Thomas Dehaene

NLP engineer and meme sommelier

[Satya Nadella @satyanadella]

We remain committed to our partnership with OpenAI and have confidence in our product roadmap, our ability to continue to innovate with everything we announced at Microsoft Ignite, and in continuing to support our customers and partners. We look forward to getting to know Emmett Shear and OAI's new leadership team and working with them. And we’re extremely excited to share the news that Sam Altman and Greg Brockman, together with colleagues, will be joining Microsoft to lead a new advanced AI research team. We look forward to moving quickly to provide them with the resources needed for their success.

[Raphaël Millière @raphaelmilliere]

I asked ChatGPT to rewrite Bohemian Rhapsody to be about the life of a postdoc, and the output was flawless:

[Farid @ai_fast_track]

😍 Lama Cleaner: A free and open-source inpainting tool powered by SOTA AI model

🎁 Completely Free and Open-Source

• Fully self-hosted
• Multiple SOTA AI models
• Classical image inpainting algorithm
• CPU & GPU
• Various inpainting strategy
• Run as a Desktop APP

[Thomas Dehaene @TDehaene]

It's that time of the year again 👀! https://t.co/AJlWUecwNk

Everything to know about the past year of happenings in #AI 🎁. My favorite graph: a showcase of how cross-modal #transformers have become 👉

[Thomas Dehaene @TDehaene]

@ClementDelangue @ml6team @huggingface Thanks for the shoutout @ClemDelangue , great to be on HF as well 😎! We're totally hooked

[clem 🤗 @ClementDelangue]

Cool to see @ml6team on @huggingface! https://t.co/RqFO67oR5L

[Thomas Dehaene @TDehaene]

Note: if the interactive version is down for some reason, you can check out the Medium version: https://t.co/RJnVqeO2Xy

[Thomas Dehaene @TDehaene]

#Transformers 🤖 for #summarization are sometimes prone to hallucinations 😵‍💫. Our #NLP team investigated some post-processing steps in an interactive blogpost on @huggingface
👉 https://t.co/PSWr5WzGky

[Suzana Ilić @suzatweet]

Machine Learning Operations (MLOps):
Overview, Definition, and Architecture
https://t.co/2MHAoBjfRN

[BigScience Large Model Training @BigScienceLLM]

▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓ 99%

[@emilymbender.bsky.social @emilymbender]

For those playing along at home, here's a "AI is sentient!" argument bingo card.

[Niels Rogge @NielsRogge]

🤩 LayoutLMv3 by @Microsoft is now available @huggingface!

🔥 The model replaces the CNN backbone of its predecessor with (much simpler) patch embeddings à la ViT.

🚀 SOTA performance on all document AI benchmarks, both image-only and text+image! (1/2)

[Thomas Dehaene @TDehaene]

💡 Adapting semantic search to a new domain with unlabeled data 🔍? Using #GPL from @UKPLab, we demonstrate how to do this on non-English datasets 🇩🇪 . 🔗: https://t.co/ToiqA8dHr0

[Hugging Face @huggingface]

Introducing the biggest change to https://t.co/dKZjob5Hcx since its inception 🔥

The Community Tab 🥳

read more and discuss: https://t.co/SI07ac47Z1

[Thomas Dehaene @TDehaene]

📝 I wrote an #NLP blogpost detailing how we optimized our Dutch #GPT2 model to reduce our Google Cloud bill by a factor of 2.4 🤑 ! 👉 https://t.co/DCU4XJV7Bt

[Thomas Dehaene @TDehaene]

🚨 new model alert 🚨. We're releasing: ✌️ translated Dutch summarization datasets (CNN-nl & XSUM-nl)
➕ a finetuned mBART model for Dutch summarization on @huggingface !
https://t.co/WQjkNsFL2t https://t.co/RUsmTF8ck7

[Thomas Dehaene @TDehaene]

It do be like that with #AI 🙈

[Thomas Dehaene @TDehaene]

When your funding round raises about as much dough as a BERT-Base model has parameters 🤯💵! Big congrats to @huggingface 💪! https://t.co/iJPhF0QugE

[Thomas Dehaene @TDehaene]

@RasmusToivanen @huggingface I'm assuming you'll come up with the next architecture after Transformers 😁

[Thomas Dehaene @TDehaene]

Our official @huggingface HuggingMugs (or MuggingFaces 🤔) have arrived ☕️! And already our Rouge scores have improved, our Perplexities are down and my eyesight is better 🤯