In my personal experience, Gemini seems to have the best image editing and creation capabilities but when it comes to coding, Claude is my favorite. However, Claude is slow and errors out for me often. Grok is also very good when it comes to coding but I seek to run into their rate limits much easier, so I don't use it.
My comment regarding Google victory was not related to the merits of its current tooling. But to me, it is the only company with already established cash cows. and its speed of adaptation shows that it is not left behind for very long by competitors in delivering state of the art LLM models.
Subjectively, OpenAI's image generation is the only one that consistently works for me and renders what I describe in the way I describe it at an acceptable level of quality. I don't understand how NB gets so much credit compared to what OpenAI is offering.
I felt like that at your age. Now 40 and realizing my energy was wasted worrying when there's so much more potential left. I believe I'd feel similar at 50. There's a lot of living left to do.
> Silent data corruption (SDC) is the most expensive, where a faulty GPU produces wrong answers without crashing anything, which means a multi-day training run can complete normally and the resulting model weights are quietly poisoned.
I went with the self host route by putting it on a few years old computer with much better specs than cheap vps. Cloudflare tunnels to make the web server accessible on the internet.
Nice. How fast is your home internet connection? I’ve thought about putting an old laptop to this use. But I don’t want my day to day internet to suffer if my site gets traffic spikes. And also, I’m nervous about non-ecc ram.
My internet is very fast for sure (can give more concrete numbers when I am home later) but I don't think you need to worry about it unless you are operating some massive website with huge traffic concurrently.
reply