Key facts

- Cited by 20 papers in the map (data through 14 Oct 2026).
- Placed in Process reward models for reasoning, one of 10 directions in LLM alignment / RLHF / preference optimisation.
- Builds on 25 of its 41 references that are in the map.
- First posted on arXiv 2 Mar 2024.

## Placed in

- LLM alignment / RLHF / preference optimisation / [Process reward models for reasoning](https://mlsanity.com/alignment/process-reward-models-for-reasoning/)
- Deep reinforcement learning methods / [RL training of language and agent models](https://mlsanity.com/deeprl/rl-training-of-language-and-agent-models/)

## Builds on (most cited first, 25 of 25)

- [LoRA: Low-Rank Adaptation of Large Language Models](https://mlsanity.com/paper/2106.09685/) (2021)
- [Llama 2: Open Foundation and Fine-Tuned Chat Models](https://mlsanity.com/paper/2307.09288/) (2023)
- [Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks](https://mlsanity.com/paper/1908.10084/) (2019)
- [Evaluating Large Language Models Trained on Code](https://mlsanity.com/paper/2107.03374/) (2021)
- [Direct Preference Optimization: Your Language Model is Secretly a Reward Model](https://mlsanity.com/paper/2305.18290/) (2023)
- [BERTScore: Evaluating Text Generation with BERT](https://mlsanity.com/paper/1904.09675/) (2019)
- [Large Language Models are Zero-Shot Reasoners](https://mlsanity.com/paper/2205.11916/) (2022)
- [Reflexion: Language Agents with Verbal Reinforcement Learning](https://mlsanity.com/paper/2303.11366/) (2023)
- [HuggingFace's Transformers: State-of-the-art Natural Language Processing](https://mlsanity.com/paper/1910.03771/) (2019)
- [Fine-Tuning Language Models from Human Preferences](https://mlsanity.com/paper/1909.08593/) (2019)
- [Program of Thoughts Prompting: Disentangling Computation from Reasoning for Numerical Reasoning Tasks](https://mlsanity.com/paper/2211.12588/) (2022)
- [Can Large Language Models Be an Alternative to Human Evaluations?](https://mlsanity.com/paper/2305.01937/) (2023)
- [Contrastive Learning with Hard Negative Samples](https://mlsanity.com/paper/2010.04592/) (2020)
- [Large Language Models Are State-of-the-Art Evaluators of Translation Quality](https://mlsanity.com/paper/2302.14520/) (2023)
- [Automated Distractor and Feedback Generation for Math Multiple-choice Questions via In-context Learning](https://mlsanity.com/paper/2308.03234/) (2023)
- [Mathematical Language Processing: Automatic Grading and Feedback for Open Response Mathematical Questions](https://mlsanity.com/paper/1501.04346/) (2015)
- [Evaluating ChatGPT's Decimal Skills and Feedback Generation in a Digital Learning Game](https://mlsanity.com/paper/2306.16639/) (2023)
- [Automatic Short Math Answer Grading via In-context Meta-learning](https://mlsanity.com/paper/2205.15219/) (2022)
- [Automated Personalized Feedback Improves Learning Gains in an Intelligent Tutoring System](https://mlsanity.com/paper/2005.02431/) (2020)
- [Novice Learner and Expert Tutor: Evaluating Math Reasoning Abilities of Large Language Models with Misconceptions](https://mlsanity.com/paper/2310.02439/) (2023)
- [Code Soliloquies for Accurate Calculations in Large Language Models](https://mlsanity.com/paper/2309.12161/) (2023)
- [Using Large Language Models to Assess Tutors' Performance in Reacting to Students Making Math Errors](https://mlsanity.com/paper/2401.03238/) (2024)
- [Math Operation Embeddings for Open-ended Solution Analysis and Feedback](https://mlsanity.com/paper/2104.12047/) (2021)
- [ALL-IN-ONE: Multi-Task Learning BERT models for Evaluating Peer Assessments](https://mlsanity.com/paper/2110.03895/) (2021)
- [Algebra Error Classification with Large Language Models](https://mlsanity.com/paper/2305.06163/) (2023)

