Marius_Manola

you didn't get into yc? No problem, now you can just talk to pg. made this for fun

There's basically so much educational content out there, that you could argue there's no need for AI to "generate" lessons, when it could just restructure existing ones.

Some make it the case that you need "personalised education" for everything, but it's not at all clear that's the solution. For deeply technical things, you have to explain them from the ground up, and that can't be personalised in n ways...

I thought about making a tool that can search for videos based on context, and jump to the specific moment in the video that you needed. So for the past couple of days I've been working on notclass. We focus mostly on university lectures and long form podcasts, because most of the time valuable information is skipped due to the sheer length of such clips.

Try it and let me know what you think.

P.S. It is 100% free, no registration required, and I intend to keep it that way.

So, should we expect Perplexity to post that "OpenAI killed my startup" kind of article? (just saying)

but for some reason it didn't get that much traction... I know there are just 2 learning paths, I just finished the whole thing in 5 days... Currently the learning pathways are done manually, but the idea would be to embed lots of high quality videos so that they can be re-arranged into courses based on what the user needs.

I wanted to make this a B2C product for those who can't afford to buy an online course, but don't want to scroll through all the content that's on youtube...

As for the B2B idea, it might work, but this would be just another gpt wrapper.

I'm honestly more interested into doing something useful, by embedding a ton of youtube content, then creating 100% free learning paths with free resources.

I intend to keep notclass free forever. No hidden fees.

I'm an 18 years old high school student, and I just finished my first month since I started learning integrals. (Studying based on the European system, calculus is mandatory - which I love. We do mostly derivatives in 11th grade and integrals in 12th)

Seeing this video of Neil deGrasse talking about his experience with calculus pushes me to study it even harder, so I thought that if it motivates me, maybe it might motivate someone else here.

I so much loved watching the pervert's guide to ideology

Was it even a startup, if what I was offering was to generate synthetic data for fine-tuning AI models using various frameworks to understand the patterns out of 3 examples given by the user of input-output pairs, some questioning done by a chatbot and then use some fine-tuned gpt models and some big prompts to generate data at scale.

I feel like I knew openai was going to do this, but I decided to proceed with the project anyway... I am currently like 2/3 of the way to finishing the project, but I'll obviously drop it now... I feel like I was just fooled by the gold rush to creating million dollars gpt wrappers (what an idiot...), but I felt like what I did was a bit more than just a gpt wrapper, maybe I just fooled myself.

The fourth one hit really hard because we live in a world where making mistakes has been normalised to an extreme extent, where we stop being aware that we even made a mistake. And I feel that's the case with me (at least that's what my tennis coach tells me, and I kind of see that in the rest of my life)

Fortunately I don't have a problem with not making mistakes, I made a great deal of them (and many more to come) as I have been trying various online ventures for 7 years now (from youtube and freelancing to startups). And in the end some of those ended up working so well I make more than most of my school teachers do per month.

Finally, I wish I knew how not to care about what others think of me. I'm trying to find a way not to worry about it, but I'm still trying to figure it out. I think this will go away with age, as it is with most bad experiences we have in life; the cure to those experiences is to relive them again and again and realise it's not as bad as we make it up in our own heads.

Thank you for your response. The gift of your keystrokes is much appreciated.

thank you. Will be improving the model in the upcoming days and weeks maybe. can t promise tho.

I did it for fun mostly

but do you have the question?

Also you could achieve that with RAG quite easily. training is good mainly for the tone and style (which still isn't perfect) but better than most things out there that simply use the llm with no fine tuning

It was. But the dataset for training is rather small and focuses only on the first book.

I might improve it in the future

thank you for your kind words. Will be making it even better (making the fine tuning dataset larger and more similar to how the guide would answer) if there'll be enough people wanting that.

but trained on synthetic data with elements from the first book