svaha1728

7 ms·

31.
by svaha1728 2y ago
If you have the source code for TikTok this can be done. GenAI is a nice cover for theft.
32.
by svaha1728 2y ago
They were bought by a private equity company in 2021. Would love to know more about the Berlin programming scene in the 90s. The initial minds behind Generator/Reaktor were incredibly inspiring.
33.
by svaha1728 2y ago
Nothing wrong with the senior engineer level. I’ve had the experience of managing 75 developers and I enjoy the fact that right now I’m “just” a Senior Software Engineer.
34.
by svaha1728 2y ago
No disrespect to the other developers on the team, but James Newton-King is no ordinary developer.
35.
by svaha1728 2y ago
It’s also because in real life I know musicians and artists who are struggling to survive.
36.
by svaha1728 2y ago
The next index fund should use AI. What could possibly go wrong?
37.
by svaha1728 2y ago
Ask the model to provide an image of its answers written on a chalkboard. Or, ask it to provide you an image of pathological eye conditions and look at the text labels.
38.
by svaha1728 3y ago
And the gems you are trying to excavate might not even be in the training set.
39.
by svaha1728 3y ago
Yup. Once we have every possible prompt categorized we will achieve AGI. /s
40.
by svaha1728 3y ago
for the low price of 7 trillion dollars you too can order from Grubhub and have it delivered to your door without the need for a single human! /s
41.
by svaha1728 3y ago
These types of arbitrary rules are why projects end up depending on deprecated packages. Nobody can do it so everyone whistles around a slowly smoldering dumpster fire.
42.
by svaha1728 3y ago
Those in power seizing control of a product once it’s created is a timeless story. See the Prague clock tower or Daedalus the Labyrinth builder who was imprisoned there by King Minos. Not a new story, or a particularly wise business decisio
43.
by svaha1728 3y ago
Fair number of apps out there running on a skeleton crew after a successful acquisition.
44.
by svaha1728 3y ago
The multi modal version will read your palm.
45.
by svaha1728 3y ago
They split the document. Here’s an example of Markdown splitting. All this is far more an art than science at this point. https://python.langchain.com/docs/modules/data_connection/do...
46.
by svaha1728 3y ago
The problem is the training set. If the training set contains something disruptive or of value, there is a nonzero percent chance it will be the next token. Boxes can attenuate that risk, but cannot set it to zero.
47.
by svaha1728 3y ago
A parallel universe where the internet is built with no code Windows Forms. That’s not exactly heaven.
48.
by svaha1728 3y ago
AI will only make this worse. Even my son, 12, assumes I’ll be out of a job soon because of what he heard on YouTube.
49.
by svaha1728 3y ago
Confidence is all you need.
50.
by svaha1728 3y ago
Management won. We are replaceable widgets now.
51.
by svaha1728 3y ago
Regex to the rescue!!!
52.
by svaha1728 3y ago
This type of AI is going to be far better at breaking things than it is at putting them back together.
53.
by svaha1728 3y ago
vector math in a 1536-dimensional space?
54.
by svaha1728 3y ago
I would read Francois Chollet's explanation of this. It's very good: https://fchollet.substack.com/p/how-i-think-about-llm-prompt... For words that are not in the model's vocabulary, like 'fluftable
55.
by svaha1728 3y ago
It’s crazy to think that if he would have just made a training set of JSTOR documents for a large language model he might have skirted around the copyright issues entirely.
56.
by svaha1728 3y ago
Most credible theory I have read today.
57.
by svaha1728 3y ago
He makes it rain H100s.
58.
by svaha1728 3y ago
Parrots may live 50 years. H100s probably won’t last half that long.
59.
by svaha1728 3y ago
I remember the Pylon days! I'd recommend Michael Kennedy's FastAPI course on MongoDB and Beanie. It's a quick introduction to how things can be done these days, and a great springboard: https://training.talkpython.
60.
by svaha1728 3y ago
I'm guessing it will dip for a couple years, then rise as we are called on to refactor the mess that was made.

More ›