# A Self-Training Method for Machine Reading Comprehension with Soft Evidence Extraction

arXiv: https://arxiv.org/abs/2005.05189 · HTML: https://mlsanity.com/paper/2005.05189/

## Key facts

- Cited by 9 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 11 May 2020.

## Built on by (most cited first, 9 of 9)

- [Self-training Improves Pre-training for Few-shot Learning in Task-oriented Dialog Systems](https://mlsanity.com/paper/2108.12589/) (2021)
- [Robustifying Multi-hop QA through Pseudo-Evidentiality Training](https://mlsanity.com/paper/2107.03242/) (2021)
- [Self-Training with Differentiable Teacher](https://mlsanity.com/paper/2109.07049/) (2021)
- [Understanding and Improving Zero-shot Multi-hop Reasoning in Generative Question Answering](https://mlsanity.com/paper/2210.04234/) (2022)
- [REPT: Bridging Language Models and Machine Reading Comprehension via Retrieval-Based Pre-training](https://mlsanity.com/paper/2105.04201/) (2021)
- [Do Answers to Boolean Questions Need Explanations? Yes](https://mlsanity.com/paper/2112.07772/) (2021)
- [ECoRAG: Evidentiality-guided Compression for Long Context RAG](https://mlsanity.com/paper/2506.05167/) (2025)
- [Multi-grained Evidence Inference for Multi-choice Reading Comprehension](https://mlsanity.com/paper/2310.18070/) (2023)
- [U3E: Unsupervised and Erasure-based Evidence Extraction for Machine Reading Comprehension](https://mlsanity.com/paper/2210.02621/) (2022)

## Abstract (arXiv)

Neural models have achieved great success on machine reading comprehension (MRC), many of which typically consist of two components: an evidence extractor and an answer predictor. The former seeks the most relevant information from a reference text, while the latter is to locate or generate answers from the extracted evidence. Despite the importance of evidence labels for training the evidence extractor, they are not cheaply accessible, particularly in many non-extractive MRC tasks such as YES/NO question answering and multi-choice MRC. To address this problem, we present a Self-Training method (STM), which supervises the evidence extractor with auto-generated evidence labels in an iterative process. At each iteration, a base MRC model is trained with golden answers and noisy evidence labels. The trained model will predict pseudo evidence labels as extra supervision in the next iteration. We evaluate STM on seven datasets over three MRC tasks. Experimental results demonstrate the improvement on existing MRC models, and we also analyze how and why such a self-training method works in MRC. The source code can be obtained from https://github.com/SparkJiao/Self-Training-MRC