For example, in semantic search, we index a corpus of documents, with each document containing valuable information on a specific topic. Due to the way embedding models work, those documents will need to be chunked, and similarity is determined by chunk-level comparisons to the input query vector. Then, these similar chunks are returned back to the user. By finding an effective chunking strategy, we can ensure our search results accurately capture the essence of the user’s query.

If our chunks are too small or too large, it may lead to imprecise search results or missed opportunities to surface relevant content. As a rule of thumb, if the chunk of text makes sense without the surrounding context to a human, it will make sense to the language model as well. Therefore, finding the optimal chunk size for the documents in the corpus is crucial to ensuring that the search results are accurate and relevant.

These models find semantically similar sentences within one language or across languages:

sentence-transformers/distiluse-base-multilingual-cased-v1: Multilingual knowledge distilled version of multilingual Universal Sentence Encoder. Supports 15 languages: Arabic, Chinese, Dutch, English, French, German, Italian, Korean, Polish, Portuguese, Russian, Spanish, Turkish.

The issue with multilingual BERT (mBERT) as well as with XLM-RoBERTa is that those produce rather bad sentence representation out-of-the-box. Further, the vectors spaces between languages are not aligned, i.e., the sentences with the same content in different languages would be mapped to different locations in the vector space.

In my publication Making Monolingual Sentence Embeddings Multilingual using Knowledge Distillation I describe an easy approach to extend sentence embeddings to further languages.

Chien Vu also wrote a nice blog article on this technique: A complete guide to transfer learning from English to other Languages using Sentence Embeddings BERT Models

Characteristics of Sentence Transformer (a.k.a bi-encoder) models:

Calculates a fixed-size vector representation (embedding) given texts, images, audio, or video.

Embedding calculation is often efficient, embedding similarity calculation is very fast.

Applicable for a wide range of tasks, such as semantic textual similarity, semantic search, clustering, classification, paraphrase mining, and more.

Often used as a first step in a two-step retrieval process, where a Cross-Encoder (a.k.a. reranker) model is used to re-rank the top-k results from the bi-encoder.

The two families of similarity

The first family is lexical similarity. This is where local libraries make light work of the problem. If I want to match names, organisations, cities, account references, or noisy OCR output, then normalisation, entity extraction, fuzzy matching, and character n-grams are often fast, cheap, and surprisingly effective.

The second family is semantic similarity. This is where embeddings start to matter. If two phrases are conceptually related but use different wording, fuzzy matching often falls apart. Embeddings give you a vector representation that lets you compare meaning rather than spelling.

Die Patientinnenanwältin der Steiermark stellt klar: „Solche Tools dürfen nicht versorgungs-ersetzend sein“.

Wer ist der Landeshauptmann von Vorarlberg?

Ich konnte keine Informationen zur aktuellen Landeshauptfrau oder zum aktuellen
Landeshauptmann von Vorarlberg finden. Bitte prüfen Sie die aktuellen Informationen direkt auf
der Website der Vorarlberger Landesregierung oder wenden Sie sich dort für aktuelle
Details.

Die mögliche Errichtung eines Rechenzentrums des Online-Giganten Google in Nickelsdorf (Bezirk Neusiedl am See) hat am Freitag die ÖVP auf den Plan gerufen. Die Volkspartei sieht viele offene Fragen und fordert Antworten von der Landesregierung. Landeshauptmann Hans Peter Doskozil (SPÖ) erteilte dem Projekt unterdessen bei einem zu hohen Wasserverbrauch eine Absage.

Everyone has a different EMF, and arguably a very different EMF. This makes it almost impossible to describe to anyone who hasn't been. This time there were lasers and LEDs and robotic things, and also embroidery and a plant swap and a planetarium. I didn't go to any scheduled talks or workshops and I took hardly any photos. I chatted to lots of very diverse people, went to and organised some meetups, danced, wrangled with (and a few times solved) puzzles, volunteered, and wandered around admiring the art.

