วีดีโอ

I believe I did man, it is in the GitHub in my description already I think

You are right my friend, I was trying to avoid providing too many rules and let the LLm reason its way to the correct direction. I’ve noticed over the past year LLMs are not very aware of spatial dimensions which is kind of obvious in heinsight. Thank you for your comment and I’ll back a better one someday

Thank you but nemotron deserves all the kudos

You can try this, it’s a longer video of how I set it up Follow along | deep fake live cam | Slow video th-cam.com/video/OvjWOwUktxw/w-d-xo.html it’s a bit boring so I don’t have it a long video on my channel

Thank you, I tried a more advanced thing with support bots but I think it would still be training for a week or more before anything intelligent came out. I will try again soon thought

Thank you very much man. I was not happy with it overall and plan to make better ones when I have a bit of time.

I have not seen it on Huggingface and I think this is because of local camera access. It could be possible but the lag effect would probably increase.

Sorry could you explain your request a bit more for me? I have not dived into all the hidden prompt states in most LLms mainly because when users figure out how to access them the company behind it tries to make it harder to find. I might be missing something in your question I’m sorry

I know, it’s a very good question but I couldn’t tell you. I’m personally not using it anymore either. Too many times the applications are limited or don’t work for me

@@eointolster Yeah I just installed it for a specific program while following yt instructions and I could have just installed the program directly. New PC I was setting up.. now I'm considering just refereshing/wiping the drive and installing windows from scratch in case this stupid program had malware. Oh well.

@@paul479 honestly, if you are already near a new install, I’d do it. If you didn’t, two months down the line you will still be thinking about it. In saying that I have not found any problems with my setup after installing or uninstalling pinokio.

@@eointolster Yeah I just did it. Lesson learned!

@@paul479 I should do it too, I installed a 500gb c drive and. 1 terabyte d: drive been driving me crazy for a year. Next time round I reinstall I’m doing a 2 terabyte c drive

Thank you very much aimademerich, you made my day

Thank you again, not many people check my videos, so it means a lot

You sound like you have a lot of experience already my friend. I’m super happy I could help in anyway. I’m actually working on using it for a robot at the moment was hoping to make a video later on it but having technical difficulties with my arduino not connecting to the network. I figure I could use video streamed from my phone on my robot and let ell call the function for reading the image and maybe I can get it to move properly. Might be a few days after work to get it to work though. I like the tidiness of ell and want to see grow

I think it is possible but I have not got access to one currently. There are options for turning the gpu off in the documentation, have you tried that?

Might be awhile before I get access to an integrated gpu computer my friend. I’ll will try just the cpu at some point. What are you running?

I see what you are saying and thank you for the comment. I still think in 40 thousand years plus he could have thought of a better way than to murder a large portion of the Galaxy to achieve his dream of a good traffic system ( web way ). The Emperor is dishonest, dishonourable and bad at planning imo but I still really enjoy the setting and I don’t think these attributes I see in him take away from the story, in fact I see it as a positive and enjoy the dichotomy or the Imperium I.e doing everything they can to save humanity except for things that might work.

Thanks, have you read/listened to the children of Dune? The golden path is very much like Warhammer 40k

@@eointolster I have yes I'm a huge dune fan and I do agree that the god emperor and Leto the second are similar in a lot of ways. There's also other dune references in 40k like plasteel. I think the concept of abomination in dune is also like a chaos possession in 40k.

@@LoganOpals considering warhammer 40k is kind of loose with timelines I like to connect all the stories but I think Games workshop is a bit more strict about it. Maybe Dune could be like the interex and they get destroyed by the God Emperor on his crusade

Thank you

I figure everyone else is “ for the Emperor” but nobody is for Nurgle, which makes me kinda sad, he actually cares about his offspring

Thank you for your response Rexmagi, in all honesty I am sure you are right on all counts. I don’t mean that sarcastically either. I will say I do feel his planning ability sucks but as you say maybe I don’t have a good grasp of it. There are a lot of TH-cam videos on the “factual” parts of warhammer 40k and I really enjoy them but I’m going to take the approach of taking the piss out of the factions for my videos, I’ll try to mention it is all in good fun and I am always happy to be corrected. Thank you very much for your message I do appreciate it and sorry if my takes annoy you

Wives? It’s a fun dystopian universe full of action

@@eointolster Yes. But I love them nonetheless.

