Homepage of Dr Pasquale Minervini Researcher/Faculty at the University of Edinburgh , School of Informatics Co-Founder and CTO at Miniml.AI ELLIS Scholar, Edinburgh Unit
Our group has a few papers at ACL 2026 in San Diego (2–7 July) and ICML 2026 in Seoul (6–11 July, at the COEX Convention & Exhibition Center), covering GUI agents, efficient generation, reasoning, robustness, and mechanistic interpretability. If you are attending either conference, come say hi! Here is where to find us and what we will be presenting. ICML 2026 – GUI Agents & Efficient Generation…
I’ve been running a small benchmark, harness-bench , that pairs local LLMs (served via llama.cpp ’s llama-server ) with agent harnesses (Aider, Claude Code, OpenCode, Pi, Qwen CLI) on 16 software-engineering tasks across Python, PyTorch, JAX, C, C++, Rust, and SQL. Each (model, harness, task) cell is sandboxed: the agent only sees a scratch workspace/ and grading is done by a hidden test.sh that…
To make my life a bit easier, I built deep-research-mcp , a small Python agent that exposes several “deep research” backends through a single Model Context Protocol server ( Anthropic, 2024 ). It allows Claude Code, Codex, Gemini CLI, or any MCP client to fire off long-running research tasks against whichever backend the user prefers. What the server exposes The MCP server exposes deep_research ,…
Assume we have a scalar function $f(x)$ of interest, such as a reward we want to maximise or a loss we want to minimise, and that $x$ is drawn from a distribution $p_{\theta}(x)$ parameterised by $\theta$. A natural quantity to study is the expected value of $f$ under this distribution, \[J(\theta) = \mathbb{E}_{x \sim p_{\theta}(x)}[f(x)],\] and we would like to optimise $J(\theta)$ with respect…
Academic research sometimes risks to be disconnected from real-world applications. Research from our group demonstrated significant real-world impact across multiple domains, from improving the efficiency of LLM inference and training to new state-of-the-art evaluation protocols, and contributing to several industry products. Industry Adoption KV Cache Compression Our work on KV cache compression…
We have been working on language model evaluation, knowledge utilization, efficiency, and multimodal reasoning. We had papers at ICLR 2025 , NAACL 2025 (x3) , AAAI 2025 , and others, along with several ongoing works. NAACL 2025 – Controlling Knowledge & Reasoning Steering Knowledge Selection Behaviours in LLMs via SAE-Based Representation Engineering , by Yu Zhao et al. – We introduce SpARE , a…
Amazing opportunity to join our team at the School of Informatics, University of Edinburgh ! The School of Informatics is seeking a Postdoctoral Research Associate to work on evaluating and improving multimodal foundation models, with a particular focus on Vision-Language Models (VLMs). About the Position This is a full-time position running until January 2029, fully funded by the AI Hub in…
My amazing collaborators will be presenting three papers at EMNLP 2024 (main track), a leading conference in natural language processing, happening in Miami later this month! A few weeks ago I also blogged about our ACL 2024 , ICML 2024 , and CoLM 2024 papers – you can check the post here . Our work at EMNLP 2024 We will be presenting three papers this year at EMNLP, a flagship NLP conference: A…
My amazing collaborators will be presenting several works at ACL 2024 , ICML 2024 , and CoLM 2024 in the upcoming weeks/months! Our work at ACL 2024 We will be presenting four papers this year at ACL, the flagship NLP conference: Analysing The Impact of Sequence Composition on Language Model Pre-Training , by Yu Zhao et al. – we analyse several language model pre-training schemes and find out…
We have an opening for a 3-year postdoc – more details are available here – on a project funded by Huawei via the Huawei-Edinburgh Joint Lab initiative, with me as the Principal Investigator (PI). The researcher will work on projects involving the design and application of improving the robustness and trustworthiness of Large Language Models when solving complex reasoning tasks, while improving…