← Search

Hua Farn

2 accepted papers

2025

Safeguard Fine-Tuned LLMs Through Pre- and Post-Tuning Model Merging

EMNLP 2025

Fine-tuning large language models (LLMs) for downstream tasks often leads to catastrophic forgetting, notably degrading the safety of originally aligned models. While some existing methods attempt to restore safety by incorporating additional safety data, the quality of such data typically falls sho

Cited by 0SourcePDFScholar
2024

Task Arithmetic can Mitigate Synthetic-to-Real Gap in Automatic Speech Recognition

EMNLP 2024main

Synthetic data is widely used in speech recognition due to the availability of text-to-speech models, which facilitate adapting models to previously unseen text domains. However, existing methods suffer in performance when they fine-tune an automatic speech recognition (ASR) model on synthetic data…