sa-code

11 ms·

151.
by sa-code 3y ago
This sounds like the worst game of telephone
152.
by sa-code 3y ago
Oh for sure. We're in the golden age of LLMs before the enshittification commences and we don't appreciate that enough
153.
by sa-code 3y ago
There was a paper published by Microsoft that seemed to leak this detail. I'm on mobile right now and don't have a link but it should be searchable
154.
by sa-code 3y ago
Gpt 3.5 turbo is 20B
155.
by sa-code 3y ago
I hear a lot of complaints about refusals but rarely any examples of said refusals, likely because they are embarrassing. Is it fair to assume that I won't get refusals for code generation and RAG on documentation?
156.
by sa-code 3y ago
Because not every company that uses GCP gets offered this discount
157.
by sa-code 3y ago
Yeah it's only nice because they haven't enshittified yet. I would give it 2 years
158.
by sa-code 3y ago
How is WanDB different from postgres?
159.
by sa-code 3y ago
It would be cool if there was a wasm package manager that you could use from any language, and each lib could autogen interfaces in the language that you're working in
160.
by sa-code 3y ago
The point you missed is that when insurance is hyper tailored to specific people, you are basically paying for your own healthcare and can cut out the middle man
161.
by sa-code 3y ago
This reads like it's written by someone who works at vercel
162.
by sa-code 3y ago
Their keynote at the llvm conference looked quite substantial
163.
by sa-code 3y ago
Wall hacks are also a type of cheating that doesn't require changing your communication to the server
164.
by sa-code 3y ago
IMO the same could be said about voting. If someone states that they voted for (in this case with their wallet) for a worthwhile cause I wouldn't respond to it by reminding them about the insignificance of a single vote.
165.
by sa-code 3y ago
Sounds like a good reason to give your business to someone else
166.
by sa-code 3y ago
Lucene's WFST is absurdly fast
167.
by sa-code 3y ago
Being employed, so that Google can say they employ them
168.
by sa-code 3y ago
Mojo could potentially fill this gap in a year. They still have a long way to go, but they're working on traits and sum types right now
169.
by sa-code 3y ago
That's fair, I'm not really a compiler person but what you said makes sense
170.
by sa-code 3y ago
IMO the reason Julia gets to be this fast is because of LLVM, and the guy who created LLVM is also the creator of Mojo so there is something to be said about that
171.
by sa-code 3y ago
The language is really new, it only supports amd64 Linux at the moment (at least publicly)
172.
by sa-code 3y ago
In your opinion, what are some examples of things that work and don't work? And can you talk about what's the tools do?
173.
by sa-code 3y ago
Indeed, and its existence was mentioned in the llama 2 paper too
174.
by sa-code 3y ago
The ability to hook it into frameworks like streamlit would be huge. Sorta like llama-cpp-python
175.
by sa-code 3y ago
Thank you!
176.
Ask HN: What is the best Llama model that you can deploy on a single A10G?
2 points by sa-code 3y ago|2 comments
177.
by sa-code 3y ago
What about just using a cloud VM with an ansible script? I find ML deployment solutions to be very over engineered