It's stable for me so far. Are there any problematic prompts?
It's been helping me with some code and I made the prompt below to double check.
A Reddit comment in the style of Edgar Allen Poe:
Good sir (or madam, for the veil of anonymity shrouds us all), I find myself utterly disquieted by the omnipresent algorithm that governs this digital realm. Is it not akin to a raven, perched upon a forgotten server, croaking pronouncements that dictate the content we see?
A chill wind whispers of a bygone era, when the written word flowed freely, unfettered by the unseen hand of calculated manipulation. Now, are we but phantoms in a spectral dance, forever doomed to see only that which the algorithm deems fit for our consumption?
Let us raise a collective glass, not of merriment, but of mournful acceptance, for perhaps, in this digital necropolis, true freedom of expression lies buried six feet deep.
Hanlon's razor: (A) An unimaginably large coordination to attack the most computing power on earth temporarily out of unspecified malice (B) It's finals week and a cascading tidal wave of millions of kids are texting each other that chatgpt is down and they need an alternative
I'd love to use local models, but seems like most of the easy to use software out there (LM Studio, Backyard AI, koboldcpp) doesn't really play all that nicely with my Intel Arc GPU and it's painfully slow on my Ryzen 5 4500. Even my M1 MacBook isn't that fast at generating text with even 7B models.
But even if that worked, I'd still have the problem that IDEs and whatever else I have open already eats most of the 32 GB of RAM my desktop PC has. Whereas if I ran a small code model on the MacBook and connected to it through my PC, it'd still probably be too slow for autocomplete, when compared to GitHub Copilot and less accurate than ChatGPT or Phind for most stuff.
This may be tangential, but situations like this just make me want to try to spin up my own LLM. But I suspect I won't benefit from economics of scale, right?
I wonder if the OpenAI issue has anything to do with Apple doing some large performance/integration testing ahead of potential WWDC announcement on Monday
Okay, I'm struck that 90% of the comments in a post about Claude being unavailable are complaints that ChatGPT is unavailable. I guess that probably corresponds to the companies respective share of the market?
Anyway, my tale of woe and recovery re the great centralized chatbot outage was turning to use MLCChat on my phone running a Mistral 7b model and being happily surprised it could actually summarize and answer questions about (though slowly) an article I didn't want to read in its entirety. Although, I didn't quite trust it, so I read the article anyway. The summary wasn't too bad, though. Good to know.
"This website is using a security service to protect itself from online attacks. The action you just performed triggered the security solution. There are several actions that could trigger this block including submitting a certain word or phrase, a SQL command or malformed data."
37 comments
[ 3.0 ms ] story [ 85.4 ms ] threadIt's been helping me with some code and I made the prompt below to double check.
A Reddit comment in the style of Edgar Allen Poe:
Good sir (or madam, for the veil of anonymity shrouds us all), I find myself utterly disquieted by the omnipresent algorithm that governs this digital realm. Is it not akin to a raven, perched upon a forgotten server, croaking pronouncements that dictate the content we see?
A chill wind whispers of a bygone era, when the written word flowed freely, unfettered by the unseen hand of calculated manipulation. Now, are we but phantoms in a spectral dance, forever doomed to see only that which the algorithm deems fit for our consumption?
Let us raise a collective glass, not of merriment, but of mournful acceptance, for perhaps, in this digital necropolis, true freedom of expression lies buried six feet deep.
The singularity is here.-
... by whom?
(Cue spooky music)
2024-06-04 13:51 - ChatGPT enabled a feature that double-checks responses with Perplexity
2024-06-04 15:04 - Claude enabled a feature that double-checks responses with ChatGPT
2024-06-04 17:42 - The internet ran out of bandwidth as LLMs starts going around in infinitive loops, double-checking double-checked answers
https://status.openai.com/incidents/rs49lmsh6xxw
Do they make calls to each other?
The OpenAI API meanwhile is still up :)
I wonder if llama.cpp with SYCL could help, will have to try it out: https://github.com/ggerganov/llama.cpp/blob/master/README-sy...
But even if that worked, I'd still have the problem that IDEs and whatever else I have open already eats most of the 32 GB of RAM my desktop PC has. Whereas if I ran a small code model on the MacBook and connected to it through my PC, it'd still probably be too slow for autocomplete, when compared to GitHub Copilot and less accurate than ChatGPT or Phind for most stuff.
Anyway, my tale of woe and recovery re the great centralized chatbot outage was turning to use MLCChat on my phone running a Mistral 7b model and being happily surprised it could actually summarize and answer questions about (though slowly) an article I didn't want to read in its entirety. Although, I didn't quite trust it, so I read the article anyway. The summary wasn't too bad, though. Good to know.
This is what I'm getting from Claude.