311–320 of 328 posts
Earlier quoted context omitted.
Openrouter is not counting tokens used by Kilo or Cline. They have own endpoints.
Yet if you go to the actual model’s page: https://openrouter.ai/x-ai/grok-code-fast-1 Cline and Kilo code are in the top 3. So how does that work? It’s considerably cheaper than competing models like 2.5 flash, though. So its not that surprising
Earlier quoted context omitted.
> it isn't tuned much for political correctness It was tuned to be edgy and annoying though (I mean his general style of speech not necessarily the content).
Nothing in AI is more edgy and annoying than beginning every response with a mandatory glazing, like ChatGPT. “That’s a really insightful question, and shows that you really understand the subject!”
Earlier quoted context omitted.
Here's an on topic question: all the frontier model companies "promise" that they wont store and train on your api use if you pay for it. Who do you trust? I for sure will absolutely assume grok will just use the data I submit to train in perpetuity. Thats a scary thing for me and if anyone else does anything thats real work this should be great cause for worry if they wish to use grok.
Do you really think Google isn't logging all our prompts?
OpenAI will go to zero unless it agrees to be acquired because they're messing with public company stock valuations using funky purchase orders leaving those public companies no choice but to cancel their credit (at least unless they get a "government backstop" that they say they don't want or need). Those who compete with OpenAI will also "take a hit" if/when that happens, so they would be wise to be looking to make…
Clarifying, because there's no way a company (public or private) is going to reduce the credit line of a major customer until it's obvious that the orders "aren't real" But if Wall Street realizes it before they do, they can lose control of their business too. This is not quite Enron or WorldCom/MFS, but it's a very similar storm on the horizon. (BTW, ever wonder why Sprint never could remain airborne and eventually was merged with TeenMobile? It's because they overspent on CapX trying to keep up with the fraud at Worldcom and could never dig out to actually use all that spectrum. Likewise, we are still dealing with the fallout of the Enron collapse on the US domestic energy grid a quarter century later.)
Earlier quoted context omitted.
[flagged]
Oh the hubris.
This is quote seared into my head because my father says anything that disagrees with his conspiracies, it is a liberal bias. If I say “ivermectin doesn’t cure cancer”, that’s my liberal bias. “Climate change is not a hoax by the you-know-who’s to control the world” == liberal bias. “Bigfoot only exists in our imagination”… liberal bias (I’m not joking on any of these whatsoever).
So I’ve been saying this in my head and out loud to him for a looooong time.
Earlier quoted context omitted.
[flagged]
Are you implying that avoiding use of services controlled by a fascistoid oligarch is cope?
If we are gonna avoid using the services by an oligarch, then just say that. No need to go through the 5 steps of denial.
I hope we are not against being honest.... but you can see a bunch of comments here going through those steps already.
What matter is not context or the recod token/s you get. But the quality for the model. And it seem Grok pushing the wrong metrics again, after launching fast.
I thought the number of tokens per second doesn't matter until I used Grok Code Fast. I realized that it makes a huge difference. If it take more than 30s to run, I lose focus, and look at something else. I end up being a lot less productive. It also opens up the possibility to automate a lot more simple tasks. I would def recommend people try fast models
Earlier quoted context omitted.
I think they would want a more optimized regex. Like a long list of swears, merged down into one pattern separated by tunnel characters, and with all common prefixes / suffixes combined for each group. That takes more than just replacing one word. Something like the output of the list-to-tree rust crate.
Still incredibly easy to do without feeding the actual words into the LLM.
Earlier quoted context omitted.
I thought the number of tokens per second doesn't matter until I used Grok Code Fast. I realized that it makes a huge difference. If it take more than 30s to run, I lose focus, and look at something else. I end up being a lot less productive. It also opens up the possibility to automate a lot more simple tasks. I would def recommend people try fast models
If you are single tasking, speed matters to an extent. You need to still be able to read/skim the output and evaluate its quality. The productive people I know use git worktrees and are multi-tasking. The optimal workflow is when you can supply it one or more commands[1] that the model can run to validate/get feedback on its own. Think of it like RLHF for the LLM, they are getting feedback albeit not from you, which…
It's a shame that the top comments are focusing more on Elon Musk, his personality and politics rather than the quality of the model per se. Speaking about Elon, regardless of what you think of him, he really does get things done, despite naysayers -- SpaceX, Tesla, Neuralink and even get Trump elected ( despite subsequent fallout) etc. Even Twitter is finding a second life by becoming a haven for the free speech adv…
Is a billionarie getting a politician elected - even by promising payment to voters (that is, buying votes) - something positive?
The US is supposed to be a democracy, it's the people that get politicians elected, not billionaries