Elisa Beshero-Bondar

Here's us unfurling one of the new Keystone DH banners--beginning of a great #keystonedh24 adventure! @KeystoneDH

@LorrieAnn25 @RmThaman @krassenstein If the past admins were so successful at it than why does the Big Beautiful Bill pour millions of dollars into ICE? Also, your comparison numbers are pretty skewed considering Trump has only been in office a few months. Look at the rate of deportations and project out 4 years.

@RmThaman @krassenstein 🤣🤣🤣🤣 nope! 75% of Obama deportations were done with zero due process!!

BREAKING: An Unlawful assembly has been declared by DHS at the federal building in Los Angeles this 4th of July. I don’t recall Americans being this upset about what was happening when Biden was in charge.

"Rolled round in earth's diurnal course with rocks and stones and trees" open.substack.com/pub/ebeshero/p…

@patrickDurusau No, it’s the writing and synthesis of those sources that worries me. Not so much the location of them. We have ways of telling whether sources are good and authentic. But the writing itself? Hmm.

@epyllia Do we really need to spoon feed sources to students? Sorry, I'm feeling cranky tonight.

#LLMs #Humanities Are students required to use sources in non-English languages because of the translation capabilities of LLMs? It seems like a no-brainer to expose them to a broader range of literature and how to compare LLMs. @epyllia

@patrickDurusau ? Every research assignment is different. Some done in class, some in mentored projects. I don’t worry too much about use of AI to locate sources in projects I mentor bc we use Zotero and I guide them in stockpiling interesting library and web resources w good metadata.

@roopikarisam Now ask it to help you with some regex matches…

I dictated an upcoming talk while driving to work then had Otter transcribe it. The AI-generated summary I did not ask for is so good I'm horrified.

@dmimno Thanks for the heads-up! I never liked Google CoLab’s space restrictions and have tended to want my students to work more independently on their own machines—with all the bespoke trouble thereunto appertaining.

PSA: if you assign python notebooks for homework, you CANNOT allow students to use Google colab anymore, it reads instructions and autosuggests good answers. I’m trying “You’ve earned the right not to play on easy mode”, but the temptation to just say “that looks good” is there

Want to test Medieval TrOCR Models on an entire manuscript page? This is a @Gradio app on @huggingface that let's you do that! Kraken is used for line detection. I'll add the ability to output the data as JSON if there's interest. Let me know if you want to see that added. #Medieval #HTR #MachineLearning #DataScience

The work done to revise sex + gender representation in TEI markup is SO IMPORTANT. Previously it was 0 = Not known 1 = Male 2 = Female 9 = Not applicable We literally had to (or work around) encoding women/females as #2. Among many other probs that structure posed. THANK YOU!

@UTHumanitiesCtr @utksoa @UTDTGallery @ArtsSciencesUT @UTKDailyBeacon @marcoinstitute @MedievalAcademy @jgob @UTKLibraries Hi all! I am excited to deliver this presentation, but you have not listed my current institutional affiliation! I am (full) Professor of Digital Humanities at Penn State Behrend since 2020!

@utksoa @UTDTGallery @ArtsSciencesUT @UTKDailyBeacon @marcoinstitute @MedievalAcademy Elisa Beshero-Bondar will round out our Distinguished Lecture Series' fall semester lineup with a talk that is part of our "Dialogues" mini-series in #DigitalHumanities. #PublicHumanities #Knoxville #UTK

We're excited to announce our 2024-2025 DCHA Distinguished Lecture Series! All talks are free & open to the public, with a livestream option available for those unable to attend in person. Details on our website: humanitiescenter.utk.edu. #PublicHumanities #Knoxville

Do you know anyone doing something inventive or cutting-edge with the TEI? Would you like the chance to celebrate their achievement? Nominate them for the Rahtz Prize for TEI Ingenuity by August 14th: tei-c.org/activities/rah….

@jamescummings Seems ironic for X to be so assiduously protective of youth (and youthful institutions like DHAwards). I wonder if anyone at #X central will answer if you write to them directly?

@jamescummings @Ted_Underwood @roopikarisam Not forgetting @friede , @heatherfro , and some authors I care about like @KateElliottSFF , etc. But yes, I do believe migration is past due for the likes of us!

@jamescummings @Ted_Underwood I’m still here only bc you, Ted, @roopikarisam and some others still post things here that I care about. But fwiw, the W3C isn’t even on X anymore. I’m more active on Mastodon and Bluesky these days and enjoy checking in there.

@kchonyc I agree, we need to be making our own models, but also need to work on lowering that barrier for entry.

@kchonyc In practice, at least in the past year, rolling your own model isn’t so easy to just soup up in an undergrad class or project… I think it could be better. We need up-to-date documentation. $$ Resources required (like processing capacity) does seem to be lower these days…

it is necessary to fully understand your system powered by an LLM, but these lack of transparency, maintainability and compliance prevents us from doing so. so, what can we do? - we must be able to train them ourselves. (9/12)

@kchonyc But the documentation for how to make/work w open-source models on HuggingFace gets dated quickly. Not enough people working in here, and script libraries age fast. If the docs you’re reading are from 2022, something will break inexplicably. It’s a time suck.

Not quite accurate ^^^ ! 4 new attributes! But not many. :-D

And we managed to introduce a chapter with only one new element and two new attributes, pulling largely from resources available across the TEI. Looking forward to seeing what we do with CMC! tei-c.org/release/doc/te…

Check out the new chapter in the TEI Guidelines on Computer-Mediated Communication (now chapter 9)…new with this release! Very happy we were able to complete this.