SPARTA
[EACL 2026] SPARTA: Evaluating Reasoning Segmentation Robustness through Black-Box Adversarial Paraphrasing in Text Autoencoder Latent Space
Install / Use
/learn @emb-ai/SPARTAREADME
SPARTA
<div align="center">Official Implementation of "SPARTA: Evaluating Reasoning Segmentation Robustness through Black-Box Adversarial Paraphrasing in Text Autoencoder Latent Space"
</div>📢 News
🚧 More updates coming soon! Stay tuned for additional features, documentation, and improvements.
📋 Overview
This repository contains the official implementation of the paper SPARTA (arXiv:2510.24446).
🚀 Getting Started
Installation
# create conda env
conda create -n rtext python==3.10.16
conda activate rtext
# install python packages
pip install -r requirements-attack.txt
pip install -e fairseq2
pip install -e SONAR
pip install -e LISA
pip install -e LISA_v0
pip install -e GSVA
pip install -e rtext
Related Skills
node-connect
345.4kDiagnose OpenClaw node connection and pairing failures for Android, iOS, and macOS companion apps
frontend-design
104.6kCreate distinctive, production-grade frontend interfaces with high design quality. Use this skill when the user asks to build web components, pages, or applications. Generates creative, polished code that avoids generic AI aesthetics.
openai-whisper-api
345.4kTranscribe audio via OpenAI Audio Transcriptions API (Whisper).
qqbot-media
345.4kQQBot 富媒体收发能力。使用 <qqmedia> 标签,系统根据文件扩展名自动识别类型(图片/语音/视频/文件)。
