@felix_ved: What does the Top P setting do with LLMs? Instead of just adjusting probabilities like Temperature, Top-P intelligently filters the list of possible words, creating a 'nucleus' of the best candidates #llm #aiengineer #softwareengineers #developer #ia
So increase temp. if you want randomness and instead of lowering temperature to get more "certain/correct" answer lower the top p? Can you do entire video on some real world scenarios where you should use one approach over the other or use them together in some way.
2025-08-01 16:28:44
4
Alphie :
We need more content. I learned this at uni but I’m glad it’s here too, more people would appreciate this
2025-08-02 23:20:12
0
drozd :
Thanks for sharing, very nice visualization!
2025-08-01 21:48:25
0
Bob James :
Yes 👏👏 more please and maths too please 🙏
2025-08-01 21:30:43
0
0xl16 :
Love how you simply explain it
2025-08-01 16:30:24
4
Apple User936445 :
Sounds like it would make sense to have a multi agent system adjust these parameters based on the problem…. 👀
2025-08-01 19:16:13
0
Water, Harry! 🪄💦 :
That’s interesting and something I didn’t know, but I still don’t understand how we could meaningfully know what value to choose? Do you have any examples of when we’d want a low verses high top-p?
2025-08-01 19:21:00
0
Gordon Ramsay :
This is helpful thank you
2025-08-01 23:47:21
0
To see more videos from user @felix_ved, please go to the Tikwm
homepage.