Відео

That was great, thank you Harper! How awesome is it that you can do all this in such a straightforward way! Also how crazy was it that you meant Google Gemini and it did it as was needed. Such an amazing element of AI to point you in the right direction. Loving your content and followed your Github, thanks again!

Seems to save a step with hopping from code editor to LLM, to browser. Seems easier to navigate and not as cluttered compared to visual studio code with copilot. For some reason when I saw the title with Claude 3.5 Sonnet, I assumed it would have an artifact window included and got too excited, but that would be amazing to have in cursor.

Anthropic is surpassing OpenAI with Claude 3.5. Playing with Cursor is addictive. The AI space for developers is becoming increasingly competitive. The new Replit Agent launch also looks promising, though it’s currently limited to Vanilla JS for the frontend, Python/Flask for the backend, and Postgres for the database. The full-stack approach will unlock immense creativity. Choosing a framework like React/NextJS, Svelte/SvelteKit, connecting to a BaaS like Firebase or Supabase, installing packages, debugging, documenting… it gets tedious. If AI solves that, it’s a game-changer. Keep up the great work! Awesome video. One suggestion: for a paid community, check out Circle or Skool. Speaking as a former marketer myself, your credentials/expertise and your passion will attract many paying customers.

Nuclear power

Dibs 10 out of 10 would smash

Can Cursor use vision modalities? E.g., taking a snapshot of a napkin drawing of a UI and have the model use the drawing to generate the code?

Please fine tune mistral model on our own data

I would love to know about RAG from you :)

What is neurosymbolic AI

How are they regulated? If they were instructed a vague instruction, are they given the agency to do all that it takes (e.g. with Internet access) to achieve that?

What do you think about Project Sid? I find it fascinating.

This isn't about imposter syndrome. It's bc we don't really understand why AI and LLMs work. If he wrote code for traditional applications like TurboTax, the imposter syndrome would go away.

Thanks for putting it out, its something which I really need to hear today.

This is inspiring and can make so many kids excited about learning and make them successful in life. I feel it can also help people with disabilities to learn, adapting to the needs of e.g. dyslexia or ADHD. Looking forward to using it with my kids.

I love the idea of student-focused learning paths: some want breadth first, some want depth first, some need different orderings (graphics before calculus or vice versa), and this makes holistic/holographic or fractal-based paths possible… every lesson can be understood as worthy of knowing by itself, not as “just a prerequisite” that needs to be understood before the “next barrier.” There’s a real chance for LLMs to be good story tellers here!

Whisper whisper whisper tiktok voice.

This channel is underrated.

He had imposter syndrome because he was programming The end of humankind and that weighs on your conscience just a little bit

So basically when we train the model on different datasets,we update model parameters and we try to find parameters based on which model performs as expected for unseen data,is it so?

So basically model forms a mathematical a numerical representation of real -world object and when it sees unseen object of similar kind then it tries to identify the new object based on that mathematical representation?

What are parameters?

thanks for making the basics series, finding it very useful while doing the Andrew Ng courses!

Did you ever get to see Yann Lecun while working at Facebook?

What

Here a guy educated in Switzerland on a Pestalozzi school (like Montessori). I can tell, I'm weird 😅. I'm an artist, AI programmer and entrepreneur. Full mode into create.

Hi Harper, quick question. Is the series relevant for non-tech AI enthusiasts?

Oh look! Two more airheads shilling for the nuclear swindle! Nuclear is a crime against humanity! Nuclear produces toxic radioactive waste that you have to pay for to store, forever. That radioactive threat just increases every day and keeps growing larger and larger as time goes on which costs you more and more every day but it does make the nuclear swindlers rich you poor...

FinalSpark in Switzerland are using brain cell organoids instead of silicone chips. They claim it uses 1million times less energy. Their biocomputer server is being used to do machine learning - so maybe wetware computing is also a solution?