@@DGH-q3b I’m just saying it’s normal to love things it’s not the same as loving wives or kids. Like I often say “ I’d love a cup of tea” or I’d kill for a cup of tea. It’s not entirely literal most the time

​@@eointolsterof course. I only kill fpr cofee

@@turtleanton6539 lol

Agreed but not a great plan though imo. Still warhammer 40k is awesome and I enjoy it a lot.

Thanks man, hope you find some interesting use from it

True though, look at Vulkan of the Salamanders

Crazy right, still can’t wrap my head around it

@@eointolster You can't wrap your head around it because you imagine "the particule sent one by one" as being a physical "small marble" object. It's not. Its position is blurry and spreaded out. The only thing that is real is the wave function (and it's a probability wave, not a position wave). What you see on the screen is the probability wave collapsing in a single location. So fundamentally, anyway, we live in a simulation. lol.

As I understood it, all the way up to the slits we can measure it and know its location. But if we stop measuring at the slits then we don't know which one it passed through and it even passes through both, and the probability wave creates an interference pattern. I also love the simulation hypothesis.

Thank you, very different video than my normal.

It just overlaps the face above the tongue so you would see a face over the hand

Just for you man I recorded this video th-cam.com/video/OvjWOwUktxw/w-d-xo.html

Ok Ganesh, I will do it from the beginning soon.

Still waiting 😊​@@eointolster

@@Ganeshkumar-ch1kg sorry man, I had other stuff to finish yesterday, I’ll make a step through video tonight my time

th-cam.com/video/OvjWOwUktxw/w-d-xo.html Hey man, personalised video, should work, it is unlisted but should demo everything I did to get it working

Give the choice I would take the 3090 as it gives you more options in general, more vram but deep face live will work on both. At present it is not using a lot of vram and I believe an work on cpu also

Yes it is live Ganesh, in the description is a line by line of how to do it and this video covers the fix, th-cam.com/video/5jZJnNE1kA0/w-d-xo.htmlsi=3uDwsp21jxNJ4yLc

Thank you for watching the channel

4070 or 3090 which is better​@@eointolster

I prefer 3090 for the more vram, it has more capabilities with llms

Thank you let me know if there is something additional I can do for you in this tech

I love it too, it’s from the game Valheim. Amazing game. Was the music too loud or too low?

Sorry about that I’ll try make my text bigger in the future. Thank you for the feedback, it really helps

Yes I believe it is very similar, I think the bit more to it is the reflection part. In a video I made just after this I show what you are talking about with Claude also.

Unfortunately it’s not possible the A6000s are in another building I just have access to them through an online portal. Believe me if I had one in my hand it would be all over the internet :), sorry I can’t help there

I have seen reflection a couple of times but honestly not consistently. It seems to be reflecting / thinking at the same time. Could be my questions

No but it’s a good idea, I only got access to this card yesterday and was suffering from food poisoning on the same day. Weekend with kids now so hopefully tonight I get a proper whack at it

Thank you for letting me know. I’m hoping to get a Mac with the same kind of specs as you

​@@eointolsteroh, it's PC of course, i would never buy any Apple stuff, altrough i have a Macbook & Mac mini.

@@marpleka I’m getting a Mac and I also feel bad about it but it will work for more models more easily and more often. Last year was brutal watching everything come out on Mac and me with my 3080 couldn’t do half the stuff I wanted

@@eointolster i've chosen different way. There's just lack of memory, i bought used Gigabyte server motherboard($150) with 12 Ram slots and can now run 405B Llama in q8 (it uses 465Gb Ram=$90 each 64Gb), yes on Cpu it's slow, but speed not affect quality at all.

