naillo

18 ms·

211.
by naillo 4y ago
Their repetition of and use of the term 'Year of Efficiency' always feels like branding that'd come out of chinese government.
212.
by naillo 4y ago
ML would have been fine without torch, we would have just used tensorflow instead.
213.
by naillo 4y ago
The biggest thing here is that it's semi capable and improving. I feel safe about my job right now but it is worrying to invest time to compete with a machine that will continue to get better over the years where previously I felt safe
214.
by naillo 4y ago
This is really cool but the output is such garbage at that weight size that you might as well be running a markov chain.
215.
by naillo 4y ago
Fonts still being able to make money makes me think that all this AI art generation probably won't put artists out of a job. Surely we have enough fonts at this point to be a 'solved' problem, and yet somehow people keep payi
216.
by naillo 4y ago
Are these done with NeRFs? Feels like they are but could just be lots of photos around the objects. Curious if anyone knows.
217.
by naillo 4y ago
Well yes and no. Image 'search' has kinda been radically improved via stable diffusion. It can help you find things that search never would have enabled you to (e.g. because of 'anti biasing' or because the search just i
218.
by naillo 4y ago
I appreciate mercilessly curating your feed. But man even having to do that is such a huge waste of time and life energy.
219.
by naillo 4y ago
ChatGPT runs a highly fine tuned (and pruned) version of `text-davinci-003` so it's probably much much smaller and thus cheaper than 003. Possibly as cheap as 10x less or as much as the `text-davinci-002` or earlier models anyway.
220.
by naillo 4y ago
Also even if that were the case, is that worth diverging from the main mission of the creation of the company to do? If they created the company with non profit in mind (since the consequences of a profit maximizing AI organization is the r
221.
by naillo 4y ago
Somehow this tutorial never clicked for me. I finally looked elsewhere and it got way easier for some reason. I think it has to do with him using his own little library in almost example which hides away some important setup (which obviousl
222.
by naillo 4y ago
To be clear I'm saying that you claiming that data won't leak into 3rd parties by storing it in aws is deceptive. There's no way you can protect the user of your product from having their data being soaked up by openai as lon
223.
by naillo 4y ago
Every app using GPT will go through OpenAIs API and all that data is stored and used for training future models. No amount of good will or non nefarious intentions from IngestAI side makes any difference.
224.
by naillo 4y ago
Interesting that just a few days ago they posted about opening basically the same thing but focused on ML https://developers.googleblog.com/2023/02/MLOlympiad2023.htm... Interesting contrast and I guess a sign of
225.
by naillo 4y ago
Google has never looked more tired and more prone to be overtaken.
226.
by naillo 4y ago
So happy this company is going to be the first in charge of deploying AGI to the world.
227.
by naillo 4y ago
There are some open source LLM models already such as the one this repo is running and mentioning like OPT-175B
228.
by naillo 4y ago
Makes me think karpathys "bet" on vision being the only sensor was maybe misguided
229.
by naillo 4y ago
Sounds like a great opportunity to buy high value ads then at a huge discount.
230.
by naillo 4y ago
I've used it quite a lot and I agree with the original post. It seemed really useful at first but then it started introducing several bugs in large blocks of code. I've stopped using it in the end since the small snippets on the o
231.
by naillo 4y ago
Especially in the context of getting fired. I doubt google would call them 'googlers' in the very email they're getting laid off in.
232.
by naillo 4y ago
Regarding the last point, does this mean developing UIs like this is illegal in some parts of the world? I have no idea asking if someone is more knowledgable than me (e.g. europe).
233.
by naillo 4y ago
Never realized how little data it was fed with. 570GB can fit on my laptop.
234.
by naillo 4y ago
I've been using this https://github.com/JustAnotherArchivist/snscrape and with 5 threads and no vpn, just my laptop, have been getting about 1 million tweets a day (vs 1000 before being rate limited on the api). L
235.
by naillo 4y ago
Still works though (mostly) and still a pretty cool service.
236.
by naillo 4y ago
We're definitely like 5 years from being able to have really realistic NPCs in games. But then probably another 10 until they're actually widespread (RL AI have been able to be incorperated into games for years now but these thing
237.
by naillo 4y ago
This is a great from-scratch (non bloaty) tutorial on this imo https://colab.research.google.com/github/huggingface/diffusi...
238.
by naillo 4y ago
I couldn't really parse the header to tell if he authored this essay or if he asked chatgpt to make it up. (If the latter, pretty impressive!)
239.
by naillo 4y ago
Also as part of this exchange Microsoft will own 49% of the company.
240.
Relative Memory Access Speeds (interactive)(overbyte.com.au)
2 points by naillo 4y ago|0 comments

More ›