The concern driving all this is the US CLOUD Act. It lets American authorities request data held by US firms, even in overseas data centres. Last year a Microsoft executive told a French court, under oath, that the company could not guarantee data sovereignty.

Die Logik des Gesetzes wird sein, süchtig machende Mechanismen“ für unter 14-Jährige zu verbieten, sagte Pröll der „Presse“. Was die Plattformen betrifft, für die es zukünftig eine Altersgrenze gelten soll, werde es „keine taxative Aufzählung der Plattformen geben“.

Everyone is an artist.
Lots of people consider themselves to be "hackers" or "makers", but wouldn't call themselves "artists". One of our goals should be to show people that they are. We can do this by creating a level playing field between more established artists and people just bringing a little thing that they've made. We want people to be proud of their work and see how valued it is.

Being encouraging:

• lots of people are just starting out making things and need encouragement

Even if you've done EMF multiple times before and are an experienced installation artist, making something new is always hard. It's easy to get sucked into negative thoughts about your work and sometimes someone just saying "yes! that sounds amazing!" is what you need to find the motivation to complete it.

Equally, sometimes a project just doesn't work out how you'd hoped, no matter how hard you work on it. Having someone say, "hey! that's cool, don't worry about it" can restore your confidence.

inside-voice is a terminal app that helps you keep your speaking volume down while you are on Zoom, Google Meet, or other calls. It watches your microphone level and plays a short local chime when your voice crosses above an adjustable dB threshold, or after it stays above the threshold for a configurable duration.

Google hat für sein Rechenzentrumsprojekt in Kronstorf (Bezirk Linz-Land) weitere Ausbaustufen zur Genehmigung eingereicht. Nach Angaben von Wirtschaftslandesrat Markus Achleitner (ÖVP) ist der Antrag deutlich größer als die bisher genehmigte erste Baustufe und umfasst nun den gesamten Campus.

Much of AI "productivity gains" come from managers implicitly trusting an LLM to do a good job, and suspending the onerous micromanagement that humans employees get subjected to.

You can do that without paying Anthropic a single penny. It just requires letting the experts you hired do their jobs.

Falls sich jemand fur 1450.at und ein paar ähnliche Websites verantwortlich fühlen sollte …

Das TLS-Zertifikat ist heute abgelaufen.

Autowaschen, Rasen sprengen und Pool auffüllen verboten: Das heißt es ab Freitag in drei Gemeinden im Bezirk Grieskirchen. Wer sich nicht daran hält, muss mit einer Geldstrafe rechnen. Die Maßnahme wurde nötig, da Aufrufe, mit Wasser sparsam umzugehen, nicht erfolgreich waren.

Die Grünen Oberösterreich fordern für das geplante Google-Rechenzentrum in Kronstorf eine verbindliche Strategie zur Nutzung der Abwärme. Landesrat Markus Achleitner (ÖVP) müsse sicherstellen, dass die Region von Abwärme und Wertschöpfung profitiert.

Über 70 Prozent der Steirerinnen und Steirer kaufen laut KMU Forschung Austria online ein. Experten warnen vor Fake-Shops, die zur Fußball-WM mit günstigen Trikots und Fanartikeln täuschen. Besonders aus Nicht-EU-Ländern können Lieferungen problematisch und Produkte mangelhaft sein.

I believe that vibe coding — irrespective of whether it’s useful for enterprises, which I doubt — is being marketed towards consumers in a deeply unethical way. One that’s worryingly reminiscent of multi-level marketing schemes like Herbalife and Amway, or the crypto grifts of the 2010s.

I believe that Replit’s decision to target younger people at a time when they’re struggling to find work, or are convinced that the future workplace has no use for them, is deeply predatory.

Any creator that promotes Replit without being transparent about the likelihood of building a million-dollar app, or about the costs of building software with AI, is either willingly complicit in a cynical, harmful scam, or otherwise promoting a technology that they themselves do not understand.