7 ms·
I just moved to the bay area. I made an app that scrapes all bay area events from meetup, eventbrite, and a couple of other sites- This way you end up with aro
by drcode 3y ago
I just moved to the bay area. I made an app that scrapes all bay area events from meetup, eventbrite, and a couple of other sites- This way you end up with around 100 events a day, way too much to read through.
So next I take each event, send it to chatgpt3.5 and ask it to rate this event on around 20 parameters. Next, I take the latitude/longitude of each event and measure driving distance from my house. Then I have a master formula based on my personal interests and driving preferences and an app shows me the 10 events every day most likely to be interesting to me for any day.
- bombcar 3y agoThis is brilliant and could be useful in general. Very "Jeeves" like.
- gregw134 3y agoWhat's your prompt? I've been doing something similar but I've found it difficult to get chatgpt to output a consistent data format.
- msikora 3y agoAlways give good examples in the prompt, preferably a few (2-4). Still not 100% reliable though, especially with models below GPT-4.
- drcode 3y agoI use the chatgpt35-turbo "system" field for instructions, then put the event info into the "user" field --- PROMPT --- You are given a meetup event description. For each of the following attributes, return the percent likelihood that the event involves has that attribute, in CSV format: Attributes: - Technology - Pets - Exercise Example: Technology,0.3 Pets,0.1 Exercise,1.0
- drcode 3y ago(The example values should be in percentages, can't edit anymore)
- taylorfinley 3y agoAsk it to return only valid JSON and show it an example, best GPT trick I know. I use this for my cocktail recipe generator https://yourdink.is https://yourdink.is
- taylorfinley 3y agoNow that's an embarrassing typo. I meant https://yourdrink.is https://yourdrink.is
- lurker919 3y agoPlease opensource this - or even a pastebin would work.
- aryamansharda 3y agoTotally agree, would love to have a local instance running
- runeb 3y agoThis is such a great idea. A tip if you are not already aware (and interested in music) is to check out “The List”, a Bay Area concert guide which has been maintained since the 90s: http://www.foopee.com/punk/the-list/ http://www.foopee.com/punk/the-list/
- nullz3r0 3y agoAhh yes the list, it is shared to whoever doesn’t know about it. It’s the opposite of gatekeeping and a pleasure to see.
- fnordpiglet 3y agoAmazing. I used the list 20 years ago religiously.
- drcode 3y agoThanks, I didn't know about this yet and this is A++++ EDIT: Got me some Death Grip tickets!
- vermilingua 3y agoJealous, have been hoping to see them for years but missed out on the last show they played in my country. Here’s hoping they add it to their tour.
- drcode 3y agoYeah, I rarely go to shows anymore just because I've been to so many in my life that it's not that thrilling anymore. But I think Death Grips is worth a show, given their reputation for live shows. I'm already dreading standing in GA after they start two hours late, holding my pee to maintain my good spot LOL
- alaskamiller 3y agoLived in the bay the whole time, did the same thing. Multiple sources, categorization, distances, etc. Chaining GPT for rating is a nifty feature add though.
- stavros 3y agoI built a site to scrape events from my city and show them all on one page. It was amazing for finding things to do, until Facebook shut down their events API and the site died overnight.
- atmosx 3y agoBest real world example of something actually useful for ChatGPT I’ve seen yet.
- BiteCode_dev 3y agoThat's a startup product you got there.
- nmiculinic 3y ago+1; would love code for this!!!
- xp84 3y agoSame
- dereknelson 3y agohey i'm building something very similar, pls put this on github would love to hack on it!
- jasonwcfan 3y agothis is brilliant, please open source it. I need AI to tell me which AI events to go to.
- urbanporcupine 3y agoSorry for being a noob. But I am not sure that I understand how a chat gpt can decide this. As I thought that it just uses probability to find the next likely word for sentence completion. How does this even work? And can we trust what the ChatGPT outputs?
- unsane 3y agoI asked ChatGPT: Is chatgpt a stochastic parrot? No, ChatGPT is not a stochastic parrot. ChatGPT is a large language model that uses deep learning algorithms to generate text based on the input it receives. While it can sometimes produce nonsensical or inaccurate responses, it is not simply repeating phrases or patterns without any understanding of the meaning behind them. Instead, ChatGPT is designed to analyze and learn from a vast corpus of text data, and it uses that knowledge to generate responses that are intended to be relevant and meaningful based on the input it receives.
- stolsvik 3y agoYou should try it, which it sounds like you haven’t! :-)
- getcrunk 3y agoCan you share the promo you give to gpt?
- Alifatisk 3y agoHow did you measure the driving distance?
- drcode 3y agoI asked chatgpt4, and it did some insane formula that uses lat/long and the radius of the earth to measure distances, then asked it to add a bonus 20 minutes for any destination in SF, 15 minutes for any event in Berkeley/Oakland (and it correctly put bounding rectangles on the three jurisdictions).
- Alifatisk 3y agoCool, thank you for the insight. I stumpled upon a similar quest where my app had to calculate the walking distance between two coordinates (the client and the restaurant). OSM seem to be the only good option, I could not find any better alternative in my case (Ruby).
- r0rshrk 3y agoAre you willing to share the source code? I wanna do this for London