# Ai generative cloud based GPU

**URL:** <https://forum.logik.tv/t/ai-generative-cloud-based-gpu/7846>\
**Category:** Hardware Questions\
**Tags:** jobs\
**Created:** [March 2, 2023, 5:22pm UTC](https://forum.logik.tv/t/ai-generative-cloud-based-gpu/7846 "2023-03-02T17:22:00Z")\
**Posts on this page:** 6\
**Page:** 1

<div class="post-metadata">

**Author:** ![theoman](https://avatars.discourse-cdn.com/v4/letter/t/94ad74/32.png) [@theoman](https://forum.logik.tv/u/theoman)\
**Post date:** [March 2, 2023, 5:22pm UTC](https://forum.logik.tv/t/ai-generative-cloud-based-gpu/7846/1 "2023-03-02T17:22:00Z")

</div>

Hi everyone, I have another deep fake head replacement job possibly coming up. As the turnaround is SUPER quick i have to find a cloud based solution for training 7 shots concurrently. I can only do 1 at a time in my home office so am looking into Google Cloud, Lambda to do all 7 at the same time, but they are not easy to understand or impossible to book instances ahead of time for less than $100k. Has anyone here have experience in getting through multiple shots at the same time using a cloud based GPU facility? Im using Deep Face Lab for the training. I sent a message to Thiago here who seems to have great experience in some de aging techniques but opening it up to the group as i need to bid this soon. thank you!!

---

<div class="post-metadata">

**Author:** ![ALan](https://avatars.discourse-cdn.com/v4/letter/a/a87d85/32.png) [@ALan](https://forum.logik.tv/u/ALan)\
**Post date:** [March 2, 2023, 6:19pm UTC](https://forum.logik.tv/t/ai-generative-cloud-based-gpu/7846/2 "2023-03-02T18:19:08Z")

</div>

look at the offerings of Vultr or DigitalOcean. I happen to like Vultr. Both of them are much easier to use than AWS or GCP.

 ![Screenshot 2023-03-02 at 10.31.03 AM](https://us1.discourse-cdn.com/flex020/uploads/thedepartmentofexternalservices/original/2X/e/e32dd2d1e09df20f86bde9cc6cce4a1f09630309.png)

---

<div class="post-metadata">

**Author:** ![theoman](https://avatars.discourse-cdn.com/v4/letter/t/94ad74/32.png) [@theoman](https://forum.logik.tv/u/theoman)\
**Post date:** [March 2, 2023, 8:31pm UTC](https://forum.logik.tv/t/ai-generative-cloud-based-gpu/7846/3 "2023-03-02T20:31:49Z")

</div>

thanks Alan, ill take a look at them!

---

<div class="post-metadata">

**Author:** ![TimC](https://sea2.discourse-cdn.com/flex020/user_avatar/forum.logik.tv/timc/32/11881_2.png) [@TimC](https://forum.logik.tv/u/TimC)\
**Post date:** [March 10, 2023, 2:06pm UTC](https://forum.logik.tv/t/ai-generative-cloud-based-gpu/7846/4 "2023-03-10T14:06:00Z")

</div>

@theoman did you wind up going down the cloud GPU route? how did it work out for you?

---

<div class="post-metadata">

**Author:** ![theoman](https://avatars.discourse-cdn.com/v4/letter/t/94ad74/32.png) [@theoman](https://forum.logik.tv/u/theoman)\
**Post date:** [March 13, 2023, 4:16pm UTC](https://forum.logik.tv/t/ai-generative-cloud-based-gpu/7846/5 "2023-03-13T16:16:44Z")

</div>

For 5 days i was trying to install Deepfake and all its associated bits and bobs ( tensorflow, cuda, python, conda etc) on a Ubuntu partition ready for migration to 7 instances in the cloud. Damn it was near impossible! I have a whole document of linux installs to get the scripts to work, Last Thursday the job went away so i didnt get a chance to see! I was bummed! BUt damn its harder than the first time i installed flame on a linux - ridiculous!

---

<div class="post-metadata">

**Author:** ![TimC](https://sea2.discourse-cdn.com/flex020/user_avatar/forum.logik.tv/timc/32/11881_2.png) [@TimC](https://forum.logik.tv/u/TimC)\
**Post date:** [March 13, 2023, 8:57pm UTC](https://forum.logik.tv/t/ai-generative-cloud-based-gpu/7846/6 "2023-03-13T20:57:29Z")

</div>

yikes! well, thanks for the feedback! glad you didn’t get burned on it