I would love to watch you guys. It’s something I curiosly want to know to about

So happy I stumbled upon this. Yay! Binging.....

Hi, thanks for the video! I’ve a couple questions: What if one did try to use dropout during inference and not just training? Would that be a good way to get multiple, varied samples from a model? (e.g., instead of just using temperature for sampling) In training, does the order of the various batches matter? For example, after a few iterations and the weights start to settle, if you had a choice, would you want the next training examples to be the ones with the highest gradient differences or would you want to sharpen things first by picking the examples with lower perplexity? I anticipate the answer is going to be “random is better”?

As it relates to this topic, it would be interesting to hear about AI Machine Learning being used to adjust the magnetic confinement around a tokamak to keep the plasma inside and the fusion reaction going and the AI outperforms humans at this task. Only read an article about it a while ago. Did anyone ever really try this or just in a sim? What if there's a little mistake or a big mistake? Has anyone bothered to improve this magnetic confinement ai? Who made it? When reinforcement learning AI figures out the optimal strat to get the max number of rewards/cookies for doing what its meant to do, does it stop trying to do a better job because it sees as diminishing return? is there anyway to have that not happen? I heard china just got a thorium nuclear power plant. in the race for ai it sounds like that's an important step, is it? whats america doing? "llama 3 design me an even more efficient thorium power plant for episode 1!". lol. cheers

Lets talk about AI that creates UNCANNY VALLEY LEVEL PROPAGANDA and how it effects VOTERS. Lets talk about requirement of a LICENSE and REGISTRATION To legally use Groq 2 so it cannot be ABUSED or those who DO and ARE RIGHT NOW abusing it can be held TO ACCOUNT!

That's cool to see you both ma'am 😃🤍

Oh my goodness i thought I found a great tool I supposebit is in a way but I'll stick to what I know

I'm a startup founder currently in the beginning stages of launching an AI business. I’m very interested in learning from those who have successfully navigated this path. Would you be open to an interview about your experience creating your AI business? If so, I’d like to ask you a few specific questions: 1. Should I prioritize hiring a developer or securing initial investment first (for payroll and computer supply)?? 2. What key metrics or milestones should I focus on to attract early-stage investors? 3. How can I leverage my existing network to find potential co-founders or advisors? 4. What is the core technical skills and experience I need in my first 2 or 3 hires? 5. What specific roles will be critical in the early stages of my startup. P.S. Please for the love of coffee and all that is “holy moly with guacamole”. Seeking practical insights for a developed concept, target market and growth projections, and ready to move forward. need more actionable advice. Thank you in advance for your time! Best regards, Michael

FAI

Wow that was so fast, I feel like we just unlocked a world of custom knowledge. Especially on ultra niche topics. Imagine a book about the history of your local government, a specific type of bird or forgotten historical figure!

Here's the link! www.neuronpedia.org/gemma-scope

Great! Can you do the next on Jupyter notebooks?

WOW! It is lightning fast. Do I understand correctly that the API key allows the GROQ servers to do the work? Am I right to assume you do not require a new API key for every book you want to generate?

good sources for learning about authorship are KM Weiland she has a few books that are really good, structuring your novel and creating character arcs especially, KM also has a youtube channel and news letter, its surprisingly not a lot of traffic for how good her books are. Abbie Emmons is an authorship youtuber who has a whole bunch of videos about writing. Brandon Sanderson, a very accomplished author has a creative writing workshop at a college that's on youtube in his playlist.

Imagine all the epsteins elon will hire

Ewww

I feel like he would make a cool kind of societal different kind of school and teaching. The only draw back is his transphobia, but kids probably don’t think too much about there gender at that age so it’s probably fine.

My only fear is how much is he going to charge parents.

Dear gods! I pity any child sent to this school.

Hey Harper! Thank you so much for this. Super Awesome! Do you happen to have any methode on generating nice ebook covers?

Could you share the name and link to access the tool?

Hardware is so difficult, congratulations!