🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

Katya Khramtsova presenting our paper "Selecting which Dense Retriever to use for Zero-Shot Search" @SIGIR_AP #SIGIR_AP2023: can we select which dense retriever to use on your new dataset, without needing labels for evaluation? w Shengyao Zhuang Guido Zuccon arxiv.org/abs/2309.09403

now at #EMNLP2023 Bevan Koopman is presenting our paper "Dr ChatGPT tell me what I want to hear: How different prompts impact health answer correctness" ielab.io/files/emnlp202… Apparently the first Human-Centered NLP session at an NLP conference (so said the session chair)

🎉 Excited to share that our abstract "Incorporating Changes in Review Outcomes in the Evaluation of Systematic Review Automation" has been accepted for a short oral presentation at the Global Evidence Summit GESummit 2024. Work done with Allan Hanbury Petr Knoth 🇺🇦 Guido Zuccon 🙌

"The new paradigm in machine learning – foundation models, large language models and beyond: a primer for physicians" by Ian Scott & I (Guido Zuccon) is now available in Internal Medicine Journal: doi.org/10.1111/imj.16…

Our review describes the different types of LLM, their emerging applications and potential limitations and bias and likely future translation into clinical practice

Our new work "PromptReps: Prompting Large Language Models to Generate Dense and Sparse Representations for Zero-Shot Document Retrieval" (Shengyao Zhuang Xueguang Ma Bevan Koopman Jimmy Lin Guido Zuccon) arxiv.org/abs/2404.18424 a 🧵

This paper is receiving good attention, also featuring on the May 2024 Trends in AI by @ZetaVector. I am excited about this work, for several reasons. Here they come...

(1 -- perhaps the most trivial) this is the first work in Information Retrieval that reports results on Llama3, including the 70B model: and this was achieved within 10 days of model release.

This was possible thanks to @CSIRO_News new Virga GPU compute facility, that boasts 448 H100s. This is a great commitment to AI and in particular LLM research by CSIRO. It also highlights the need for a national GPU infra to foster research at Australian universities

(2) this work breaks away from previous attempts to use generative LLMs as encoders by being the first method that can create dense representations in a zero-shot manner and without requiring finetuning

(4) all code and results are publicly released (see github.com/ielab/PromptRe…) You can fully reproduce our results with just a few command lines as there is no expensive LLM training

(5) and finally, this is work in progress: we have new experiments running and we plan to roll out iterative updates to results as we get them finished and written up. so, stay tuned!

It was great to have Prateek Jain from Google DeepMind visiting UQ School of EECS last week, and talking to us about Efficient and Elastic Large Models. Efficiency in LLM inference and elasticity of models certainly big ticket items to solve for improved LLM adoption UQ News

We have a PhD scholarship and 2 postdoc openings in the ielab team to work on Information Retrieval and LLMs. Come and join us! 🧵

PhD scholarship: “The Future of Search: Reducing the Environmental Impact of Large Language Models with Quantum Computing" -- You will be part of the joint PhD program b/w UQ School of EECS & University of Exeter More info at scholarships.uq.edu.au/scholarship/qu… Quick, closes 28 June

Postdocs: we are seeking 2x postdocs for up to 3 years. In these positions you will develop your expertise and an emerging research profile in IR, NLP, and LLMs, with particular applications in the health domain. More info: uq.wd3.myworkdayjobs.com/en-US/uqcareer… Applications close on 2 August

Me, Bevan Koopman and part of the team will be at SIGIR 2025 2024 if you want to chat about the postdocs. I also will be at ESSIR #ESSIR2024 #SIGIR2024

Come and join us at The University of Queensland. UQ is in sunny Brisbane, the 3rd most populated city in Australia, close to great beaches and forests. Brisbane is very lively and liveable; it will host the 2032 olympics. UQ ranks among the top-50 universities in the world.

The ielab is very active in IR research with publications across SIGIR, EMNLP, WebConf, WSDM, ECIR among other top venues. You will join a dynamic and collaborative research team, with strong support for your research and career development

🎉 Excited to share the list of our ielab papers accepted at #SIGIR2025 ACM SIGIR! Topics include efficiency in IR, domain-specific IR (SysRev, podcast), screenshot retrievers, hallucination detection, and more. See you in Padua 🇮🇹! Details coming soon, stay tuned!