Farcaster Devs

Trying to build a frame that uses an LLM, and I'm realizing the 5 second timeout is going to be pretty prohibitive for what's possible

Neovim enjoyer, "a gift to the culture" at /thirdweb

Building @warpcast and @farcaster, new dad, like making food

@jpfraneto.eth also suggested this! I think its the best option so far

you could do it with multiple actions 

one to initiate and a follow up to fetch the results

Gotcha, I'll work on some options. Also a lot of funky behavior when using advanced framework features like loading pages on Next -- lots to be explored

Side note, the frame preview tool currently disables all inputs so I can't test 😕

I'm eth newbie..just learning
(working on @farcaster)

terminally.online

Yeah, async patterns are still underexplored, hopefully this constraint leads to some creativity. For some things you might save a claim check server side and let the user come back for it or notify when it's ready. (Could do this with a reply like the cookie bot).

MrKevinOConnell.github

prev @neynar, @hypeshot

don't take my casts seriously please (most of them at least)

My current app responds in exactly ~5 seconds when slow, so its 50/50 whether or not the error hits. A faster model definitely helps, I'm just concerned of these scenarios where you can't guarantee the frame will work every time

Dad, geek, conversation liquidity provider. vrypan.net (home), /weatherxm (work), @l--o--l (fun). Also https://paragraph.xyz/@purplesubmarine

Yes I think so, I was getting errors last night that it timed out after 5 seconds but I'm not finding it anywhere in the specs

Is this a 5sec limit to the response time?

EigenDA DevRel - onchain thinker and sharer of information
~
Prev @ intuition.systems

https://warpcast.com/gregfromstl/0x20572d71

Only if you're relying on a generic GPT-4 API! Try running mistral locally and observe the possibilities