That voice has a normal language model behind it
That voice has a normal language model behind it
LBut the demo is well done, it even included tts errors. so it's probably quite honest in capabilities.
However it was a perfect use case example, normal people are hard to predict.
In any case it's more for illegal spam callcenters than for businesses.
If it doesn't tell me right away that it is an AI I will be super pissed once I find out.
If it tells me it is an AI I'll hang up
However it was a perfect use case example, normal people are hard to predict.
In any case it's more for illegal spam callcenters than for businesses.
If it doesn't tell me right away that it is an AI I will be super pissed once I find out.
If it tells me it is an AI I'll hang up
XI’ve been trying to automate a call center
LIt's the future of callcenters - just that we might not need people or centers anymore at one point.
Right now - everyone will absolutely hate AI calls. In EU it's illegal to use AI to "influence" people so they will probably just ban calls (and ban anything else with AI sooner or later).
In maybe 5 years you might actually want AI, especially for support. Because it's smarter and without waiting time until you need a transfer to a human supervisor.
In maybe 15-20 years you'll not find a human anywhere in that chain
Right now - everyone will absolutely hate AI calls. In EU it's illegal to use AI to "influence" people so they will probably just ban calls (and ban anything else with AI sooner or later).
In maybe 5 years you might actually want AI, especially for support. Because it's smarter and without waiting time until you need a transfer to a human supervisor.
In maybe 15-20 years you'll not find a human anywhere in that chain
LI'm heading to bed, gnight
XSucks, but I agree ai has a huge side eye from a lot of call center kind of businesses. However, I don't think we'll see call center employees reciting the same script over and over again in the next two years.
XSo overall conversational AI for scheduling appointments isn’t really good yet, you think?
XThis is very dangerous lol!!!
LOh I think it definitely is good enough but most people will have problems implementing something that does the job sufficiently. It won't work out of the box.
The image I posted is just a good example what happens if you put someone incompetent behind it.
Models, finetunes and commercial APIs are all improving at a pace that is almost unseen in human technological progress, so it will get easier
The image I posted is just a good example what happens if you put someone incompetent behind it.
Models, finetunes and commercial APIs are all improving at a pace that is almost unseen in human technological progress, so it will get easier
Fit will cause some unemployment certainly
Fby the way
Fif i i understand ai talking i dont talk anymore
XWhat do you mean?
Ac
Fi mean in customer services i hate AI 
MYouTubeSECourses
Master Stable Diffusion XL Training on Kaggle for Free! Welcome to this comprehensive tutorial where I'll be guiding you through the exciting world of setting up and training Stable Diffusion XL (SDXL) with Kohya on a free Kaggle account. This video is your one-stop resource for learning everything from initiating a Kaggle session with dual ...
Master Stable Diffusion XL Training on Kaggle for Free! Welcome to this comprehensive tutorial where I'll be guiding you through the exciting world of setting up and training Stable Diffusion XL (SDXL) with Kohya on a free Kaggle account. This video is your one-stop resource for learning everything from initiating a Kaggle session with dual ...Mand does lora need regularization images? if so which set would you recommend for a style lora?
Ahopefully we will get a dreambooth guide for pixart soon, i got it working, but im still a bit uncertain on the parameters.
also is it possible to get around the 120 token truncating with pixart?
also is it possible to get around the 120 token truncating with pixart?
Fyou can
Ffor style you usually dont need
Fjust have good and big dataset
Fi never had chance to try yet sadly
Mfor style dreambooth training (to later extract lora) can i use 10-15 epoch and lesser repeats like 25? also save every n steps at 0?
Mbecause in your training video for faces you set epoch to 1 and repeats to 100 i was wondering if my settings would be suitable for style if i keep everything the same in your v12 json but just change those settings i mentioned above
Mand one last thing because the vae is set to stabilityai/sdxl-vae in the v12 json file, when using the trained lora later is it going to be a problem if i use it with a checkpoint that needs a vae? should i delete that setting if i'm going to extract lora from the dreambooth training later?
Fdepends on number of training images
Fit shouldnt matter much
M"I recommed tweaking your repeats so that a single epoch (number of images × number repeats) is 200-400 steps." is this a good formula?
Mso 200-400 divided by the number of training images we have?
Mthank you
Fthat would be too few steps
Fusually above 2000 steps necessary
Mbut if i set epoch to 10-15 it would make the steps much higher than 2000
Mor is setting epoch to one and using save every n steps a better method? so i guess my question is to train 10-15 epochs with 40 images, also considering that i'm training for style, how many repeats would be plausible?
Moh there is not enough space for more than 7 checkpoints so i can't set epoch to above 7 anyway 
Mso what i did is 100 repeats for 40 images with caption and without regularization Save every N steps set to 572
Yhas anyone played around withthis https://stability.ai/news/stable-video-diffusion-open-ai-video-model
Stable Video Diffusion
Stable Video Diffusion
Stability AI
Stable Video Diffusion is a proud addition to our diverse range of open-source models. Spanning across modalities including image, language, audio, 3D, and code, our portfolio is a testament to Stability AI’s dedication to amplifying human intelligence.
Yokay cool. do you have any plans for tutorials .. kind of addicted to your tutorials now
Fyep i plan for that too
MHi again
can you help about this ?
https://vxtwitter.com/zhengyiluo/status/1740506602615214231
https://github.com/ZhengyiLuo/PerpetualHumanoidControl
i dont know how to install it correctly probably the way it say about conda line cant be done for me and i dont know why
can you help about this ?
https://vxtwitter.com/zhengyiluo/status/1740506602615214231
https://github.com/ZhengyiLuo/PerpetualHumanoidControl
i dont know how to install it correctly probably the way it say about conda line cant be done for me and i dont know why
vxTwitter / fixvxYou can now ask your simulated humanoid to perform actions, in REAL-TIME 👇🏻
Powered by the amazing EMDM (@frankzydou, @Alex_wangjingbo, etal) and PHC.
EMDM: https://frank-zy-dou.github.io/projects/EMDM/in…
💖 248 🔁 49
You can now ask your simulated humanoid to perform actions, in REAL-TIME 
Powered by the amazing EMDM (@frankzydou, @Alex_wangjingbo, etal) and PHC.
EMDM: https://frank-zy-dou.github.io/projects/EMDM/in…
248 
49
Powered by the amazing EMDM (@frankzydou, @Alex_wangjingbo, etal) and PHC.
EMDM: https://frank-zy-dou.github.io/projects/EMDM/in…
248 49GitHub
Official Implementation of the ICCV 2023 paper: Perpetual Humanoid Control for Real-time Simulated Avatars - GitHub - ZhengyiLuo/PerpetualHumanoidControl: Official Implementation of the ICCV 2023 ...
Fhello it looks like a complex and very specific program
Fi can only help as private consultation
Freally busy right now :/
LHey, could a Langchain-backed ChatGPT client filled with specialized attorney/judge/psychiatrist agent be able to ask each party for their viewpoint, and read whatever court docs would follow, and make a proper judgement?
It's for this courtcase where a woman has lost custody over her child in a very cruel way.
It's for this courtcase where a woman has lost custody over her child in a very cruel way.
Fi think GPT-4 should be able to do it pretty good right now
FUSA court documents are open
Fso they train on them
LOh hey that's great news! Cheers
LWe're taking Norwegian child custody service to court
This GPT will destroy them
This GPT will destroy them