@@marpleka oh wow that’s smart. I don’t care much for speed myself, I’d just rather the right output in the first place. Was going to make crewai work with sonnet and the Reflection model for that purpose. I really like what you did with the Ram, clever

My head is hurting thinking about this, you are possibly right, I just wanted to show it thinking and which allows you to see how it got the wrong answer in this case

@@eointolster yeah I forgot the answer too, it's funny how I solved it in 2 minutes the first time now I can't anymore

@@sephirothcloud3953 yeah no idea myself it’s a weird one, I can always remember the puzzle existed but for the attributes of the puzzle and often the answer

I don’t have a Mac but I believe you can. I’ll be getting a Mac next month but Mac 3 so won’t be able to tell you. Please let me know if it works for you

@@eointolster Ill try installing, mac 3 ? you meant M3 chip one ? my M2 pro max is 32 gigs 1 terabite so i hope its good

only quantised, but bad platform choice, Apple don't allow Ram upgrades too much historically, for future better to migrate on PC (i would say Linux, but Nvidia driver ability to use system Ram in help to GPU are only on Win for now-very needed in most Ai tools in Pinokio, or no memory Cuda error always showing)

@@marpleka 39 years of windows issues and Intel issues, i switched, and im happy now, but i know what you saying,

@@LuciousKage ah, current Ai started censoring my replies - that's a harsh reality, you can check Exo cluster project for Macs

Thank you so much man. I really appreciate it, I’m just lucky today though I think

Yeah I’m not 100% certain if it will but I think there is CPU mode on a lot of large models that still run pretty well. Also I’d love a 4090. I’m using online compute here so not entirely mine and it’s literally just the first 24hours of me having access. Let me know if you want me to test out anything for you.

I have an idea what to test. To play to Reflection's alleged strength, try to get it to proofread a text for content, style, syntax and grammar as well. If the output is excellent, it has beaten all the other models. No other model promises few to no hallucinations.

@@abelware8083 thank you man, that’s a good idea. I started up Anything llm earlier to test it there and the output + thinking sections, I should have tried that. Will test it out, thanks

@@eointolster I'll be very curious what you find! Let me know if you need any material or help for the test.

I made three videos in total on Reflection and I had to stop after it failed to be any good at coding, which I mentioned in my video. I was amazed the next day to find out the world was coming to the same conclusion. It was not terrible but was just giving the same ish responses as Claude which turned out to be where it was getting a lot of its data. Crazy week, I do dislike the AI gurus who push this tech like it’s the greatest thing since sliced bread, they hype machine is turning people away from the benefits of AI imo

Thank you, sorry about that I will do it again with better audio, sorry about that

@@eointolster drop the music.

@@eointolster dont worry it was a nice video

Will do man, thank you for the feedback, usually not sure how I should go about it

Sorry about that, I’ll do it again more clearly later, my honest apologies

You can try turning on auto captions, which should help you understand better

I really should be having captions on the screen, I get too lazy at times or just too excited to post.

In all honesty I can’t answer the question yet. I have two A6000s and I don’t know if it is using both or just one. I have to do more testing to see. I assume it is running on one as I believe in total I have 96gb of vram or maybe 128. I’ll make a better explanation or video today. I literally just got them and then reflection came out, I build a UI and made a video. Sorry for the confusion but I myself was confused

Sorry I made a response now as I felt bad th-cam.com/video/vVXQMEeSpY0/w-d-xo.htmlsi=GK82MbnUYABcHdsa

@@eointolster thanks for the explanation, no need to feel bad. I have a single A6000, but have been tempted to get a second one. These 70b models are just out of reach for 48gb and the quantized models become pretty dumb and are still rather slow. I'd be curious to know more about your setup. Thanks for the video.

@@artificialcitizens2399 not to sound dumb but I’m curious about my setup also. I got given access to the other day and have never used online gpus before. I know I have 2 a6000s to use but I did see with my own eyes Reflection using only 1 of the GPUs. Yeah it is a bit slow but I’m hoping to leverage ollama ability to switch between models and attach it to crewai and see what happens. I’d appreciate any comments on the other videos I make regarding this as I’m newer to it than you I think

40 gb? more than I have.

70

I’m going to get more info on this tonight and see, I think 40gb will work but slower as I believe it can still run on cpu but it is just it runs slower. Sorry I’ll have a proper answer later today

Are you sure that is the case? I am not 100% certain of what I have even, I have two a6000s and literally just set it up and ran them made a video. It was like one of my fastest least thought out videos

I made a new video now demonstrating it, sorry I was not more clear in the beginning th-cam.com/video/vVXQMEeSpY0/w-d-xo.htmlsi=GK82MbnUYABcHdsa

Will do, I’ll make it later tonight and see, thank tou

I tried the two guard question here for you th-cam.com/video/vVXQMEeSpY0/w-d-xo.htmlsi=GK82MbnUYABcHdsa