Picture for Trevor Cohn

Trevor Cohn

University of Melbourne

SEEP: Training Dynamics Grounds Latent Representation Search for Mitigating Backdoor Poisoning Attacks

Add code
May 19, 2024
Viaarxiv icon

Transferring Troubles: Cross-Lingual Transferability of Backdoor Attacks in LLMs with Instruction Tuning

Add code
Apr 30, 2024
Viaarxiv icon

Revisiting subword tokenization: A case study on affixal negation in large language models

Add code
Apr 04, 2024
Viaarxiv icon

Backdoor Attack on Multilingual Machine Translation

Apr 03, 2024
Viaarxiv icon

Pre-training Cross-lingual Open Domain Question Answering with Large-scale Synthetic Supervision

Add code
Feb 26, 2024
Viaarxiv icon

Simpson's Paradox and the Accuracy-Fluency Tradeoff in Translation

Add code
Feb 20, 2024
Figure 1 for Simpson's Paradox and the Accuracy-Fluency Tradeoff in Translation
Figure 2 for Simpson's Paradox and the Accuracy-Fluency Tradeoff in Translation
Figure 3 for Simpson's Paradox and the Accuracy-Fluency Tradeoff in Translation
Figure 4 for Simpson's Paradox and the Accuracy-Fluency Tradeoff in Translation
Viaarxiv icon

Predicting Human Translation Difficulty with Neural Machine Translation

Dec 19, 2023
Viaarxiv icon

Noisy Self-Training with Synthetic Queries for Dense Retrieval

Add code
Nov 27, 2023
Viaarxiv icon

Boot and Switch: Alternating Distillation for Zero-Shot Dense Retrieval

Add code
Nov 27, 2023
Viaarxiv icon

Multi-EuP: The Multilingual European Parliament Dataset for Analysis of Bias in Information Retrieval

Add code
Nov 03, 2023
Figure 1 for Multi-EuP: The Multilingual European Parliament Dataset for Analysis of Bias in Information Retrieval
Figure 2 for Multi-EuP: The Multilingual European Parliament Dataset for Analysis of Bias in Information Retrieval
Figure 3 for Multi-EuP: The Multilingual European Parliament Dataset for Analysis of Bias in Information Retrieval
Figure 4 for Multi-EuP: The Multilingual European Parliament Dataset for Analysis of Bias in Information Retrieval
Viaarxiv icon