วีดีโอ

Hey! Normally you can tell the model what sheets you want it to look at. If you want it to look at the entire workbook then let the model know. So for example, you might say, "Look at this Excel workbook and give me an analysis of the first and third sheets." If the sheets are named, you can use the names like this: "Summarize the data in the "Sales" and "Products" sheets in this Excel workbook."

@@reallyeasyai I'm working on a "DB-based question answering" project for users who aren't familiar with how gpt works. Therefore, when a user asks gpt about an xlsx file, he or she may not explicitly ask about a specific sheet or workbook. For example, if you ask "How much is an apple?", GPT have to find the value in Excel and answer. However, if user ask a question like this and use code_interpreter, xlsx cannot be read and the TPM exceeds 30,000. btw all questions are asked in Korean. thank you for answer.

@@RyuJG Ahh yeah the cost of something like that could get pretty high depending on the amount of information needed to sift through.

@@reallyeasyai So is there anything else I can do about this?

Sounds like a great idea! Let me know how it works out.

Hey @camelendezl ! Unfortunately, for vector stores, there is currently no way to set minutes. The "days" argument only accepts integers as whole days and not decimals so we could put in fraction of days: expires_after={ "anchor": "last_active_at", "days": 7 }

Hahaha Nice! 🎆

🎆🎆🎆🎆 Congrats!

It will absolutely take PDFs. Make sure you are using Claude Sonnet 3.5 but, both, the Pro and Free versions will consume PDF files. 👍

Thanks for the tip! 😀

😀

Glad you like it! 😀

Glad you like them! Happy to have you in the community! 😀

I was thinking about it later on. I hear good things.

@@reallyeasyai it took me a minute to wrap my head around the concept of ; give a detailed prompt of what you want to do and wait as it suggest all the actions it will need to accomplish the task. You add these actions and then run and it does a test run. If it then discovers another action is required, it asks you to add it. It continues until finished.

@@user-iz1kn3ti6d That's actually pretty cool!

It's sad, too, because it is really a good LLM 😢

Yeah it's pretty cool and, as the LLM improves, it will get insanely good 😀

Thanks for the suggestions. I hope I don't have TB 🤔 I'll definitely be hiding my tabs going forward if possible. Sometimes I need to switch quickly between them so it really depends on the context. 👍

I guess it's just me. The technology is my passion. Politics as it impacts the technological space is important to me but whether some Senator is mad about the carbon tax (on either side of the issue) just holds no interest for me. 🤔

@@reallyeasyai right, you're a glorified calculator. Fair enough. Technology is a tool. And again, a person with a broader and more powerful intelligence would realize that politics is in fact a technology in and of itself, it is no different than a spear or a sword, its impact and importance is far greater than either though. It's funny how the autists always believe the machine is greater than the machine that built the machine. They can never see the bigger picture, they never realize that companies and societies themselves are machines, and are far more complex than any computer embedded in the machine. Anyways. It doesn't matter.

Hey! It's been a minute since I did this video. What was the issue? If you have a good fix, share with the folks. Much appreciated. 😀

Yeah I eventually figured that out. It is just one more item in a long list of grievances with OpenAI. I love it when the product works and there is decent documentation; however, since the launch of 4o, they really seem to be struggling with service issues and the documentation isn't keeping up with the changes.

@@reallyeasyai agreed, as a total noob I was suffering with the documentation until I found your channel. And to add assault to injury, their own chatgpt gives outdated code. Had to figure out the migration feature. Either way I love it and learning this whole thing

Thanks! I was getting a little down because of the views but it's folks like you who keep me going! 😀

Unfortunately, no. Currently only those generated can be modified. Now if you want to modify a picture you made then you should look at the Adobe Firefly suite of products. I believe they offer that type of functionality. Of course they aren't cheap.

Interesting 🤔

He's definitely a character 😂

🤔

I can't wait! 😂

@@reallyeasyai Bill Gates probably already has one I bet you.

Unfortunately not yet. That level of integration hasn't happened but I can imagine they will begin having the products interoperate in a future version. AI is basically just "bolted on" for now so they can say they have something.

@@reallyeasyai , thanks for the feedback. Your tutorial is great. I’ll check out the features.

