stan_kirdey

10 ms·

61.
by stan_kirdey 3y ago
the beginning of Krang I really hope there are tiny turtles learning martial arts right this time
62.
by stan_kirdey 3y ago
Chatting with Falcon 40B feels like chatting with GPT4, very capable model.
63.
by stan_kirdey 3y ago
We might end up with a new character set like gpt-utf-8 or something, offers at least some watermarking capabilities
64.
by stan_kirdey 3y ago
how do one watermarks "hello world" if gpt says "hello world"?
65.
by stan_kirdey 3y ago
now I have to buy Quest3 and trash it, just like Quest2...
66.
by stan_kirdey 3y ago
I've read the title and thought "oh this seems to be a weird name for a footgun. It would totally make sense if it was about timezones". Then I clicked the link.
67.
by stan_kirdey 3y ago
I am being sarcastic. Nobody even uses terms `programmer`, and I've never heard `professional programmer` in any engineering context. The title should state "Software Engineers ...`
68.
by stan_kirdey 3y ago
it is well known fact that most programmers are unprofessional
69.
by stan_kirdey 3y ago
can I get 100$ discount for going slower, please? driving like a gran-pa-pa finally pays off.
70.
by stan_kirdey 3y ago
it is not good news for the only currency in the world that can generate wealth by issuing ... more currency. welp.
71.
by stan_kirdey 3y ago
... "but it will depend on how well of a job is done by OpenAI" ... :D (just being pett)
72.
by stan_kirdey 3y ago
Maybe it is just semantics and they do train GPT-5.1? Or UniversalPT-1 that is multi-modal?
73.
by stan_kirdey 3y ago
lack of analytics offering over stolen data seemed like a missed opportunity too :(
74.
by stan_kirdey 3y ago
Do hackers ask for ransom as a single payment, or there is more of a scalable repeatable business model with recurring revenue? A subscription service seems convenient.
75.
by stan_kirdey 4y ago
Would be great if ChatGPT interface had used token count instead of throwing an error after submission.
76.
by stan_kirdey 4y ago
it is ironic that documentation is lacking when it can be generated with an LLM, using LangChain itself
77.
by stan_kirdey 4y ago
... -> hallucinate your answers to the team mates, as if you are an LLM
78.
by stan_kirdey 4y ago
back to peloton times
79.
by stan_kirdey 4y ago
breaking news: machine learning models including LLMs are fuzzy and make mistakes!
80.
by stan_kirdey 4y ago
sadly, join waitlist button doesn't work on google chrome ._.
81.
by stan_kirdey 4y ago
anti-recruiter, in this economy???
82.
by stan_kirdey 4y ago
their early analysis also showed that "metaverse" is a thing ^_^
83.
by stan_kirdey 4y ago
I guess, voice is not your password
84.
by stan_kirdey 4y ago
Another Deepspeed?
85.
by stan_kirdey 4y ago
amazing read
86.
by stan_kirdey 4y ago
i just wanted to see the good bois
87.
by stan_kirdey 4y ago
nothing is quantum safe until there is quantum compute
88.
Addition of fine-tuning for large language models in Cerebras AI Model Studio(cerebras.net)
3 points by stan_kirdey 4y ago|1 comments
89.
by stan_kirdey 4y ago
Quite cool development, Cerebras make largest GPU accelerators that can fit entire LLM on a single chip. Making their tech available and affordable for fine-tuning can democratize LLM space even further.
90.
by stan_kirdey 4y ago
revolutionary, just like metaverse

More ›