MarineLives

What's top of my mind? My Generative Lives agenda for the next three months

Here's my agenda for the next three months: generative-ai, historical research and archival practice: generativelives.substack.com/p/whats-top-... #earlymodernhistory #history #archives #ai #ml

GitHub - Addaci/HCA: English High Court of Admiralty deposition volumes [HCA 13/20 to HCA 13/79, covering the years 1570 to 1685] English High Court of Admiralty deposition volumes [HCA 13/20 to HCA 13/79, covering the years 1570 to 1685] - Addaci/HCA

Welcome to BlueSky. Our website is currently down for maintenance and improvement), but you can check out some of our transcriptions for English High Court of Admiralty records late C16th/early to mid-C17th here: github.com/Addaci/HCA

Introducing myself: Colin Greenstreet, research public historian A user's perspective on generative ai for historical research and archival practice

I am starting a Substack called Generative Lives. This offers a user's perspective on generative-ai and its potential application to historical research and archival practice. The Substack is free and seeks to encourage hands-on experimentation. open.substack.com/pub/generati...

Thanks. Looks interesting. I see Rainer Simon is a former member of the tech/research team.

Ten years ago to this day, THE FUTURE OF THE PROFESSIONS, by Daniel Susskind and me, was published by Oxford University Press. Together, Dan and I set out to explore the impact of 'increasingly… |... Ten years ago to this day, THE FUTURE OF THE PROFESSIONS, by Daniel Susskind and me, was published by Oxford University Press. Together, Dan and I set out to explore the impact of 'increasingly capabl...

Nice creative link. Thank you. In return, a little exercise I just ran using three LLMs in response to a LinkedIn posting by Richrd Susskind, author (with Daniel Susskind) of "The Future of the Professions (2015). www.linkedin.com/feed/update/...

Hello BlueSky. I haven't been active for a while. I am looking for examples of museums, libraries and archives doing interesting things with generative ai and would welcome suggestions of websites to look at. #earlymodern #archives #libraryscience

Here are the Yorkshire data. Scarborough and Hull.

Did ships get bigger in the C17th? Yes. Contact me if you would like access to the data and would like to discuss. And of course the devil is in the detail, but this is a relatively and well defined large dataset #earlymodern #marinehistory @thiagokrause.bsky.social @brodiewaddell.bsky.social

Note that it is broadly correctly identifying the pen strokes, and is reading data such as "1645". For marks, where the notary has added the actual name and latin text, it can distingusih and read pretty accurately

It is analysing the signature. I have run a load of tests with and without the caption

A potential use of AI image analysis which could be done at scale as part of an analysis of historic literacy? I loaded images of 140 signature, initial(s), mark(s) into NotebookLM & interrogated them palaeographically (with zero training). @thiagokrause.bsky.social @timhitchcock.bsky.social

From memory HCA 13/66 is one of the HCA 13/ series without folio markings. So to reference your truffle trove you end up referencing your own photo IDs?

Thanks @thiagokrause.bsky.social You should have access already to this specific GoogleNotebookLM. Though you may need a paid Gemini AI account to get mindmsp and 50 languages podcast features. Will resend you link

Looking for a Turkish speaker who would be willing to sample a couple of minutes of a Turkish podcast generated by Google NoteBook LM Plus [podcasts in 50 languages, handles 300 sources, very useful mindmaps]. Can send .wav file or give NB access @thiagokrause.bsky.social @timhitchcock.bsky.social

Many different types of vessel in the SOLM-QUANT database, all waiting to be characterised quantitatively and qualitatively

Thanks for getting in touch @ametria.org . See my detailed response in chat

Looking for two or three Early Modern marine historians, marine archaeologists, and knowledgeable enthusiasts interested in exploring novel, as yet unpublished 5,000+ ship database [SOLM-QUANT] for C17th commercial (and naval) ships - Get in touch to learn more #marinehistory #earlymodern

Much appreciated Brodie. I will see if I can get you a chapter in a couple of weeks

An extract from four case studies of ship to shore and shore to ship operational communication, a section within CH 6: Strategy, Operations and Geography, CH 6: Strategy, Operations and Geography. @brodiewaddell.bsky.social

Thanks. I am putting together a synopsis and couple of sample chapters together with sample appendices. The full draft will be finished by end June.

Here is a brief intro

I have been busy these last four months. Now looking for a publisher. Any suggestions. Can supply draft chapters and appendices for anyone interested in taking a look. @timhitchcock.bsky.social @brodiewaddell.bsky.social

Enrollment is now open for the 4th edition of the Master's in Computational Social Sciences at UC3M! No prior knowledge of programming or stats is needed—perfect for social sciences grads. Questions? Contact us or ask to be connected with our student ambassadors!

Please, share!

We have a nice new shiny organization card for MarineLives on Hugging Face, where you can find models and datasets huggingface.co/MarineLives

4/ I am performing the test fine-tune from a Colab notebook, and am monitoring the evaluation, training and systems data from a Weights and Biases monitoring screen. I'm using a paid T4 GPU @beismo.bsky.social

3/ Everything looks good so far. I am three epochs into a test run of a ten epoch fine-tune. @beismo.bsky.social @timhitchcock.bsky.social

2/ My first fine-tuning run is going to be a fine-tune of mBert (110M parameter LLM) using paired line Yiddish Hebrew text. I am running it for 10 epochs as a test. I got Gemini to modify some existing code I have @beismo.bsky.social @timhitchcock.bsky.social

I am working with @beismo.bsky.social to fine-tune two small large language models to correct raw machine transcription of Yiddish in Hebrew script. We plan to show the results this coming Tuesday at our next ai-and-history-collaboratory @timhitchcock.bsky.social @mark-l-thompson.bsky.social