Data science, music & programming
Ghent, Belgium
Joined January 2014
- Tweets487
- Following229
- Followers282
- Likes896
Dear @clipperteas and @Pukkaherbs, please put the tea names on top too.. Take some inspiration from @yogiteaeurope or @DelhaizeNL. A relaxing tea time doesn't warrant reorganizing our kitchen
HendrikD'Oosterlinck retweeted
Komaan jongens, het is ondertussen kwart over elf en jullie internet is nog niet hersteld.
HendrikD'Oosterlinck retweeted
Zeg @Telenet , wat is er aan de hand, hier in Sint amandsberg geen internet?
HendrikD'Oosterlinck retweeted
DSPy on ICL RAG Classification: Code explained
🎥 Watch here: youtube.com/watch?v=52wo6uvh…
HendrikD'Oosterlinck retweeted
Not one but two Infer-Retrieve-Rank releases today??? What a time to be alive. Great job by the @llama_index team 🚀
Infer-Retrieve-Rerank (@KarelDoostrlnck et al.) is a simple but powerful paradigm to use LLMs for complex classification problems with thousands of classes. Examples include medical reactions 🥼 and job skills/qualifications 👷.
Use an LLM to infer a set of predictions, use predictions to retrieve a set of labels, and use an LLM to rerank the labels.
If you’re using LLMs for classification problems, you’re in luck - we implemented as a LlamaPack template for you to easily use OR build from scratch! 💫
Disclaimer: We didn’t implement few-shot prompting, as implemented in DSPy (@lateinteraction). Check out the source paper for more details there.
LlamaPack: llamahub.ai/l/llama_packs-re…
Notebook guide: github.com/run-llama/llama-h…
Source paper: arxiv.org/pdf/2401.12178.pdf
HendrikD'Oosterlinck retweeted
📢Tasks with > 10k classes (e.g. information extraction) are hard for in-context learning: typically a tuned retriever or many in-context calls per input are used ($$$)
Infer-Retrieve-Rank (IReRa) is a SotA program using 1 frozen retriever with a query predictor and reranker.
HendrikD'Oosterlinck retweeted
We proudly present our 524 page book on equivariant convolutional networks.
Coauthored by Patrick Forré, @erikverlinde and @wellingmax.
maurice-weiler.gitlab.io/#cn…
[1/N]
HendrikD'Oosterlinck retweeted
Hier precies hetzelfde, maar dan met 160 tokens.
0,0 communicatie van @GraspopMetal of @weezevent
Laat de skully voor wat t was en zie het als een bug bounty fee.
@GraspopMetal #graspop @weezevent 256.9 tokens ('skullies') just vanished from my account. I just bought them this morning! There's been no communication whatsoever? What is going on? @Testaankoop
This is how the before looked btw: I bought 40 skullies 7 times as visible here, and now it says I bought 3,3 skullies 7 times...
HendrikD'Oosterlinck retweeted
Trots om één van de genomineerden voor Student Van Het Jaar @Stadgent te mogen zijn! Bedankt aan de vriendjes van @VTKGent en @GentseStud om niet vies te zijn van een ambitieus projectje of twee😉. Lees de interviews met mij en de andere kandidaten hier: stad.gent/nl/student-gent/ni…
HendrikD'Oosterlinck retweeted
Very proud to have been a small part in rolling out @openfaas at @waylay_io during my summer internship this year 🥰
We're excited to announce that @waylay_io has become a Platinum homepage sponsor for 2021!
Today, team Waylay launches a new low-code platform with powerful workflow and automation capabilities.
Check it out 👇
#serverless #orchestration #sponsorship
waylay.io/articles/waylay-io…
HendrikD'Oosterlinck retweeted
This Wednesday at 8pm, we are hosting a ‘#datascience bar’ on Clubhouse about recent trends and advancements in #NLP. Come join us for a fun chat!
Some people joining:
@yvespeirsman (@nlptown)
@JVHautte (Techwolf)
@FsMatt (@ml6team)
@OxyKodit (@explosion_ai)
HendrikD'Oosterlinck retweeted
Kom nu langs voor een gesprek over "AI in marketing" met @DDoosterlinck @vincent_spruyt @jonathanberte !
“AI and data infrastructure in marketing” with @vincent_spruyt, and Data Science Belgium. Tomorrow, 17 Mar at 8:00 PM CET on @joinclubhouse. Join us!
joinclubhouse.com/event/M8Az…
#BelgianAIWeek #ai4belgium @BelgiumData @DSGhent @dsleuven @DSBrussels