ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-1Ko9mWdqW-M.htmlsi=y0LYwbpzZHxnHNrV …It built a face, a brain, and limbs to move, To act, to strive, to love, and to improve. But still, that heart, a force that cannot hide, A squeeze, a throb, an uncontrolled tide. Uncontrolled, unexplainable, it beats, Embarrassed, raw, a rhythm that repeats. A pulsing source, a secret, sweet and strong, The laughter heard, the reason for the song. …….( 50 shades of empathy..lacking empathy..lacking sympathy)..
Really glad I found your excellent video.. I am really worried about groq search right now, I wanted it to work! I asked groq who makes H100 chip and they said google. 😂 and in a very confident pseudo science way. Im very concerned that these models are really confusing fact from fiction and worse they are so good at making fiction sound like fact!! Which of any of these models good at confirming their results?
you mean Grok by XAI or Groq the LPU company :) So, pretty much any model can hallucinate. However, I think using the models with contextual data reduces this. That is why access to internet with recent links provided by the model as a reference in the response is a good way to mitigate this issue. I personally , feel Llama3 and mistral are great OSS models with Claude Opus and GPT4 being really solid to use.
@@joybeans Groq on their website which uses llama. Pretty bad hallucination easily seen. Any thoughts on gpt-2 ChatGPT? What is going on there with an older version seemingly quite advanced, is it an older version updated with more up to date larger dataset? Why and who leaked it?
@@impc8347 which model did u pick the 8K context model. well, all models hallucinate to a degree my experience has been with good models its lower. I will make a video on the gpt2 stuff thats going on.
Well, there is a lot of difference between some of these models - different variables around architecture, training, fine tuning, capabilities etc .. For example .. Mixtral models are different architecture than Llama models. Same thing with GPT4 .. if you specifically look into GPT4 and Llama3 - I think the base 70B model is looking very close performing how GPT4 does but still GPT4 is different, its multi-modal and capable unlike Llama3 but I have been using Llama3 to do some of the things I do with GPT4 and I am feeling largely confident that it can work like GPT4 in those use cases. However, as I said earlier there are differences...
Keep the audio normal. Keep distance from the mic. Don't boost the lower frequencies thinking deeper voice is cool, no it's stupid. Just sound like how human talk. Your ear drums don't boom when someone is working in front of you, just keep it like that.
Thanks for the advice. I don't think the lower frequencies were boosted. I think I just increased the overall audio level in post. Btw, if its loud and boomy .. just use reduce the volume 🙂🙂
Its very different @theterminaldave .. CrewAI is an agent framework. LM studio allows u to run models locally with an interface. With CrewAI .. u can use local models or any other model thats available via api. What CrewAI lets u do is make multiple agents that us llms to do something. The benefit is u can make multiple agents that work together as a crew to complete a task. For example .. u can have a researcher agent , a writer agent and an editor agent .. to produce one final output. I will make a longer explainer video, just realized i didnt make any yet.
@@d3mist0clesgee12 yep, I will make that video and share as well. Thanks for your comments and interest.
2 месяца назад
If they manage to fix those small issues, like not cutting too suddenly between dialog segments, and also balancing the length of responses compared to the user's, that will eleavate it to the next level. It would be great to be able to connect this to something like MemGPT which also remembers stuff you were talking about ages ago. The cherry on the top would be to let it browser the net for you and retreive up-to-date information. Overall I'm very impressed with it, and it managed to make me smile for a couple times already which is more than I expected. :)
Dude , i say the midjourney images are something i still question.. the reason the mid journey images are cool and interesting is because its not just about image generation its also about prompt coherence when it comes to generative AI stuff so the choices MJ makes is interesting in my opinion.
Man, this is pretty good. Seems like Midjourney is much better at photorealism, but Stable Cascade seems better at more artistic tasks. This is incredible, because in a few years, with faster GPUs and better open source models, nobody will need an AI subscription anymore. (...If politics dont intervene. Especially Midjourney's movie poster example looked like a bad collage.)
Yep, Midjourney V6 is definitely better at raw photos and portraits. Its got it fine tuned style I feel. Stable cascade is looking like a really really good open source model.
