Видео

Real nice explaination, Could u help me out with the open ai api key? It only works if you have a paid openai account , any other alternative?

How do I add a chat history here as a reference for the bot's answers?

is not there a way to use faster-whisper with groq?

You misspelled "Annoucement".

Just enrolled...happy that i did.... Half way through already.....

Joined already 🎉 just hope that you will keep the content updated with today’s dynamic AI landscape

do we have to add the tiny pixel model to colab?

why do so many "use your own dataset" videos are just using online datasets? this has nothing to do with my data, it is custom yes, but not mine. it is online dataset from hf. my dataset wouldnt be there

Interesting, and I am prolly going to drop $ the for this course. The only thng holding me back is..... where are the vector stores. I really like your approach to 'local' for everyone, and tell everyone I know interested in local AI to look at your channel ....... but I don't want to be pointed at 'cloud based' vector stores that use pg-vector plugin extension .....when Postgres is already free, open source, and what the plugin (pg-vector) plugs into. That's the 'decider' for spending $200. Do we get pointed back at something like Pincecone, when postgres running locally overly fits the bill with the very. Same. Technolgy, sans giving our data to Pinecone (ect ilk) and paying them a fee, when I can run the same postgres faster from the Desktop. Are you pointing people (with code) to cloud based vector store, or pointing out the same free version of it on the desktop? It really matters when developing a flow. I'd like an answer to that before spending $200 Other than that, keep doing what you do. It's great.

Will you add this to Udemy?

Thanks, this does sound v tempting and v interesting - congrats on putting this together. It would have been helpful to have had a brief demonstration of the best version of the solution that you have put together. ie does this really work in practice on real research type problems that we encounter, covering hundreds of documents and being able to ask complex questions requiring inputs from multiple documents (as opposed to the typical "who is the president of the US"-type questions that we see on so many other videos of RAG demos). Also, assuming it can do such tasks, what is the average run cost of a typical query, as obviously there are ways of simply cramming the context window with massive chunks of text to the extent that each query costs a fortune). Lastly, kudos for offering to maintain the content over a period as it does seem like the market is changing v quickly in this space.

Good luck i would like to learn about AI but not RAG maybe next course :)

First of, thanks for all the great content ! Any plans for a nodejs version of this course ?

artifacts does not seem to be working for me. I entered the same prompt and got “I apologize, but I do not have the ability to generate edit, manipulate, or produce images.” I’m on the professional plan. Any advice?

Nice outline and love your way of delivery 🎉. But honestly speaking your course price is high. You may cosider udemy platform as well to publish your course with a discounted price. Wish you good luck.

is it better than DeepSeek coder?

8:50 Wrong!!!! The answer is -5 !!!

Hey, heads up the correct answer is -5 @9:02 not -4

Very informative video. Could you try to build a system that can run on a large number of PDFs and further convert these to .md files for an LLM to query or generate specific prompts with a UI?

9:54 it didn't ignore position 9, it just didn't count it as a NEW appearance. It is presuming not a count of the R, but a unique appearance between other letters. Interesting. Once I an prompt again, I'll try "How many times is the letter R used in Strawberry."

When I was a kid I was giving an old computer that used those black floppy disks. There was this one game on one of those disks that was a game with spaceships on a two plane with planets all stationary. The Planets had gravity to them and the goal of the game is to type a number between 0 and 360 and a speed number and the ship would shoot a missile that would fly around the planets and try to hit the enemy ship without getting hit by your own missile or the enemy's missile. For what I seen this game could be easily recreated in Claude Artifact.

Could you add the description of each image to the text with the aim of having a single Markdown file, similar to the original PDF? This way, it would be possible to pass a file to a language model that is readable and maintains its content.

Hi, could you convert complex PDF documents (with graphics and tables) into an easily readable text format, such as Markdown? The input file would be a PDF and the output file would be a text file (.txt).

Please clarify. Is Artifacs only available for a paid subscription? As I'm only getting saying that Claude cannot execute code? Please create a video which clearly demonstrate the enabling Artifacts. Perhaps I'm missing a step or two? I'm using Sonnet 3.5 in the free version.

I saw that ozgrozer has an repo chatgpt-artifacts basically recreating the interface aspect. It’s only a matter of time until Open AI will implement this native, but in the meantime that might be nice to have for people on ChatGPT or with a longer commitment like the yearly teams plan as I have. I must confess, I am also paying for Claude currently… just to much fun to pass up on this latest innovation.

Does OPen AI API cost ? Good video man

i have pdf of scanned documents it is like images os using tesseract i am converting this into images and then reading it but the thing is my pdf has also some table and when i am generating embedding of this and then passing into to an llm it is unable to answer the question which i am asking from the table...

how do you find that chunksize =800 and overlap 20 is a good number?

A lot of thanks for this tutoriel. I have created an avatar with your clear explanations !!

Thanks for the good content. I wonder how this compares to OLLAM? 🌹🌹🌹

Brilliant! Thanks for this one

Gemini 1.5 Pro: This video is about Gemini Flash, a large language model, and its capability for function calling. The video also covers how to build a customer support agent using Gemini Flash. The speaker starts the video by mentioning that Gemini models recently had an update to both Pro and Flash versions. This update allows users to access improved rate limits and the ability to fine-tune the Flash version on their own dataset. The speaker is particularly interested in Gemini Flash because it seems to offer a good balance between quality of outputs, price, and throughput. Then the videos dives into the concept of function calling in large language models. Function calling allows the model to access real-time information through external APIs. The speaker provides an example of using an external API to get stock prices. Next, the speaker explains how to build a customer support agent that can make sequential and parallel function calls using Gemini Flash. The agent is built using Python code. The speaker demonstrates the agent by simulating a conversation between a customer and the agent about an order. The conversation involves the agent getting the order status, initiating a return, and updating the shipping address. The speaker concludes the video by mentioning that Gemini Flash was able to handle all the requests accurately, including the complex ones that involved multiple function calls. The speaker finds Gemini Flash to be a good model with good speed and relatively lower cost compared to other models in its class. Overall, the video provides a positive assessment of Gemini Flash's capability for function calling.

It will work well for a while. Then cause dependency conflicts because of some library that implements a redundant data structure that only Giggle feels we need. Then Giggle will lose interest and move on to something else leaving developers in the pit. Thanks but no thanks. 😂

I Was telling my friend about the power of the SONNET! more than chatgpt 4.O

Can it generate image?? 🌚

Brilliantly explained with clarity and insight, thank you! Also really pleased you point out that RAG emerged from IR ideas and wasn't brand new: when I saw it I was like, haven't people seen Facebook's DrQA from 2017?!? And even that wasn't out the blue, there's a long established history with IR 👍

500 likes, keep it up !

Real GPT 5/6 level intelligence This is not a trick i’ve extensively tested to get these answers first try if anyone wants to access this superior GPT with 10 x vision capabilities 10 x smarter outputs for chat GPT i figured out something crazy, you’ll be surprised of what it can really do *Its called Smart Vision image/text analysis* Or visit my channel description for the direct link (Paste your own custom instructions for a superior smart chat aswell *It can put text into images, solve highly complex math and physics problems even reasoning images* and so much more This is not a trick i’ve extensively tested to get these answers first try, read *IMPORTANT* in the GPT’s conversation starter

it's one of the most useless video on youtube. You literally opened a notebook, and read it. You didn't added a single sentence that would be considered as a plus value. That was like watching a text to speech model in action.

By YouSum Live 00:00:10 Cloud 3.5 Sonet: State-of-the-art model surpassing previous benchmarks 00:00:23 Artifacts feature: Enhances user interaction with Cloud 00:01:00 Performance superiority: Beats GPT 4 and previous models 00:01:30 Model availability: Accessible through API for immediate use 00:02:42 Image understanding: Enhanced capabilities demonstrated 00:03:00 Artifacts demonstration: Interact with model-generated content 00:04:20 Future model releases: Teasing upcoming Cloud versions By YouSum Live

Claude 3.5 Sonnet looks amazing!

Nice comparison! Please continue work on scraping for AI applications. Hot topic!

I wish videos would stop using the phrases 'Killer', 'Changes everything' and the like.

🎯 Key points for quick navigation: 00:00 *Cloud 3.5 Sonnet surpasses GPT-4.* 01:19 *Costs $3 per million input tokens, $15 per million output tokens.* 03:09 *Artifacts feature enhances user interaction.* Made with HARPA AI

Me parece igual complejo, ni un brillo.

in really it kills harder than in benchmarks

Excelent and concise description. Thank you.

still no agents or web search?

no way gemini is that high...

Isn't it funny how they do their versioning? Now, Claude-3.5 isn't just better than GPT-3.5, but also surpasses GPT-4. So, they're basically saying that whatever the latest version of OpenAI GPT is (let's say N), their N-0.5 version will surpass it. This will continue unless OpenAI stops releasing updates until Anthropic's versioning catches up! :D