BeNN
1.11K subscribers
915 photos
82 videos
115 files
153 links
From simple ML algorithms to Neural Networks and Transformers β€” and from Number Theory to Topology, Cosmology to QED β€” dive into the world where code meets the cosmos.πŸ‘¨β€πŸ’»πŸŒŒ

For Any Questions @benasphy
Download Telegram
Someone asked me what if I fail and said it all will be for nothing?
lol It looks scary when you fail after doing big project for six month, but here is the funniest thing...If I fail I'm getting back to 9-5 Job as usual and there is nothing i lost besides I still gotta a ton of projects I'm gonna build😁! At least I have tried and gave all I have to finish what I started and most of all I'm man of my word and like in Karamazov I won't betray myself! Most of all any developer out there you should know Tech products will soar up later not around first time, even most of big tech companies started by losing money for few years and the most important is its future.

Have a nice time may the force be with us!
❀14
BeNN
I know Those big companies aren't gonna achieve AGI by 2030 as they bluff, but I can't prove it.
I have changed my mind on this..... I could say they won't achieve AGI by 2035 unless they find robust system than LLM
πŸ‘5❀1🀬1
Finally It is pleasure to see the legends are realistic!
❀3
I know that today's AWS outage is coz of vibecoding, but I just can't prove it.
🀣15
I should've started C long time ago! It's SUPERIORπŸ”₯
πŸ”₯5⚑2🌚1
BeNN
I should've started C long time ago! It's SUPERIORπŸ”₯
The funniest part is I found C more easier to grasp and straight forward than any other language... C is so GOATED
πŸ’―10❀‍πŸ”₯1
Forwarded from EdueraSkills
❀4
πŸ‘€Interesting! Try it out and let us see in comment 😁
😁7
Forwarded from EdueraSkills
πŸŽ‰ Big Update from EdueraSkills (EdueraSkills.com)! – Up to 50% OFF πŸš€

After a lot of students DM’d us saying β€œthe price is a bit high for us”, we listened. ❀️
Because our true goal has always been to enlighten as many students and people as possible, we’ve decided to take the risk and make our courses even more accessible!

πŸ’₯ For a limited time, we’re offering up to 50% OFF across all our courses πŸ‘‡

πŸ”°Python Programming – from 800 birr to 500 birr
πŸ”°Machine Learning – from 1000 birr to 550 birr
πŸ”° Python Libraries (NumPy, Pandas, Matplotlib, Seaborn) – from 800 birr to 400 birr
πŸ”° Introduction to Computer and Technology – from 1000 birr to 420 birr

🎯 Learn real, job-ready skills at a fair price - made for you.
🌐 Visit EdueraSkills.com and grab this offer while it lasts!


#EdueraSkills #SkillUp #Python #MachineLearning #DataScience #Technology #Ethiopia #OnlineLearning
❀9
EdueraSkills
πŸŽ‰ Big Update from EdueraSkills (EdueraSkills.com)! – Up to 50% OFF πŸš€ After a lot of students DM’d us saying β€œthe price is a bit high for us”, we listened. ❀️ Because our true goal has always been to enlighten as many students and people as possible, we’ve…
I have heard many people saying it's hard for student to pay 1k and lower it.... Here's it then and actually the profit will be super lower and may even be negative 😭 15% tax, payment fee, cloud billing most of all!
Enjoy it.
πŸ‘3
GOT Timeline
πŸ‘€4❀1
Which method is better for finding parameters in Linear Regression?
Anonymous Quiz
55%
Gradient Descent
26%
Normal Equation
19%
Moore–Penrose pseudoinverse
BeNN
Which method is better for finding parameters in Linear Regression?
Most people said GD, but Gradient Descent is like trial and error: it keeps adjusting the parameters step-by-step until it finds the best fit. It works great for very large datasets, but it takes time, needs a good learning rate, and might not always land exactly on the true optimal solution.

The Normal Equation, on the other hand, directly gives you the best parameters in one formula β€” but only if your data matrix behaves nicely (that is, if X^TX can be inverted). When it can’t, the Normal Equation simply breaks down.

That’s where the Moore–Penrose Pseudoinverse shines. It’s like an upgraded version of the Normal Equation that never fails β€” even if your data has redundant features, missing rank, or weird shapes. Instead of forcing an inverse, it uses Singular Value Decomposition (SVD), which is a little bit sophisticated math in Linear Algebra😁, to find the most stable and mathematically perfect least-squares solution possible. In simple terms, it always works, it’s more accurate, and it’s mathematically elegant.
πŸ‘6πŸ‘1
BeNN
Most people said GD, but Gradient Descent is like trial and error: it keeps adjusting the parameters step-by-step until it finds the best fit. It works great for very large datasets, but it takes time, needs a good learning rate, and might not always land…
Note: This is for LInear Regression alone not for all of algorithms. By the way, when you call model.fit() in scikit-learn it uses PseudoInverse under the hood to find optimal parameters and that is why you don't have to specify like lr, n_iters and etc.
πŸ‘5
All my Linkedin is filled with feed of people graduating from INSA and I just remembered sth. I'm not sure if it's good idea to talk about it, but since they ain't CIA i'll share ya one story..... The thing is I and My friend had opportunity to present our work when we were at Grade 12 in one Inauguration ceremony. I built chatbot with local language + English and I had opportunity to present to Prime Minister and he was amazed and the best thing was he truly understands AI and it was simple to explain.

The thing is at that moment The INSA director was telling to one guy to contact me and after he took assignment from her he comes on and asked me to build sth I'm not gonna talk here😁(got sound Spy movie), which is really scary and NSA shit honestly and offered me a job to work there as AI Engineer at grade 12 starting and I asked him some question and his response was really scary and The surveillance they do on people is pretty scary!! I surely declined the job and I remembered that I declined job offer from them 🀣I sometimes wonder the outcome if I have accepted it.
😁5😱4❀1πŸ‘1πŸ‘€1
with all due respect please If your knowledge is limited to GenAI or any of AI subfields don't call yourself AI Engineer🀨It is really Harassment for those who spent more than decades and master a lot of subfields and then could call themselves AI Engineer.
❀4πŸ‘1
This media is not supported in your browser
VIEW IN TELEGRAM
When C cooks u😭😭
😁8
Forwarded from BeNN
"Mathematics is the language with which God has written the universe." Galileo Galilei
❀3
BeNN
"Mathematics is the language with which God has written the universe." Galileo Galilei
Math was one of the main reason I changed my focus from web dev and app dev to AI. All the magic thing AI is doing today is basically a math! It is just ELEGANT
πŸ‘6πŸ”₯5πŸ’―3