| Challenge: | Existing methods to replace human annotation are expensive and limited. |
| Approach: | They investigate the use of synthetic data in Fact Verification and Evidence-based Question Answering by replacing human-generated data with synthetic points on eight diverse datasets. |
| Outcome: | The proposed method shows promise but performance declines when replacing up to 90% of training data with synthetic data are severe . the proposed method can be used to improve models trained on purely synthetic data by including as few as 125 human-generated data points. |
Similar Papers
Synthetic Data in the Era of Large Language Models (2025.acl-tutorials)
Copied to clipboard
| Challenge: | 'synthetic data' is a data generated with the assistance of large language models to make dataset construction faster and cheaper. |
| Approach: | This tutorial seeks to build a shared understanding of recent progress in synthetic data generation from NLP and related fields by grouping and describing major methods, applications, and open problems. |
| Outcome: | This tutorial will describe methods, applications, and open problems that have been developed and are being used to improve the quality and efficiency of synthetic data generation. |
Improving Question Answering Model Robustness with Synthetic Adversarial Data Generation (2021.emnlp-main)
Copied to clipboard
| Challenge: | a new approach to generate adversarial data is needed to improve question answering models . crowdworkers can fool a model only 8.8% of the time, compared to 17.6% for a trained model without synthetic data. |
| Approach: | They develop a pipeline that generates questions and then filters or labels them to improve quality. |
| Outcome: | The proposed approach improves state-of-the-art on a human-written adversarial dataset by 3.7F1 and improves model generalisation on nine of the twelve MRQA datasets. |
Little Giants: Synthesizing High-Quality Embedding Data at Scale (2025.naacl-long)
Copied to clipboard
| Challenge: | Synthetic data generation is an increasingly popular way of training models without the need for large, manually labeled datasets. |
| Approach: | They propose a framework that aligns open-source small models to efficiently generate large-scale embedding data. |
| Outcome: | The proposed framework outperforms state-of-the-art embedding models by using only 1/10 of the GPT API calls. |
Exploiting Asymmetry for Synthetic Training Data Generation: SynthIE and the Case of Information Extraction (2023.emnlp-main)
Copied to clipboard
| Challenge: | Large language models (LLMs) have great potential for synthetic data generation. |
| Approach: | They show that large language models can generate useful data even for complex tasks . they use a symmetric task difficulty asymmetry to prompt an LLM to generate plausible input text for a target output structure. |
| Outcome: | The proposed approach outperforms existing models by a substantial margin on closed information extraction tasks with 1.8M data points and 770M parameters. |
Better Synthetic Data by Retrieving and Transforming Existing Datasets (2024.findings-acl)
Copied to clipboard
| Challenge: | despite advances in large language models, task-specific data is not available for many use cases . a new method to improve automated dataset generation uses publicly available datasets . |
| Approach: | They propose a method to make better use of existing datasets to improve automatic dataset generation. |
| Outcome: | The proposed method outperforms existing methods on language-based tasks . it significantly increases diversity and difficulty of generated data on many tasks compared to other methods . |
Wronging a Right: Generating Better Errors to Improve Grammatical Error Detection (D18-1)
Copied to clipboard
| Challenge: | grammatical error correction is a labor-intensive task that requires large amounts of training data. |
| Approach: | They propose to use a human-annotated corpus of human-generated grammatical errors to generate a synthetic model. |
| Outcome: | The proposed method outperforms the current state of the art in grammatical error correction . human annotators achieve 39.39 F1 scores, suggesting the model generates mostly human-like instances . |
Let’s Synthesize Step by Step: Iterative Dataset Synthesis with Large Language Models by Extrapolating Errors from Small Models (2023.findings-emnlp)
Copied to clipboard
| Challenge: | *Data Synthesis* is a promising way to train a small model with very little labeled data. |
| Approach: | They propose a framework that iteratively extrapolates the errors of a small model trained on a real-world validation dataset using a large language model. |
| Outcome: | The proposed framework reduces the gap between the synthesized dataset and the real data . it improves on multiple NLP tasks and on large models with human-annotated data. |
Understanding the Influence of Synthetic Data for Text Embedders (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent advances in general purpose text embedders have been driven by training on synthetic training data. |
| Approach: | They propose to use GPT-4 to produce high quality synthetic data that expands existing training datasets for embeddings to new tasks. |
| Outcome: | The proposed dataset is high quality and leads to consistent improvements in performance. |
A Pipeline for Generating, Annotating and Employing Synthetic Data for Real World Question Answering (2022.emnlp-demos)
Copied to clipboard
| Challenge: | Question Answering (QA) is a growing area of research . state-of-the-art QA models struggle on out-of domain documents without fine-tuning . |
| Approach: | They propose a pipeline for validating and training QA data and an interface for human annotation. |
| Outcome: | The proposed pipeline improves QA performance on domain-specific datasets while preserving the accuracy of the model. |
Automatic Detection of Generated Text is Easiest when Humans are Fooled (2020.acl-main)
Copied to clipboard
| Challenge: | Recent advances in neural language modelling make it possible to rapidly generate vast amounts of human-sounding text. |
| Approach: | They compare decoding methods with popular sampling-based decoding strategies . they show that multi-sentence excerpts can fool expert human raters over 30% of the time . |
| Outcome: | The proposed methods improve with longer excerpt length, but multi-sentence excerpts fool human raters over 30% of the time. |