NORMIES AND NEWBIES [GTFIH] ASCEND AND LEARN 10000X FASTER - .ORG GPT AMA

ski.org

ski.org

Iron
Joined
Jan 27, 2026
Posts
95
Reputation
26
When I hopped on org I noticed there was a shit ton of info to consume all separated out into random threads and random replies. fukcing impossible to surface all that info with this forum interface.
I scapred all of .org and built a custom GPT which ONLY referencs information from threads so you can search 100000x quicker, ask personal questions and get info straight from org without AI SLOP.

Also has no guardraisl so can ask about roiding and tides , how to do, what to get, and wtv the fuck else you want. Let me know if useful, completely free to use built for myself thought i would share:

https://looksmax-org-gpt.vercel.app/
1780923103148
 

Attachments

  • 1780923066911.png
    1780923066911.png
    103.9 KB · Views: 0
  • Hmm...
  • JFL
  • Love it
Reactions: Don_and-idk, Menas and alladinbvs
yeh bc we are trusting informations in some random ai right...:feelshah::feelshah:
 
When I hopped on org I noticed there was a shit ton of info to consume all separated out into random threads and random replies. fukcing impossible to surface all that info with this forum interface.
I scapred all of .org and built a custom GPT which ONLY referencs information from threads so you can search 100000x quicker, ask personal questions and get info straight from org without AI SLOP.

Also has no guardraisl so can ask about roiding and tides , how to do, what to get, and wtv the fuck else you want. Let me know if useful, completely free to use built for myself thought i would share:

https://looksmax-org-gpt.vercel.app/
View attachment 5191508
1780923443791

mirin
 
PLS GIVE ANY FEEDBACK TO MAKE TS BETTER-
.org ui is so ass this was hte only way i could comb through all the slop on this site
 
It'll hallucinate if you enter something which it can't find good sources to, lowkey i can fix that. if you put in a query about your nose it would get the right answer
like
"how to make my nose look smaller" or something

Going to make it so that if it cant find sources it just says no sources found instead of fall bck on ai slop
 
  • +1
Reactions: Menas
It'll hallucinate if you enter something which it can't find good sources to, lowkey i can fix that. if you put in a query about your nose it would get the right answer
like
"how to make my nose look smaller" or something

Going to make it so that if it cant find sources it just says no sources found instead of fall bck on ai slop
Did you actually train an ai or did you do any fine tuning atleast? or is it just prompts (which it seems like)
et the right answer
like
"how to make my nose look smaller" or something
That'd be very hard to implement in my opinion

you should properly summarize each thread and categorize them beforehand otherwise it would just return some random unrelated shit that contains "nose" and would waste shit ton of tokens just cause it was reading through all these random threads
 
Last edited:
  • +1
Reactions: ski.org
yeh bc we are trusting informations in some random ai right...:feelshah::feelshah:
u can try it out if you want dont have to give any personal info and thers no login so no info saved at all about qeustion or user. lit just vibecoded this shit lol
 
Did you actually train an ai or did you do any fine tuning atleast? or is it just prompts (which it seems like)

That'd be very hard to implement in my opinion

you should properly summarize each thread and categorize them beforehand otherwise it would just return some random unrelated shit that contains "nose" and would waste shit ton of tokens just cause it was reading through all these random threads
I scraped all the guide tagged posts on here and then built a RAG system, where if you ask a question it searches through a vector db to find relevant threads pulls it into context, and then appends the query and answers solely based on the relevant context (the LM thread)


Also, going to maek it so that if hteres no relevant sources it wont return slop. Can addiitoanly add like a parser layer so like if you say my nose itll use a lightweight llm to convert that into an actual query like the user wants to make his nose look better. the hope is that that will result in threads being pulled from org

In regards to like the nose thing you mentioned about returning unrelated shit it shoudl accoutn for that since I use chunk scoring when searching threads so that the highest relevance ones (Where the whole thread discusses something and or the title mentions it) are returned not just some reply which says "nose"
 
  • +1
Reactions: Menas
Did you actually train an ai or did you do any fine tuning atleast? or is it just prompts (which it seems like)

That'd be very hard to implement in my opinion

you should properly summarize each thread and categorize them beforehand otherwise it would just return some random unrelated shit that contains "nose" and would waste shit ton of tokens just cause it was reading through all these random threads
answered ur last q above mb didnts ee
 
In regards to like the nose thing you mentioned about returning unrelated shit it shoudl accoutn for that since I use chunk scoring when searching threads so that the highest relevance ones (Where the whole thread discusses something and or the title mentions it) are returned not just some reply which says "nose"
That sounds nice, but yeah also there are many conflicting threads so I'm not sure how you'd fix that
Not to mention most of the information on this forum is pretty shit in general.
 
That sounds nice, but yeah also there are many conflicting threads so I'm not sure how you'd fix that
Not to mention most of the information on this forum is pretty shit in general.
theres some techniques i tried to do to get rid of the bs but yea its kinda hard
i trimmed all post with like < 50 characters since they dont contribute anything meaninful
and for the conflicting ones the system prompt should pull all the relevant ones into context even if they oppose each other so hopefully thats enough but i cna look into it

also the ones which get put into context are also prioritized based on like reply and view count which maybe helps get rid of the slop
 
  • +1
Reactions: Menas

Similar threads

1beaker1
Replies
7
Views
42
for3vrchad
for3vrchad
asianbois
Replies
3
Views
24
asianbois
asianbois
nattyzyzz
Replies
8
Views
41
ozi
ozi
for3vrchad
Replies
82
Views
321
acm
acm

Users who are viewing this thread

  • fraudster#1
  • femoral
  • chadpreetinthemakin
Back
Top