## Built on by (most cited first, 20 of 20)

- [Exploring Knowledge Tracing in Tutor-Student Dialogues using LLMs](https://mlsanity.com/paper/2409.16490/) (2024)
- [Training LLM-based Tutors to Improve Student Learning Outcomes in Dialogues](https://mlsanity.com/paper/2503.06424/) (2025)
- [Math Multiple Choice Question Generation via Human-Large Language Model Collaboration](https://mlsanity.com/paper/2405.00864/) (2024)
- [Automatic Generation of Question Hints for Mathematics Problems using Large Language Models in Educational Technology](https://mlsanity.com/paper/2411.03495/) (2024)
- [From Text to Visuals: Using LLMs to Generate Math Diagrams with Vector Graphics](https://mlsanity.com/paper/2503.07429/) (2025)
- [FeedEval: Pedagogically Aligned Evaluation of LLM-Generated Essay Feedback](https://mlsanity.com/paper/2601.04574/) (2026)
- [Letting Tutor Personas Speak Up for LLMs: Learning Steering Vectors from Dialogue via Preference Optimization](https://mlsanity.com/paper/2602.07639/) (2026)
- [Revolutionizing Bridge Operation and Maintenance with LLM-based Agents: An Overview of Applications and Insights](https://mlsanity.com/paper/2407.10064/) (2024)
- [From Untamed Black Box to Interpretable Pedagogical Orchestration: The Ensemble of Specialized LLMs Architecture for Adaptive Tutoring](https://mlsanity.com/paper/2603.23990/) (2026)
- [Interpreting Latent Student Knowledge Representations in Programming Assignments](https://mlsanity.com/paper/2405.08213/) (2024)
- [Comparing RAG and GraphRAG for Page-Level Retrieval Question Answering on a Math Textbook](https://mlsanity.com/paper/2509.16780/) (2025)
- [Concept-based Rubrics Improve LLM Formative Assessment and Data Synthesis](https://mlsanity.com/paper/2504.03877/) (2025)
- [Exploring LLM-Generated Feedback for Economics Essays: How Teaching Assistants Evaluate and Envision Its Use](https://mlsanity.com/paper/2505.15596/) (2025)
- [Narrowing the Gap: Supervised Fine-Tuning of Open-Source LLMs as a Viable Alternative to Proprietary Models for Pedagogical Tools](https://mlsanity.com/paper/2507.05305/) (2025)
- [Cultivating Helpful, Personalized, and Creative AI Tutors: A Framework for Pedagogical Alignment using Reinforcement Learning](https://mlsanity.com/paper/2507.20335/) (2025)
- [REFINE: Real-world Exploration of Interactive Feedback and Student Behaviour](https://mlsanity.com/paper/2603.29142/) (2026)
- [Teaching Language Models How to Code Like Learners: Conversational Serialization for Student Simulation](https://mlsanity.com/paper/2604.10720/) (2026)
- [A Multi-Agent Approach to Validate and Refine LLM-Generated Personalized Math Problems](https://mlsanity.com/paper/2604.05160/) (2026)
- [Listen, Correct, and Feed Back: Spoken Pedagogical Feedback Generation](https://mlsanity.com/paper/2604.14177/) (2026)
- [Gumbel Machine: Counterfactual Student Writing Generation via Gumbel Noise Steering](https://mlsanity.com/paper/2605.27249/) (2026)

## Abstract (arXiv)

Automatically generating feedback via large language models (LLMs) in intelligent tutoring systems and online learning platforms has the potential to improve the learning outcomes of many students. However, both feedback generation and evaluation are challenging: feedback content has to be valid especially in subjects like math, which requires models to understand the problem, the solution, and where the student's error lies. Feedback also has to be pedagogically valid to reflect effective tutoring strategies, such as explaining possible misconceptions and encouraging the student, among other desirable features. In this work, we address both problems of automatically generating and evaluating feedback while considering both correctness and alignment. First, we propose a rubric for evaluating math feedback and show that GPT-4 is able to effectively use it to annotate human-written and LLM-generated feedback. Second, we propose a framework for feedback generation that optimizes both correctness and alignment using reinforcement learning (RL). Specifically, we use GPT-4's annotations to create preferences over feedback pairs in an augmented dataset for training via direct preference optimization (DPO). We show that our methods significantly increase the correctness and alignment of generated feedback with Llama 2, an open-source LLM, qualitatively analyze our generation and evaluation systems using case studies, and outline several areas for future work.