exactly! That's what I intend to try. I've already managed to set up a login with hive keychain in that separate project. Shouldn't be ttoO dofficult to tie this to some ssort of leaderboard. But we'll see how far I can get. I ran out of compute ttokens on my claude subscripton. It renews in 1 hou
r
I tried with the 8b version running it locally, but it's not very nice. Currently installing the 70b version of llama3 on my system - 39 gigabytes. Very excited to see if it will even run
The nice thing about running them locally is that I can add as much context as I want - including the entire codebase for the application I'm building - as opposed to on groq
Yep. I tried using both GPT4o, Llama3.1 70b and Google Gemini Premium yesterday to continue developing the code. But none of them compared to Claude3.5. The closest I got to success was with Llama3.1, but that was still miles behind claude- in this particular case that is.
Using Claude, I managed to progress the development of the game significantly. Going to continue this afternoon with refreshed tokens. I'm quite excited about it
See told you that you could do it.
Here is the next question. If there a way a simple game like that could be tied to Leo?
exactly! That's what I intend to try. I've already managed to set up a login with hive keychain in that separate project. Shouldn't be ttoO dofficult to tie this to some ssort of leaderboard. But we'll see how far I can get. I ran out of compute ttokens on my claude subscripton. It renews in 1 hou
r
Can Llama3.1 help? I am not sure how we can access some of the deflationary aspects of things.
haha it will not run on my system
I do not have 16 GPUs, nor 8 lol
LOL they are getting larger as time goes by.
I tried with the 8b version running it locally, but it's not very nice. Currently installing the 70b version of llama3 on my system - 39 gigabytes. Very excited to see if it will even run
The nice thing about running them locally is that I can add as much context as I want - including the entire codebase for the application I'm building - as opposed to on groq
Guess that idea went out the window based upon some of your other threads. Too big. LOL
Words I have not heard.
Yep. I tried using both GPT4o, Llama3.1 70b and Google Gemini Premium yesterday to continue developing the code. But none of them compared to Claude3.5. The closest I got to success was with Llama3.1, but that was still miles behind claude- in this particular case that is.
Using Claude, I managed to progress the development of the game significantly. Going to continue this afternoon with refreshed tokens. I'm quite excited about it
This was my starting point: