While going all in in ML. Makes sense. https://news.ycombinator.com/item?id=40171125
Google who?!
Bard creating python code by itself 😂another stupid decision Gleaders
More like Indians, sounds likes the roles were outsourced rather than replaced by AI
Tensorflow doesn’t get the traction as it used to when it started , majority of langchain and all LLM / agent frameworks rely on PyTorch , cuda , or low level c …… also with the advent of writing code ( though it’s not ready for prime time ) it will soon be efficiency over cost
I just don’t understand why python became the default language for data science. It’s slow AF. Annoying indent nonsense. Not always clear what data types variables are. Issues could easily be caught at compile time in other languages. It’s just a dumb language.
Intuitive, easier to think and do math operations. I can’t imagine writing shape, reshape, dot product, matmul etc in goddamn Java
Easier for non cs geeks to understand. Lots of ds people didn’t study cs.
Python sucks AF. Has also the worst energy footprint. C/C++ should become the new standard for DS/ML
I fail to understand how this will benefit Google long term. For short term, I guess they save a few million in salaries?
It sounds like their internal python infra team was moved to India?
I don't think we do as much of python ML Python afaik is pretty rare
I wonder what Laurence Moroney advocates for at G? I’ve listened to many of his presentations and he always talks about TF, but still I would like to understand what is actually used in practice at Google