Research map / Adversarial robustness and security / Certified and empirical robustness
Seq2Sick: Evaluating the Robustness of Sequence-to-Sequence Models with Adversarial Examples
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Increase the targeted keyword count from 1 to 3 | Seq2Sick · Gigaword, text summarization | Success%: 99.8% -> 43.0% | Hurt |
| Increase the targeted keyword count from 1 to 3 | Seq2Sick · DUC2003, text summarization | Success%: 99.6% -> 38.3% | Hurt |
| Increase the targeted keyword count from 1 to 3 | Seq2Sick · DUC2004, text summarization | Success%: 99.6% -> 37.4% | Hurt |
| Increase the targeted keyword count from 1 to 3 | Seq2Sick · Multi30k, English-German translation | Success%: 100.0% -> 69.6% | Hurt |
| Use a subword Transformer instead of the paper's translation model | Seq2Sick targeted 1-keyword attack · Machine translation | Success rate: 17% lower | Hurt |
| Use a subword Transformer instead of the paper's translation model | Seq2Sick targeted 1-keyword attack · Machine translation | BLEU: 0.13 lower | Hurt |
Closest papers by meaning
- TextGrad: Advancing Robustness Evaluation in NLP by Gradient-Driven Optimization (ICLR 2023 · similarity 0.87)
- Detecting Egregious Responses in Neural Sequence-to-sequence Models (ICLR 2019 · similarity 0.87)
- Chain of Attack: On the Robustness of Vision-Language Models Against Transfer-Based Adversarial Attacks (CVPR 2025 · similarity 0.85)
- Revisiting Character-level Adversarial Attacks for Language Models (ICML 2024 · similarity 0.84)
- Efficient Adversarial Training in LLMs with Continuous Attacks (NeurIPS 2024 · similarity 0.84)
- Improved Generation of Adversarial Examples Against Safety-aligned LLMs (NeurIPS 2024 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.