winddude

19 ms·

181.
by winddude 3y ago
Okay, I didn't think of that on first thought, but I guess it's best to take the conservative approach, and only allow what's understood. It's almost like the principle of least privilege for the response, there's p
182.
by winddude 3y ago
[DELETED comment] fuck, I read that wrong.
183.
by winddude 3y ago
" prompt leak attacks are something you should accept as inevitable: treat your own internal prompts as effectively public data, don’t waste additional time trying to hide them." But that's relatively easily to prevent, in th
184.
by winddude 3y ago
"This is like the robot's ability to learn from its mistakes. Auto-GPT can review its work, build on its previous efforts, and use its history to produce more accurate results." Can it? I've played with it a bit and have
185.
by winddude 4y ago
it uses a lot of memory because it's an in memory db, that's by design.
186.
by winddude 4y ago
Interesting, the though had crossed my mind, and had briefly tested gpt3 years ago for this. H Have you bench marked it? I might add it too my benchmarking tool for content extraction, https://github.com/Nootka-io/wee-b
187.
by winddude 4y ago
Fuck.
188.
by winddude 4y ago
They had it in the works, but I think the timing is to try and kill some of the momentum from opensource and what people are working on from the leaked llamas.
189.
by winddude 4y ago
Yes, wait list for access to copilotX
190.
by winddude 4y ago
the release today for copilot X says it's using gpt4 "With chat and terminal interfaces, support for pull requests, and early adoption of OpenAI’s GPT-4" ~ https://github.com/features/preview/copilot
191.
by winddude 4y ago
Interesting. Cool to see python becoming more "full-stack"
192.
by winddude 4y ago
"The new A.I. chatbot will be available to a limited number of users in the United States and Britain and will accommodate additional users, countries and languages over time, Google executives said in an interview." So no, no the
193.
by winddude 4y ago
was referring to "(iii) use output from the Services to develop models that compete with OpenAI; (iv) except as permitted through the API, use any automated or programmatic method to extract data or output from the Services, including
194.
by winddude 4y ago
You can't, it's somewhere in the petals docs, but they recommend generating in parallel, and averaging, or selecting the response.
195.
by winddude 4y ago
I got worse than 1 token/sec, and yes, wasn't impressed with bloom results, but I believe it's also very foreign language heavy. I haven't tried it yet but I believe flexGen benchmarked faster as well.
196.
by winddude 4y ago
I can guarantee it's already happened, and been happening for a year.
197.
by winddude 4y ago
Don't they have in the ToS you aren't allowed to use outputs for training downstream? Which is a little ridiculous, considering it's ToS. But yea, they cheap cost and lack of training is making me a take a long hard look at
198.
by winddude 4y ago
> Instead they mandate that apps that operate in China must follow Chinese regulations, privacy laws and of course provide access to Chinese agencies as requested. And how do they do that? With the threat of banning / not allowing t
199.
by winddude 4y ago
> I hope the US takes the EU approach and regulates what data is collected and transmitted where instead of walking the totalitarian route Even if they do, how do you enforce it? The way I see it all paths of enforcement lead to a ban. I
200.
by winddude 4y ago
> Motivation is called a reward function or goal. That's generally a singular goal in RL, and is human programmed. And also you would need to assume AGI won't have a programmed goal or motivation the same way as RL, and will pi
201.
by winddude 4y ago
Anything that loads pickles from sources your unsure of can contain executable code. There were a few samples a couple month ago showing distribution on huggingface. Some solutions for checking: https://huggingface.co/docs&#
202.
by winddude 4y ago
Or you could tell them to axe the price increases. And maybe axe some management and CEO salaries at akamai, https://www.comparably.com/companies/akamai/executive-salari...
203.
by winddude 4y ago
AI existential risk being the end of all or the majority of humanity? No, no I'm not, not as direct result of AGI evolving or seizing control like is widely speculated. The risk I see is always from humans being inept, greedy and stupi
204.
by winddude 4y ago
Came here to say the same thing!
205.
by winddude 4y ago
> It has to be timeboxed into a Sprint to make sure software can be released into the market in an “iterative” cycle. No it doesn't, you just want to be able to get feedback from stakeholders at each iteration. There also exists the
206.
by winddude 4y ago
looks interesting. FYI, the link to your discord in the readme is broken
207.
by winddude 4y ago
Not sure the context on that tweet... that's three tokens from the string `1. Sydney`, so it doesn't appear ` Syndey` is index 1 in the vector list, and than why not got with index 0? Actually ` Sydney` is index 11852 in GPT3 toke
208.
by winddude 4y ago
I wonder how they named it... PM - "What should we name it?" half asleep dev "what about Bing Crosby" Pm - "hmmm, no" Other half alseep dev - "what about Sidney Crosby?" PM - "Spell it S-Y-D-N-E-
209.
by winddude 4y ago
Lighter than air aircraft. Could have been some light weight composite.
210.
by winddude 4y ago
Terrible name too

More ›