Lisa Alazraki

Hi! I am a third year PhD student at the Department of Computing of Imperial College London, supervised by Marek Rei. I am fascinated by large foundation models and work on enhancing their planning, reasoning, and general problem-solving abilities. I also have an interest in model robustness to input perturbations and adversarial attacks.
Currently I am at the Alan Turing Institute as an Enrichment Student. Previously I was a Research Intern at Cohere (2024) and Google (2022-23).

News

13 Feb 2025
Our new preprint LLMs can implicitly learn from mistakes in-context is available on ArXiv.

6 Dec 2024
Meta-Reasoning Improves Tool Use in Large Language Models' poster won first prize at the Imperial Computing Conference!

18 Oct 2024
I have been awarded a Placement Award on the Alan Turing Institute's Enrichment Scheme!

25 Jun 2024
I am happy to be among this year's IET Postgraduate Research Awards winners 🎉

24 Jun 2024
I am starting a research internship at Cohere, where I will be until ~~October~~ December.

Papers

	LLMs can implicitly learn from mistakes in-context Lisa Alazraki, Maximilian Mozes, Jon Ander Campos, Yi Chern Tan, Marek Rei, Max Bartolo ArXiv preprint, 2025 repository / arXiv

	Enhancing LLM Robustness to Perturbed Instructions: An Empirical Study Aryan Agrawal, Lisa Alazraki*, Shahin Honarvar, Marek Rei (Equal contribution) BuildingTrust @ICLR, 2025 repository / preprint

	How Can Representation Dimension Dominate Structurally Pruned LLMs? Mingxue Xu, Lisa Alazraki, Danilo P. Mandic SLLM @ICLR, 2025 arXiv

	Meta-Reasoning Improves Tool Use in Large Language Models Lisa Alazraki, Marek Rei NAACL Findings, 2025 repository / arXiv

	How (not) to ensemble LVLMs for VQA Lisa Alazraki, Lluis Castrejon, Mostafa Dehghani, Fantine Huot, Jasper Uijlings, Thomas Mensink PMLR Proceedings on ICBINB @NeurIPS, 2023 proceedings / arXiv

Teaching

I am a Teaching Assistant for 70050 Intro to Machine Learning, 70016 Natural Language Processing, 70010 Deep Learning and 40008 Graphs and Algorithms.

Supervision

2025
Kevin Zhou 'Optimal Data Mixing Strategies at Different Token Budgets for Language Model Training', MSc dissertation, co-supervised with Marek Rei (Imperial College London) and Kris Cao (Cohere).

2024
Aryan Agrawal 'Improving the Robustness of LLMs to Prompt Perturbations', MSc dissertation, co-supervised with Marek Rei (Imperial College London), Thomas Mensink (Google Research) and Shahin Honarvar (Imperial College London).

2023
Ziheng Zhang 'Knowledge-Enhanced Supportive Memory for Chatbots', MEng dissertation, co-supervised with the Algorithmic Human Development group (Imperal College London) and Zissis Poulos (University of Toronto).

2022
Alicia Law 'A Multilingual Chatbot for Self-Attachment Therapy', MSc dissertation (Distinguished Dissertation award), co-supervised with the Algorithmic Human Development group (Imperial College London).

Service

Conference reviewing: ACL 2025, COLING 2025, COLM 2025, NAACL 2025
Workshop reviewing: BuildingTrust @ICLR 2025, ICBINB @ICLR 2025, MARW @AAAI 2025

Forked from Jon Barron's website.