Filip Graliński

Day 2 of #SnowflakeSummit flew by but not before a mountain of announcements from our Platform Keynote! We announced: Adaptive Compute, Snowflake Openflow, Cortex AISQL, Semantic Model Sharing, Snowflake Intelligence, and much more. See what's new: bit.ly/4mNjiqR

@jxmnop Yeah, great paper, also for educational purposes. I'm curious whether the embeddings are still available... Whether king + woman - man = queen was already there but they didn't realize it...

did you know people have been training neural networks on text since 2003? everyone talks about Attention Is All You Need. but this is the real paper that got our field started. it was in 2003, in montreal. i read it, and it was even more forward-thinking than i expected:

@spacemanidol Todo lenguaje es un alfabeto de símbolos cuyo ejercicio presupone un pasado que los interlocutores comparten...

Kinda crazy how much random literature applies to the LLM world now. Borges would probably have a field day with LLMs and the constant talk of the incoming singularity. Truly the creation of library of babel.

How can the most accurate SQL be generated for a given question? We propose a method to significantly boost text-to-SQL accuracy while drastically cutting costs.👇 #NLProc #AI #TextToSQL #LLMs

Our Snowflake AI Research team just released Arctic Embed’s core training code into the open source ArcticTraining project — making it easier for developers and researchers to reproduce, fine-tune, and build on our embedding models. Arctic Embed is the leading small embedding model on the MTEB leaderboard and is widely used with over 1M monthly downloads. What you’ll find: ✅ Clean, config-driven workflows powered by DeepSpeed ✅ Flexible contrastive data handling ✅ Example fine-tuning recipes and ready-to-use tooling Read more here and try it out: snowflake.com/en/engineering… @SnowflakeDB @DeepSpeedAI @lukemerrick_ @pxyumass @spacemanidol @rajhans_samdani @jeffra45 @StasBekman

@eigenrobot Therefore aliens

i mean

Connor Shorten was kind enough to give me the mic for a lot of hot takes on text embedding models in the latest Weaviate podcast.

@_chenson__ @gro_tsen I remember we used them in high school (Poland, 1990s).

@gro_tsen Was this a standard notation in the 50s? Seems pretty confusing! (From the first reference)

Learned on MathOverflow: it is possible to write a finite formula for n! involving just the operations of addition, subtraction, multiplication, integer division, and exponentiation. Precise statement is here: mathoverflow.net/a/484115/17064

@magoniareview Of interest! Actually, it'd be great if it covered languages other than English... (reading weird books is the best way to learn foreign languages!), I can share a list of recent "Magonian" books in a bunch of languages, if you're interested.

I have now suspended the Pelican's 'Book News' blog, but may resurrect is as a stand-alone listing of forthcoming titles of Fortean, ufological and folkloric interest, if enough people think it will be of interest.

We are excited to share SwiftKV, our recent work at @SnowflakeDB AI Research! SwiftKV reduces the pre-fill compute for enterprise LLM inference by up to 2x, resulting in higher serving throughput for input-heavy workloads. 🧵

Obviously Germans had the reputation for copying things (just a bit earlier). Which is a standard, textbook pattern for a rising manufacturing power. Produce cheap, crappy, low quality goods. Copy everything that moves, and doesn’t move. Break every patent and every copyright.

🚀 I am thrilled to introduce @SnowflakeDB 's Arctic Embed 2.0 embedding models! 2.0 offers high-quality multilingual performance with all the greatness of our prior embedding models (MRL, Apache-2 license, great English retrieval, inference efficiency) snowflake.com/engineering-bl…🌍

Więcej informacji: linkedin.com/posts/bartosz-…

Can AI models help us create better models? 🧵 1/ It's a question that stands at the boundaries of what's possible in data science. We explored how Large Language Models (LLMs) perform as data scientists, especially in the art of feature engineering.

@jaroslavrudis Gratulacje!

Díky / Danke

@hrishioa @Hrishioa clarify what "clarification question" token is in the guide; I was, like, what is that magic token?? (but anyway, cool doc, thanks)

Entropix is beyond cool but hard to understand Here's a guide that means you don't have to walk my ragged path: southbridge-research.notion.site/Entropixplaine… Spent an hour talking to imaginary AI friends about the code, then ran it through Lumentis

@hrishioa I want to know... what the token 2564 is... ... I'd prefer 1981, for esthetic reasons