I don't disagree. I think that DALL-E definitely needs to catch up to MidJourney. As for Imagen, well, let's just say they have their own set of problems. 👍

Dang! I suspected they might do something like that. It will probably some back at some point but I think, after the Recall debacle, they are circling the wagons a bit on the products.

Thanks for the feedback. Honestly, it sis feel a little click-baity when I did it. For the news, I tend to take the titles from my Gen AI that I use to give me my summaries. With that said, I did want to convey the idea that (1) we are early stage in AI and there are going to be a LOT of good, bad, and ugly products; and (2) there will also be straight-up lies from product creators like the makers of Rabbit R1 have done. I can forgive the product flopping, products flop all the time; but the Rabbit R1 scam is unforgivable in my mind. Just my .02 👍

@@reallyeasyai Thanks again for your quick response and transparency. I won't argue because your response is your opinion and everyone is entitled to theirs. I will say that I ordered four rabbit r1s. One for each of my two adult kids, one for me and one I gave to a friend. I am really enjoying mine. But I do understand your opinion which you share with so many. I, as do so many others am enjoying being an early adopter. I enjoy the things it does well and am disappointed with the things it doesn't do well or at all yet. With that said. They are updating on a weekly basis and the anticipation of the next update is exciting. I am hopeful for what it will become. Keep up the great work!

Good point. I actually have sued someone (an ex-business partner) before so can attest to how hard it is to get things going. Just the money to get started is difficult for an attorney. Maybe these companies do these things knowing that it is unlikely that an individual will sue them. 🤔

😂

Good call on Recall 😂

Not quite to this level. The logs in windows don't take screenshots every few seconds and really only surface a limited amount of personal information. They are mostly relegated to system and application information. Recall acts more like spyware taking screenshots and other data to capture everything that is going on. It feels like a step too far for me.

Again, I would point out that, if that were the case, then why did OpenAI pull the voice entirely? Johannsson was gearing up for a lawsuit and they completely removed the "Sky" voice. So, apparently, OpenAI didn't believe they would win a suit and I agree. The voices sounded too similar and with similar inflections and intonation in my opinion. Of course, I'm not the only one that thinks this is the case... www.cnbc.com/2024/05/20/scarlett-johansson-openai-chatgpt-sky.html www.theverge.com/2024/5/20/24160621/openai-chatgpt-gpt4o-sky-scarlett-johansson-voice-assistant-her www.forbes.com/sites/roberthart/2024/05/20/openai-says-its-pulling-chatgpt-voice-sky-that-sounds-like-scarlett-johansson/?sh=1f235df95725

@@reallyeasyai so what? there is no law against sounding like somebody else - plenty of artists make a living doing impressionism. Unless they pass it off as her voice, or literally trained it on her voice, there is just no applicable law against "sounding too similar", what are you even on about?

@@RasmusSchultz Well, apparently, OpenAI thought there was a problem so they pulled the voice completely. That's kind of the end of the story there. 😏

Ew? 🤔

Excellent point! 👍

Voices aren't unique either. The collision space is too small. Lots of people sound like other people, exactly like other people.

And why everyone is moving to Linux after Microbloat introduce this privacy nightmare called Recall.

I agree with the part concerning laws to protect our privacy and safeguards for our personal AI assistants. As for the Linux answer, I just want more competition overall in the OS marketplace. 👍

I suppose. But there are better ways to do this and Recall isn't sold as that type of solution. Plus there are laws, even for employers, about spying on their staff. With remote workers, it gets even more interesting from a legal perspective.

😂

Glad you like it 😀

It has definitely caused a bit of reaction. 👍

This is why everyone is moving to Linux.

@@joshuahernandez-th6ov _"everyone"_ ??? Nah, unlikely.

Exactly. And now imagine a big court case where the artist is on public television singing without autotune. For some, it could be quite interesting. 🤔

Good one 😂

This is why everyone is moving to Linux.

I agree. I was literally told to take "influencers" to strip clubs back then. It sucked. It reminded me a lot of the fiasco with Blizzard and all the craziness they were getting up to. www.theverge.com/2023/12/15/24003556/california-activision-blizzard-gender-discrimination-lawsuit-settlement

This is why everyone is moving to Linux.

@@joshuahernandez-th6ov posting your linux comment in every reply is a pretty sad.

@@High-Tech-Geek Sad? Microbloat is a spyware including Recall.