Видео

Great pokemon catch!

hey man im getting not implemented error

All services of Hugging Face are defined in an easy way

But. Itt on cpu it doesn't not have any gpu option

Humans are -AI is- ruining the Internet 😭

Good one guys 👌

thank you

Damn, huge signal in Indian tech scene. 🙌 1littlecoder you are truly something else.

Thanks a lot. It's useful to know the profesional path

Thank you for this interview. I totally agree that to pick up skills, it definitely helps to have an interest.

I really love your videos. They're refreshing, real and down to earth. They seem to be the exception to the rule though. I mean.. the internet has become far more useless but it didn't need ai for that. That happened a few years ago. Facebook absorbing all the data and centralizing everything into a single place (that spies on you), instagram body dysmorphia in teens, the death of the regular web, google search becoming useless, most of youtube going from random silly vids to sanitized advertiser friendly slop, forums getting either shut down or selling out, the rampant spread of scams... I'm not sure there's much about the internet the way it is that i'll miss that much. I don't see AI as the threat to a useful, interesting internet.. i see corporate greed as the threat. Hopefully if it does get destroyed, it'll get replaced with something a little better. It used to be so so much better than what it is now

great to bring him here, he is an inspiration ,,know him from navrasa

Amazing

The problem is that training is so hard. One could train an AI just to create one paper or answer one hard question. We need more compute.

It's soo true, whenever I try to find a solution to a problem it just tries to correct my inefficient approach. So I do tell him don't follow my path if it's inefficient and there is a better way available, it works sometimes, sometimes not.

Amazing tutorial my friend! Was looking for this. It would have been more helpful if you could explain how to deploy LLMs created using vLLM inference engine.

It is gonna change us for sure, exactly like internet changed us and social media changed us. for innovation we need more time, to connect dots, leverage on your humanity and connect the dots on the jobs market as long as you can :)

Thank you so much for wonderful sessions!!!!

I'm currently in about month 6 of working in the AI space exclusively, currently delving into the top two of the pyramid. I must say that this is a *great* all-up introduction for anyone who is curious about this space...not a minute wasted!

It's true that ChatGPT or LLMs make me lazy and dumb. I used to use Stack Overflow for problem solving, but after ChatGPT, I just copy and paste the corrected code.

True

Google colab was paid now

I can't wait til these open source models become as powerful as Claude 3.5. I don't know how long it will take - possibly multiple years, but it's inevitable. Imagine being able to locally feed data from websites into a local model that powerful. We're living in the future!

can you compare it with pyscript?

Thank you four your videos! Definitely not making the internet worse.

Ai took away my productivity

Great topic.

The only way AI Tools can give you correct answers (not just grammatically) is if those tools understand. They don't, never did and never will. Soon the hype will die and people will see these tools for what they are. Now I leave it to you to think what happens when future AI models are trained on the crap produced by current models....(hint: gigo)

Next step is to have AIs summarize crap AI articles and to find non-AI generated content for you. It will be an interesting arms race to witness. We are steadily heading towards highly customized internet experience where everyone has entirely different set of information visible to them based on what kind of software they have in place.

trollers , spammers are pushing content than genuine content. it's the models food if it gets unhealthy food it's health will degrade

People have no common sense anymore. GPT isn't a substitute for critical thinking... We live in a world where someone rather eat glue and sue the creators of the model than listen to any common sense they might have had. The future is going to be bullshit. You wont be allowed to do anything.

The internet i grew up with died when twitter was invented. Long time coming.

I think the demonstrated behavior can be fixed in part with a persona in system prompt and agentic workflows. Of course if the alignment is strong, it still won't really work. Also the datasets may lead to a bias, especially if the data has been selected for being "clean", safe for work or to follow recent western "ideas" of how the world works - to say it in a diplomatic way. In the end it's hard to understand light if you don't know darkness. That's also why I totally hate that certain big brains in the AI scene feel responsible to determine what others can get out of AI systems and what not, all in the name of the good and "safety" of course. So all in all I'd not agree with that LLMs only rehash common beliefs. Even if the public heavy weights like ChatGPT und Claude may do this, it doesn't mean it's a per se a design error. The Internet getting filled with more and more synthetic data though might actually become a serious issue since that may unfold an own dynamic.

good reflection, thanks

Thank you so much! I was kinda lost on navigating HuggingFace, but this video helped me understand HuggingFace more.

AI (= advanced tamagotchi) isn't doing anything on it's own. The potential ONLY unfolds when combined with HS (= human stupidity).

They are never meant for doing breakthrough research. You have to rely on your brains!

Couldn't we ease the problem with a system prompt that encourages to leave the beaten paths, like "think outside the box"?

The Internet was ruined the very second that it became centralized by monopolies who are in bed with government agencies connected at the hip through intelligent backed venture-capital

Now, that Fireship channel showed how Claude is better than most LLM models, how can i add it into my Brave?

I agree

SSI is designed for super stupid investors.

Useless feature, after selection it just give you the facility to send the selection to ai...I think if I can select the text and also copy and paste it to chatgpt to get the better response

You are wrong! People are making internet as it is. What about the scammers, fake gurus and fake new BEFORE THE AI??? Who posts and likes the garbage videos - people. Now AI is doing what people are doing and you blame AI - please ... be quiet. You are not entitled to speak on the topic! COME BACK TO REALITY!!!

Good job.