Thank you for featuring my request! That's so cool of you! The AI's responses aren't there yet but who knows what it can do in a few years. I look forward to the day when I give my AI shell (complex) commands and get back the correct ones (probably that I don't know of and how it works but it works haha). Great video as always!
I'd love to see it cutting out the typing and going directly from speech to command. I expect that'd be easy with all the speech-to-text services available.
Hi, I met this erroe when run ./aish 0.0.5: Using OpenAI ~> show me my IP. 🧠 Thinking... ... values[i] = one_value.encode('latin-1') UnicodeEncodeError: 'latin-1' codec can't encode character '\u2018' in position 7: ordinal not in range(256) What's wrong to me? My machine: m1 macmini, osx mojave & monterey(same errors) Thanks in advance ~.
I was impressed by the regex and openssl examples, it's at least way faster than me at providing an answer! - I guess it's going through the manpages like "Nr. 5" did in "Short Circuit (1986)"
There are some tutorials online for running GPT-2 locally. Once that's up and running, you could adapt an example program to the aish code. I might give it a shot if more people ask.
Human: *Joking* "Create killer robot programing." AI: *Processing* "I have finished the task." Human: "Alright, let's have a look at it." AI: "You do not have permission to view that file." Human: ". . . Uh-oh."
"it knows a few regexs" REALLY? this is like saying a Lamborghini is able to move forwards when you step on the gas... gpt3 can generate regex for any example string/strings you give it, or even generate one off a description of a string or strings.
Thank you for featuring my request! That's so cool of you! The AI's responses aren't there yet but who knows what it can do in a few years. I look forward to the day when I give my AI shell (complex) commands and get back the correct ones (probably that I don't know of and how it works but it works haha). Great video as always!
I'd love to see it cutting out the typing and going directly from speech to command. I expect that'd be easy with all the speech-to-text services available.
Hi, I met this erroe when run ./aish 0.0.5:
Using OpenAI
~> show me my IP.
🧠 Thinking...
...
values[i] = one_value.encode('latin-1')
UnicodeEncodeError: 'latin-1' codec can't encode character '\u2018' in position 7: ordinal not in range(256)
What's wrong to me?
My machine: m1 macmini, osx mojave & monterey(same errors)
Thanks in advance ~.
I was impressed by the regex and openssl examples, it's at least way faster than me at providing an answer! - I guess it's going through the manpages like "Nr. 5" did in "Short Circuit (1986)"
This follow-up is awesome! How would you recommend hooking up your code to a locally-run model?
There are some tutorials online for running GPT-2 locally. Once that's up and running, you could adapt an example program to the aish code. I might give it a shot if more people ask.
@@riveducha I'd be into seeing that. +1
Great follow-up!
Human: *Joking* "Create killer robot programing."
AI: *Processing* "I have finished the task."
Human: "Alright, let's have a look at it."
AI: "You do not have permission to view that file."
Human: ". . . Uh-oh."
This needs an update
So cool .
"it knows a few regexs" REALLY? this is like saying a Lamborghini is able to move forwards when you step on the gas... gpt3 can generate regex for any example string/strings you give it, or even generate one off a description of a string or strings.
Yeah it can generate a regex for any example. Would be even cooler if the regex is correct!
@@riveducha it's correct for me... 🤷♂
"Open Leaflet"
Still try it with gpt-J 6B
well that was neat