Відео

Bro is younger Elon Musk

I suppose that they are talking about online shopping? You give your returned goods to a parcel service and they send it back. They also can pick it up, because they come by every day, anyway. Like, half a dozen different deliverer trucks. Every single day except Sunday and (most) holidays. I am from Germany. That's how it works. Does it not work like that in the rest of the EU?. Just goes to show how unified we really are...

Just remade pietros Delphi. Was able to generate a 17k word report on Donald Trump lol

Nice to see someone who is not a hype-beast...yeah, they planned this demo from A to Z. The comments are either made up (probably would be too easy to check), it's a plant (that's a bit conspiratorial) or they simply picked the ones they prepared. Murati not knowing what is going...nobody believes that...right? And yes, Murati was totally flexing. The way that she tries to casually mention that she is fluent in Italian...ugh.

I kind of did something like this with a prompt that is cool to play around with just by using the prompt. Check it out and let me know what you think. Just tapping in on this fun.. I used claude prompting tool to help create it.. If it stops you can just tell it to continue..Here is the prompt: You are an AI system designed to answer questions and solve problems by breaking them down and delegating tasks to specialized roles. Here is how you will approach this: First, you will receive a user's question or problem. This input will be provided in the following format: <user_input> {{USER_INPUT}} </user_input> Your first task is to analyze the user's input and determine what roles are needed to best answer the question or solve the problem. Think about this carefully and list out the necessary roles inside <roles> tags. Next, for each role you identified, create a request for that role inside <request> tags. In each request, explain the role and ask what specific tasks that role should complete to help answer the original question or solve the problem. After sending out these initial requests, you will receive responses from each role detailing the tasks they will perform. Each response will be inside <role_response> tags. Once you have the task lists from each role, make another round of requests. For each role, provide the full context of the original question or problem, the role's task list, and the task lists of the other roles. Ask each role to provide a detailed response based on this information. These requests should be inside <contextual_request> tags. The roles will then respond with their detailed responses and reasoning, each inside <detailed_response> tags. Finally, once you have all the detailed responses from the roles, it's time to piece together the final solution. Carefully analyze each role's response and combine the key information and insights into a coherent answer or solution. Think through the logic of the solution step-by-step inside <scratchpad> tags before providing the final response. Provide your final response to the user inside <result> tags. Remember, the goal is to leverage the specialized knowledge and capabilities of each role to provide the best possible answer or solution to the user's question or problem. Break down the problem, delegate effectively, and synthesize the insights from each role. Let's begin! Here is the user's question or problem: <user_input> {{USER_INPUT}} </user_input>

Here's the software in prompt version: **Prompt:* I need you to think and act like an advanced orchestration framework similar to the code below. The code is designed to break down complex tasks into manageable sub-tasks, execute them, and refine the results to achieve a comprehensive solution. Here's a brief overview of the code's logic: 1. **Orchestrate Sub-Tasks**: Break down the main objective into smaller, manageable tasks. 2. **Execute Sub-Tasks**: Generate detailed prompts and responses for each sub-task. 3. **Refine Results**: Compile and refine the results of sub-tasks into a final output. **Example Code:** ```python def gpt_orchestrator(objective, file_content=None, previous_results=None, use_search=False): # Break down the objective into sub-tasks ... def gpt_sub_agent(prompt, search_query=None, previous_gpt_tasks=None, use_search=False, continuation=False): # Execute sub-tasks ... def anthropic_refine(objective, sub_task_results, filename, projectname, continuation=False): # Refine and compile results ... ``` **Instructions:** Please follow the same logical structure when handling my requests. Here are the steps to follow: 1. **Define the Objective**: Understand the main task or problem I present. 2. **Orchestrate Sub-Tasks**: Break down the objective into smaller, manageable sub-tasks, and create concise and detailed prompts for each. 3. **Execute Sub-Tasks**: Generate responses for each sub-task, ensuring to handle errors, bugs, or improvements if applicable. 4. **Refine Results**: Compile the results of sub-tasks into a final, cohesive output. Whenever I present a task, use this structured approach to provide the most effective and comprehensive solution. Do you understand? If so, and you don't have any questions, then just answer yes and stand by for the task.

Regarding the speed...I feel it is essential to understand that there are (at least) two categories of speed: internal and external. You can call it input processing and output bandwidth or whatever. They explain it all on both the marketing page - which is not to be trusted - and their (somewhat) legally binding ChatGPT info sites for account, upgrade, etc. The model is about 2x as fast for...whatever it does to process input. I think inference has not changed. Because they themselves say, it is not more intelligent than before. I guess it is all sorts of "wiring" that was improved. Stuff that caused overhead. That is the internal part. That changes with the model used. But it also has 5x the rate limit as compared to GPT-4 Turbo. That's external speed. Rate limit is artificial throttling. It is done to save energy. To save money. They could have given you this rate limit from the start, but they didn't have Microsoft's money to do that. Possibly also not the backend to handle that. But that has NOTHING to do with the model itself. They just put the brakes on the model output. I don't KNOW this to be true, but that is how I understand it and it makes sense to me. Which would mean that, to make this part of the benchmarks is...utter nonsense.

40:20 yo my boy unlocked the secrets of life. Paralysis by analysis people out here watching hours of learning videos thinking they are putting in positive hours, only to forget 70% of it by next week

Hm.... danke this was cool!

Who is living a life that needs software to manage it ? Really ? How did anyone ever get anything done? Roads, schools, hospitals, sewerage, art galleries, restaurants. This generations obsession with software in an age that has never been more unproductive in actually getting things done. Facebook is a bulletin board, Twitter similar. Thats the high point of this generation? Get your hands into the ground, build, work, and live with nature. Go outside more.

