# Learning to Compare for Better Training and Evaluation of Open Domain Natural Language Generation Models

arXiv: https://arxiv.org/abs/2002.05058 · HTML: https://mlsanity.com/paper/2002.05058/

## Key facts

- Cited by 28 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 12 Feb 2020.

## Built on by (most cited first, 28 of 28)

- [Training language models to follow instructions with human feedback](https://mlsanity.com/paper/2203.02155/) (2022)
- [Learning to summarize from human feedback](https://mlsanity.com/paper/2009.01325/) (2020)
- [ImageReward: Learning and Evaluating Human Preferences for Text-to-Image Generation](https://mlsanity.com/paper/2304.05977/) (2023)
- [Aligning Text-to-Image Models using Human Feedback](https://mlsanity.com/paper/2302.12192/) (2023)
- [Humans or LLMs as the Judge? A Study on Judgement Biases](https://mlsanity.com/paper/2402.10669/) (2024)
- [Recursively Summarizing Books with Human Feedback](https://mlsanity.com/paper/2109.10862/) (2021)
- [Chain of Hindsight Aligns Language Models with Feedback](https://mlsanity.com/paper/2302.02676/) (2023)
- [A Survey of Controllable Text Generation using Transformer-based Pre-trained Language Models](https://mlsanity.com/paper/2201.05337/) (2022)
- [A Comprehensive Assessment of Dialog Evaluation Metrics](https://mlsanity.com/paper/2106.03706/) (2021)
- [RLHF Deciphered: A Critical Analysis of Reinforcement Learning from Human Feedback for LLMs](https://mlsanity.com/paper/2404.08555/) (2024)
- [Loose lips sink ships: Mitigating Length Bias in Reinforcement Learning from Human Feedback](https://mlsanity.com/paper/2310.05199/) (2023)
- [Mixed Preference Optimization: Reinforcement Learning with Data Selection and Better Reference Model](https://mlsanity.com/paper/2403.19443/) (2024)
- [A Survey on Human Preference Learning for Large Language Models](https://mlsanity.com/paper/2406.11191/) (2024)
- [Generative Transformers for Design Concept Generation](https://mlsanity.com/paper/2211.03468/) (2022)
- [RePair: Automated Program Repair with Process-based Feedback](https://mlsanity.com/paper/2408.11296/) (2024)
- [Learning from Perturbations: Diverse and Informative Dialogue Generation with Inverse Adversarial Training](https://mlsanity.com/paper/2105.15171/) (2021)
- [ChatGPT vs State-of-the-Art Models: A Benchmarking Study in Keyphrase Generation Task](https://mlsanity.com/paper/2304.14177/) (2023)
- [Learning to Predict Persona Information forDialogue Personalization without Explicit Persona Description](https://mlsanity.com/paper/2111.15093/) (2021)
- [Exploring Dense Retrieval for Dialogue Response Selection](https://mlsanity.com/paper/2110.06612/) (2021)
- [JASMINE: Arabic GPT Models for Few-Shot Learning](https://mlsanity.com/paper/2212.10755/) (2022)
- [Human-in-the-loop Abstractive Dialogue Summarization](https://mlsanity.com/paper/2212.09750/) (2022)
- [AlignHuman: Improving Motion and Fidelity via Timestep-Segment Preference Optimization for Audio-Driven Human Animation](https://mlsanity.com/paper/2506.11144/) (2025)
- [Filter & Align: Leveraging Human Knowledge to Curate Image-Text Data](https://mlsanity.com/paper/2312.06726/) (2023)
- [Towards Trustworthy Multimodal Moderation via Policy-Aligned Reasoning and Hierarchical Labeling](https://mlsanity.com/paper/2508.03296/) (2025)
- [Retrieval-based Knowledge Transfer: An Effective Approach for Extreme Large Language Model Compression](https://mlsanity.com/paper/2310.15594/) (2023)
- [Collaborative Development of NLP models](https://mlsanity.com/paper/2305.12219/) (2023)
- [A Probabilistic Approach for Model Alignment with Human Comparisons](https://mlsanity.com/paper/2403.10771/) (2024)
- [Security in LLM-Assisted Coding: A Systematic Literature Review](https://mlsanity.com/paper/2610.15482/) (2026)

## Abstract (arXiv)

Automated evaluation of open domain natural language generation (NLG) models remains a challenge and widely used metrics such as BLEU and Perplexity can be misleading in some cases. In our paper, we propose to evaluate natural language generation models by learning to compare a pair of generated sentences by fine-tuning BERT, which has been shown to have good natural language understanding ability. We also propose to evaluate the model-level quality of NLG models with sample-level comparison results with skill rating system. While able to be trained in a fully self-supervised fashion, our model can be further fine-tuned with a little amount of human preference annotation to better imitate human judgment. In addition to evaluating trained models, we propose to apply our model as a performance indicator during training for better hyperparameter tuning and early-stopping. We evaluate our approach on both story generation and chit-chat dialogue response generation. Experimental results show that our model correlates better with human preference compared with previous automated evaluation approaches. Training with the proposed metric yields better performance in human evaluation, which further demonstrates the effectiveness of the proposed model.