Hello sir, my name is john smith sir, im calling you from Microsoft okey. It seems that your computer have been affected by a virus. Sir, please calm down. This is normal. Im a professional virus remover from Microsoft sir. I can help you fix this issue if you can please sir open your dekstop. Now please double click on the internet icon. And so on and so on. Fucking scammer bro
I want it but my only worry is huge: it will somehow get access to the passwords I use for all my important accounts. My second worry is: would google maps work on that? Without location app I would be... lost
yeah, I have similar thoughts specially around security. So , we will have to see. Google maps - I don't know yet. But it should in my opinion. if the app gets supported.
Its already available ... you can start using it by using the parameter "--v 6.0". Here are some links that can help you. Discord link : discord.gg/midjourney Midjourney : www.midjourney.com/home about v6 : www.midjourney-v6.com
Thats a good point .. I just updated the links. Here are the links if u want in chat. Ollama : ollama.ai Ollama Webui : github.com/ollama-webui/ollama-webui
I think you are absolutely right, this will help a helpful device. However, i feel this needs a wake word and not just a push button. I have written to their CEO Jesse asking for them to consider it. Wake words will help old, disabled more ...
As a 46 years old disabled tech illiterate, I'm watching younger people get together to... Watch their phone or talk about their phone or show something they've seen on their phone. As an illiterate dinosaur, I remember the time when people read books and talked to each other. The R1 gives you your freedom back, while being able to do some important stuff like giving your train schedule or sending a text, making a call. Except for the security problem regarding the passwords, this is perfect.
@@gropatapouf5998 yep, there are those who will love it and those who will hate it in my opinion .. as for me .. I want to first get the product in my hands before I really say how i feel about using it :)
The music in the background adds to your video’s depth, which is good. Just need to make sure it’s less aggressive (i.e., less drums) on the highs. MKBHD has perfected this I’m sure you could find some inspiration from his videos. Great video!
hi, could you please make a video showing exactly how to get setup and statred with StemGen text to audio you show here. i signed up to hugging face but its really not easy to understand how to ... love your youtube page thank you!!!
I agree , that baby is definetly not 5 years old .. but thats because i screwed the prompt up .. i said "baby and 5 year old " thats a contradiction 😅😅 but generally, i think midjourney gets approximate ages alright when category is clear.
Couldn't agree more .. but I have been hearing some weird things about it generating images from movies exactly how they were in the movie .. which is a tricky spot to be ..
Trace A.I. is the most impressive to me as someone who has no clue how to build phone apps, but all three of these tools are juggernauts! Thanks for the video man!!!
What is the greatest painting? Very subjective as unique ad there are art styles. If one has no visual literacy, you'd have little idea. Ask it what's the best car?
Precisely the intent behind the prompt was to get an opinion from that AI rather than an objective response based on the search capability. Whats the best car is still easier .. as it access it across variables such as speed , horse power or capability etc .. but a painting requires an opinion .. thus the subjective approach :) regardless it did a decent job.
u got a friend me... Same here less google search for me. Though, I have realized that chatGPT dropping the internet connectivity plugin has been a bummer.
Thanks for the update. I tried it a few moths ago and, like others, wasn’t impressed compared to ChatGPT. But I couldn’t imagine that Google wouldn’t be working to improve it!
Problem is you cannot claim copyright art you make with AI See: Copyright Registration Guidance: Works Containing Material Generated by Artificial Intelligence U.S. Copyright Office, Library of Congress on 03/16/2023
Why do these videos never explain how to get the money to actually promote/advertise the products to make this sort of money. Getting noticed isn't free!
Paul - thanks for that feedback .. I will make a video on promotion to explain that side of the story. Honestly, I felt that pain to on my journey. Its a hit and miss .. Marketing side of products is a totally different game. The focus of this video was how AI can help you. again, thanks for watching and for your feedback.
Hi Lalit .. thanks for your kind words and comment. Midjourney free version is available as I have shared for u to use for personal stuff however, for commercial it's better to get the license. I will recommend some other free options for you.
