Saved in:
Bibliographic Details
Main Authors: Yuan, Ziqi, Zhang, Baozheng, Xu, Hua, Liang, Zhiyun, Gao, Kai
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2407.02773
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866907966670635008
author Yuan, Ziqi
Zhang, Baozheng
Xu, Hua
Liang, Zhiyun
Gao, Kai
author_facet Yuan, Ziqi
Zhang, Baozheng
Xu, Hua
Liang, Zhiyun
Gao, Kai
contents We present OpenVNA, an open-source framework designed for analyzing the behavior of multimodal language understanding systems under noisy conditions. OpenVNA serves as an intuitive toolkit tailored for researchers, facilitating convenience batch-level robustness evaluation and on-the-fly instance-level demonstration. It primarily features a benchmark Python library for assessing global model robustness, offering high flexibility and extensibility, thereby enabling customization with user-defined noise types and models. Additionally, a GUI-based interface has been developed to intuitively analyze local model behavior. In this paper, we delineate the design principles and utilization of the created library and GUI-based web platform. Currently, OpenVNA is publicly accessible at \url{https://github.com/thuiar/OpenVNA}, with a demonstration video available at \url{https://youtu.be/0Z9cW7RGct4}.
format Preprint
id arxiv_https___arxiv_org_abs_2407_02773
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle OpenVNA: A Framework for Analyzing the Behavior of Multimodal Language Understanding System under Noisy Scenarios
Yuan, Ziqi
Zhang, Baozheng
Xu, Hua
Liang, Zhiyun
Gao, Kai
Multimedia
We present OpenVNA, an open-source framework designed for analyzing the behavior of multimodal language understanding systems under noisy conditions. OpenVNA serves as an intuitive toolkit tailored for researchers, facilitating convenience batch-level robustness evaluation and on-the-fly instance-level demonstration. It primarily features a benchmark Python library for assessing global model robustness, offering high flexibility and extensibility, thereby enabling customization with user-defined noise types and models. Additionally, a GUI-based interface has been developed to intuitively analyze local model behavior. In this paper, we delineate the design principles and utilization of the created library and GUI-based web platform. Currently, OpenVNA is publicly accessible at \url{https://github.com/thuiar/OpenVNA}, with a demonstration video available at \url{https://youtu.be/0Z9cW7RGct4}.
title OpenVNA: A Framework for Analyzing the Behavior of Multimodal Language Understanding System under Noisy Scenarios
topic Multimedia
url https://arxiv.org/abs/2407.02773