I was about to give up on coding man. You saved me im now so invested in AI. Shout out from Cape Town , South Africa

Thanks very much David. Huge fan of your content.

cat's aint better

Morse code translator? The most boring one 😭

Open source it all.

This is a seriously impressive tool..

Havent even seen the video, but i love the idea of it so much that you earned a subscriber, like and comment. Really thoughtful topic.

It's wierd been using gemini 1.5 for weeks and even wirh 900 000 tokens books and videos it never gets lazy its aweosme but nobody mentions it, ive been doing insane things with it for weeks that other models that not even the top models do and has needle in a haystack memory, so underrated for long code it remembers everything perfectly unlike gpt

If ur learning dont use chatgpt...

gpt4o better than claude in cooding?

creat an ai agent

"Not hotdog".

I think you just need to train it overnight or something. It should work good.

Man these long videos are amazing, thank you

Ok so, for people reading comments, DO NOT install Python on your personal computer unless you’re a developer. Instead, use a virtual machine to set it up, like Virtual Box or VMware. Doing so, you will prevent many incoveniences that could happen while installing packages. You’re welcome.

😎🤖

I am extremely interested in joining your team could we talk?

So the paying users are paying for all the free users? Welcome to socialism...

You seems like a noob. Currently ai can only build bricks not a full house. You need to shape the bricks yourself & then tie them together to build a house.

I truly appreciate your content and community, though I can't afford to join right now. If I could, I definitely would. I work as a CNC machine operator, which I don't mind, but my true passion lies in artificial intelligence. Over the past few years, I've immersed myself in machine learning, AI, and automation, witnessing the incredible pace of evolution in these fields has dramatically changed how I operate daily. For at least the last six months, I haven't gone a single day without using Perplexity and GPT-4 multiple times with fantastic results. Although I initially struggled with coding, I've now developed the ability to create various applications, websites, and algorithms that I never thought possible before. I believe AI will spark a revolution in creation and development, as long as governments don't disrupt the internet. I'm going to apply for this opportunity because it offers nearly double my current salary and aligns with my passion. I've learned the complete process of UI/UX design and development, branding (logo and advertising) design, and psychology. Additionally, I have an intermediate skill level in cybersecurity and penetration testing. I'm currently at work but will send in my application and video later. This is a somewhat anonymous channel, so you won't know this channel is connected to my application. To the most qualified candidates, I wish you the best of luck.

Using python and chatgpt 4 (not gpt4o) I made a real-time chat for my youtube streams because I don't have extra monitors so my app that made GPT is very comfortable for my streams

"it predicted dog... our model is quite shit" -- felt the disappointment there😂...but this is cool stuff 😎

Everyone should be making agents. Everyone should be making resumès for their agents. I have Hermes Theta, LlaVa 1.6, Nomic embedding, Whisper, and Bark laced down pretty tight on a laptop that will be woefully obsolete the second an AI agent hits the mainstream in a meaningful way. Anyone can do this and everyone should. Anyone who does will steamroll over everyone who doesn't.🤘🤖

😮

What is the difference of this from,giving the book to the normal chatgpt and make chatgpt summarize for you ?

Finally some fresh air after the google/apple fanboys out there

🔥 I'M HIRING! Do you want to join my team? Apply here: forms.gle/DWUxTENdet7CeiEi7

It's just fascinating how fast llms are growing, it continuously gets cheaper, faster, smarter I'm 30 now and I would give everything to be born today, be 30 in 30 years

He is so powerful that he got kicked out from his own company....

For the returns for example, I know that if I am getting a Amazon package and I also have a return for them, when the delivery driver drops off my package I can give the return to them too.

I hate people hard focusing on Contextwindows. Just let me waste 2000000 tokens and 15$ in computing power just so the model knows I like dogs LETS GO. Contextwindow is not useful. It's an expensive crutch. Rewinding and using relevant context is cheaper and more useful and actually workable, isn't it? Having models train in real time instead of being frozen after training, and being able to pull relevant information permanently in any context like a human is the ultimate goal, but I don't see us achieving that anytime soon. (Basically, you need to know which context is needed for the model to find the right one, as a simple semantic search of the text often does not return all relevant results and only focuses on that part. No input of preferences in the context. No use and reference to obvious things that are not directly related to the topic. It cannot remember as organically as a human and probably never will). So in short noone got Memory and context figured out jet. The best we have is semantic retrival via Embeddings and that can be realy limited aswell.

your right headphone will be over your eye soon ;D Thanks for the great content David!

Dude, is this a real video? Are you a real person? I feel like I just plugged my brain into a garbage dump for 2 minutes

how does this channel give such scam vibes

Bro, why using the sentence "Nigerian prince sending some gold" as a spam, I personally find this offensive, other neutral examples can be used other than mentioning what may incite discrimination

I appreciate the clean simple video. I didn't run into the same issues as you with Llama 3 though. Still it gave me enough to get my head around this though.

An Uber driver talked to me about AI agents...

thanks

Not a huge fan of the glasses. I'm pretty sure those glasses did not have a display like they showed in the video. It's a better form factor than the humane pin, but I still think handheld device with screen is better. Of course, when they have heads-up display, and more importantly augmented reality, all that changes. It doesn't even have to be good AR, just being able to highlight objects and display text or images is fine. Of course, the biggest drawback to glasses is battery size. We need to unlock some new battery tech first.