![]() ![]() ![]() In my experience, the flakiness is pod specific. The other good news is that it's fast and easy to kill a flaky pod and switch to another. That said, it's very annoying to have to re-enter the prompt and all the other parameters after refreshing. And fixing that is as easy as refreshing the browser. The good news is that processes like training are still running, it's just the webUI that has crashed. Very frequently the SD webUI dies mid-generation. Yikes! Luckily there's a work around that's only slightly annoying (see advice). The storage fees for a non-running machine is $8/month/20Gb. ![]() Running their machines is nearly twice the cost of Vast.ai machines. Those Automatic1111 extensions are preinstalled. The search and account interfaces are very easy to use.Īfter some setup effort (see advice), it's been an easy way to train Dreambooth models and embeddings. If one is being flaky (see cons), kill it and start another with seconds. They have a template for Automatic1111, and they keep it up to date. Jupyter file browser and terminal are ready to go. No fiddling with notebooks unless you want to. Zero technical experience is required to get going on SD in seconds. I haven't used Vast.ai, so I'd love to hear from people who have.Ĭouldn't be easier. This is solely my opinion after using them for over a month. I'm not affiliated with Runpod.io in any way. ![]()
0 Comments
Leave a Reply. |