5-10 year timelines used to be a thing of the past. Companies are smarter and cost conscious now. And the tech is great and requires has an almost non existent learning curve.
@waitin4agi_
-
AI Job Displacement Could Happen Much Faster Than Predicted
By
–
People saying AI will take 15, 20 years to replace jobs: lol. We are in the era of the internet where information asymmetry is low and cost of experimentation is low. ChatGPT userbase grew magnitudes faster than anything ever before.
-
NERFs and Kaiber/Ebsynth Characterized as Reskinning Technology
By
–
NERFs + Kaiber/ebsynth should just be renamed reskinning tech.
-
PyTorch usage with fastai book for neural network theory
By
–
I mostly use pytorch too. The book comes with fastai. Using it more for NN theory.
-
Returning to Neural Networks: Catching Up After Years Away
By
–
Neural networks were my elective studying CS in college. It’s been a long time since then and lots has changed. Time to dive back in. pic.twitter.com/XNZEo2gbRi
— Varun Mayya (@waitin4agi_) 27 avril 2023Neural networks were my elective studying CS in college. It’s been a long time since then and lots has changed. Time to dive back in.
-
Bot Invasion: Forex Scams Meet AI on Instagram
By
–
20-30% of my Insta comments are bots selling the same exact forex scam. What happens when they learn to infuse a little GPT into this? Instagram's ToS allows bots btw.
-
Teaching Machine Learning Through Deepfaking Process
By
–
I have found that showing a student the entire process of deepfaking on deepfacelab the best way to teach the ground-up process of ML: from data collection to loss graphs to showing how much of it is still human.
-
Best Internet Resource on Transformers Technology
By
–
This is the best resource on transformers on the internet: https://
e2eml.school/transformers.h
tml
… -
Cold boot problem with custom model on Replicate
By
–
. @replicatehq I’m running a custom model on replicate but having the cold boot problem. Could you help me with this?
-
Transformers as Computers: English as Programming Language
By
–
Just incredible how transformers (when trained on lots of data) ended up being computers with English as the programming language. Don’t know if anyone could have predicted this.