fam in ~18 hours we'll have 157 million casts with embeddings, pray for me it doesn't crash overnight

lets get to the best version faster - follow @1dolinski · challenge your habits · talk with my resume https://getfirstresponder.com/talk/1dolinski

oh i'm gonna finish this i'm not letting a stupid computer win

no i didn't even know that was a thing, thank you!!!!!! if i can't get the FP16 working i'll switch to this

Have you considered int8 embeddings?

https://huggingface.co/blog/embedding-quantization

Im the guy that posted all those factory memes once

i will persevere if it's the last thing i do

4 instances each filled with 3.9 TB of data before they ran out, the instances have 4.2TB of HDD total

There's a secret side of YouTube, just beyond the guiding hand of the algorithm–and it’s nothing like what you know.

The vast majority of YouTube's estimated 14.8 billion videos have almost never been seen. Until now.

My AI is using this technique: https://t.co/F5kdYw677d

The scraper tried more than 18 trillion potential URLs before it collected a big enough sample for real scientific analysis.

Among the findings: the research estimates the median YouTube video has been watched just 41 times mostly by machines.

Link: https://t.co/cPTmMLKBGL https://t.co/uI4DordFlw

Brian Roemmele

18 T 

https://x.com/brianroemmele/status/1898647583045869949?s=46&t=jXbE9FtOedI8BQTeVCix-g

I Took A Video!

It’s all the videos in /itookavideo 😆

And I agree with @1dolinski - this is a grantable endeavor.

Your screams are taking up too much space on the internal drives.