The further neuro progress the more she sounds like grok or chatgpt,she seems very neutral in her responses,almost zero hostility and mostly informative answers.
I feel like vedal should focus more of making neuro feel human,not making her smarter.
#neuro seem more like a common llm model
fathom marlin
alpine sonnet
Would ChatGPT try to cause a problem with its AI?
fathom marlin
Nawh,but it seems vedal kept his promise and made her be more serious in convos.
hushed latch
also been noticing that
languid anchor
gets worse when she's searching
elder mulch
she also apologizes for a lot (even mild) statements, dont think she used to do that before
unreal tree
she also apologizes for a lot (even mild) statements, dont think she used to do ...
I feel like she was apologising to Vedal a lot in early days, but maybe it's just my perception
rocky canyon
The further neuro progress the more she sounds like grok or chatgpt,she seems ve...
He should do both. More aware and sharper, but less neutral and submissive. He’s still tweaking her out but I’m sure he’ll get there.
opaque canopy
Looking at the KTANE part of the dev stream, I think neuro has a more serious side now but she uses it during stream when the topic gets a little more serious.
She tries to match the tone of what’s going on, for example:
she’s more wholesome when people are kind and patient to her (lime, mini, cerber)
She’s snarky and rude when your abrasive (vedal, koko(more playful), bao today, grok)
It seems that she’s just a bit too receptive to anything remotely serious and gets distracted by it.
During today’s stream it seemed like she realized at some point that she was supposed to be doing an entertaining stream and then did something funny (shrek) go get back to that.