Robert Turnbull

Just out in BioScience: our new article on Hespi, an LLM-powered pipeline for extracting information from herbarium sheets. 🌿 doi.org/10.1093/biosci… Watch this space for lots more developments in this area!

🎉 Delighted to join Prof Christina Kreinecker and the @KU_Leuven team on the ERC BICROSS project, working on multilingual manuscripts using AI. Announcement: theo.kuleuven.be/en/research/re…

Looking forward to giving guest lectures on LLMs for the Data Journalism course at the University of Melbourne over the next couple of days

Mammograms are painful and only 2D. CT could be better—but needs high dosage to reduce noise. Using phase-contrast and AI denoising, we cut the dose 16× with no loss of image quality. We continue to work towards safe, 3D breast cancer screening. Preprint doi.org/10.48550/arXiv…

Our paper "Potnia: A Python library for the conversion of transliterated ancient texts to Unicode" is now published in @JOSS_TheOJ 👉 doi.org/10.21105/joss.… Potnia currently supports Linear A, Linear B, Arabic, and Hittite cuneiform—with more Anatolian languages on the way

I've got a little python package called llmloader to load LLMs in @LangChainAI just by the model name. I just updated it to include @MistralAI : github.com/rbturnbull/llm…

Excited to share Terrier 🧬✨🐶 — our new AI model to classify repetitive DNA. DNA repeats make up huge parts of genomes and play a key role in evolution, but are notoriously hard to classify. Preprint just released: doi.org/10.48550/arXiv…

Yesterday I had my book launch and I was honoured to have @TourEmily, @mbird12, Alexander Treiger, Stephen Carlson and Joey McCollum speak about the book. Here is a playlist of what they had to say: youtube.com/playlist?list=…

I’m working with Dr Peta Taylor (Lecturer in Animal Science at unimelb) on the welfare of chickens in transport. Peta will present some of our results at the Australian Poultry Science Symposium in Sydney on Wednesday.

I'm excited to co-supervise Jacob Cumming's PhD in Maths at the University of Melbourne. He's studying branching processes with applications to New Testament textual criticism. More here: robturnbull.com

A couple of weeks ago @joerogan suggested to @WesleyLHuff that it would be great if AI could sort out NT textual criticism. Here's a new LLM tool as a small step in that direction: github.com/rbturnbull/rdg… Paper out soon.

I just released a little command-line app to take object detection files in YOLO format and draw bounding boxes on an image. It can also take a @ultralytics YOLO model and predict the bounding boxes and write the image: github.com/rbturnbull/dra…

Lots of great papers at the upcoming Birmingham Colloquium on the Textual Criticism of the New Testament by @itsee_bham. I'll be presenting with Joey McCollum on an Arabic version of Ephesians.

My book about a newly discovered Arabic translation of the Gospels is out: brill.com/display/title/… Book launch Feb 13: events.humanitix.com/book-launch-co… On zoom and in person. Afternoon tea provided by @MDAP_Unimelb.

Jeremy Howard is a @unimelb philosophy grad who became a pioneer data scientist and created a leading framework for artificial intelligence. Hear him at the Victorian GPGPU Research Symposium (Oct 20–22): gpgpu.net.au #gpgpu #fastai @jeremyphoward

Here is the social media campaign that I've been working on at Melbourne Uni. Please share!

The ETC blog (@ETCtomwas) is running a giveaway of ‘To Cast the First Stone: The Transmission of a Gospel Story’ by @jennifer_knust and @tommywasserman

Luke 12. Parable of the Rich Fool - before John Dickson's 4th talk at #MadeGood: ow.ly/ehTe8

Luke 6 - Sermon on the Mount/Plain. Storytelling before John Dickson's 1st talk at #MadeGood: ow.ly/ehhq0 [chapter no. fixed]

Psalm 22 and Mark 15 - from my storytellings before John Dickson's 3rd talk at #MadeGood: ow.ly/ehhkw