In 1996/7 I had a chance to use Tcl/Tk to build one of the first stock tickers on the web called DigitalTrader [1], after that we used it to build some of the first vector embeddings in 2005 for early biological language models at Lawrence Berkeley National Lab [2,3] for space biosciences. Still a fan.

[1] https://www.orafaq.com/usenet/comp.databases.oracle.misc/199...

[2] https://newscenter.lbl.gov/2005/03/31/a-search-engine-that-t...

[3] https://patents.google.com/patent/US7987191B2/en

Agreed. Vector embeddings along with which distance calculations you choose.

Tend to avoid Euclidean distance.

When the vectors are normalized to unit length cosine similarity and Euclidean distance are equivalent.

This an optimization that many vector dbs use in retrieval since it is typically much faster to compute Euclidean distance rather than cosine.

How can it be if we have yet to fully define what human intellect is or how it works? Not to mention consciousness. Machine intelligence will always be different than human intelligence.

Ahh yes, the lost art of UNiX sys admin always comes back.

This is also where MoE shines with a mixture of small and large language models.

Unique specialized high-value hard-to-duplicate data and datasets will be frontline moats and provide new competitive edges.

And, at the heart of AlphaFold2 is the language model, the tip of the spear in AI today. 'Language' can come in many forms e.g. a protein or amino acid sequence.

AlphaFold 2 wasn't Q-learning based. It was supervised SGD and the "evoformer" they introduced is very close to a transformer. So it's not exactly an LLM, but it's a pretty close equivalent for protein data.

Yep, Scott Kelly, after a year in space, lost about 30% of his hearts muscle mass. Ref: https://www.nasa.gov/humans-in-space/nasas-twins-study-resul...

He could have gotten basically zero exercise, even against gravity as a baseline, which is better for heart muscle mass

This is just a pitch for your company hamfisted into unrelated content.

They are named 'feature' vectors with scored attributes, similar to associative arrays.Just ask MI. Jordan, D. Blie, S. Mian or A. Ng.

They are embedded into a particular semantic vector space that is learned based on a model. Another feature vector could be hand rolled based on feature engineering, tidf ngrams etc. Embedding is typically distinct from feature engineering that is manual.