Everyone, thank you so much for all your support and helping me reach 15k subs!!! I really can't believe we've made it this far. I was able to make extra time to push out another video this week about DoorDash. I hope that you enjoy and can learn something new from this video!
Appreciate the feedback! And definitely will be making more case study videos. I hope that you will tune in the next video that I'm currently working on :)
I use a combination of motion canvas and davinci resolve. Hmm the difficulty of the animation really depends on the scene which can range from super easy to quite difficult. I would definitely consider doing a tutorial on how I make my animations.
It's actually more probable they don't use nearest neighbour search, but something more overkill, like HNSW indexing. Unless you know for sure, then I'm sorry.
@@MrFram yeah, sure, my mind somehow went directly to KNN, because that's what I used to use. But he says in the video "this is done using the approximate nearest neighbour," which is an algorithm, not a problem model. My understanding is things like HNSW are indexes for speeding up ANN.
@@nexovec KNN is an algorithm for regression, not a search algorithm. It **uses** nearest neighbor search, but not implements it, so it has to rely an another algorithm to actually find the neighbors. You can use HNSW or LSH to do that. "Approximate Nearest Neighbors" just means your nearest neighbors don't have to be 100% correct, it's also a problem setting (the problem of approximating nearest neighbor search). LSH and HNSW can also be used for it and will be faster than using them for exact nearest neighbors, and the naive algorithm is always exact.
![[LIVE] : ONE ลุมพินี 88 | คู่เอก "ป้อมเพชร vs อัสลามจอน"](http://i.ytimg.com/vi/1ZqTVVbMgbo/mqdefault.jpg)
Everyone, thank you so much for all your support and helping me reach 15k subs!!! I really can't believe we've made it this far.
I was able to make extra time to push out another video this week about DoorDash. I hope that you enjoy and can learn something new from this video!
This is a really good use case of LLM’s.
agreed!
great content! Concise, appealing and more importantly making complex subjects easier to understand. Amazing job and thank you
Thank you appreciate it!
Congrats on 15k!!
Thank you so much!!! Appreciate your support!
3:54 Using non deterministic methods to label products as organic or non organic or anything else is a law suite waiting to happen.
yeah would have been nice to know the accuracy of this pipeline
Really interesting to hear these cases, great work and nice animations!
Thank you!
The editing and animation in this video is next level!
Thank you so much!
This type of content will bring you the most viewers imo, hope you’ll focus on it. Great job!
Appreciate the feedback! And definitely will be making more case study videos. I hope that you will tune in the next video that I'm currently working on :)
Also, which motion canvas do you use?
motioncanvas.io/
@@kikisbytes so I will have to learn JavaScript for creating videos like yours :( I only know python. I also own after effects
@@crazyparrot2786 you can do it!!! Once you know one programming language, it becomes easier to pick up another.
@@kikisbytes okay, I'll try my best to learn this animation and I'll reply to your comment after that.
How do you edit videos? Those animations look sick, are they easy to make? Can you launch a tutorial for this?
I use a combination of motion canvas and davinci resolve. Hmm the difficulty of the animation really depends on the scene which can range from super easy to quite difficult. I would definitely consider doing a tutorial on how I make my animations.
@@kikisbytes thanks! I'm waiting
It's actually more probable they don't use nearest neighbour search, but something more overkill, like HNSW indexing. Unless you know for sure, then I'm sorry.
You're confused. Nearest-neighbors is a problem, while HNSW is an algorithm. You use HNSW to perform nearest-neighbor queries efficiently.
@@MrFram yeah, sure, my mind somehow went directly to KNN, because that's what I used to use.
But he says in the video "this is done using the approximate nearest neighbour," which is an algorithm, not a problem model. My understanding is things like HNSW are indexes for speeding up ANN.
@@nexovec KNN is an algorithm for regression, not a search algorithm. It **uses** nearest neighbor search, but not implements it, so it has to rely an another algorithm to actually find the neighbors. You can use HNSW or LSH to do that.
"Approximate Nearest Neighbors" just means your nearest neighbors don't have to be 100% correct, it's also a problem setting (the problem of approximating nearest neighbor search). LSH and HNSW can also be used for it and will be faster than using them for exact nearest neighbors, and the naive algorithm is always exact.
@@MrFram Oh, sure, that sounds right. Thanks.
@@MrFram That's actually really well explained.
quality, subbed
Thank you!
Great explaination.
Thank you!
@@kikisbytes best of luck bro and thanks for the reply.
I have a price comparison site. I face a similar problem
that's awesome!! Did you end up doing something similar to doordash?
how do you come to know about all this?
It's from an article from their engineering blog. I have the link in the description if you're curious!
@@kikisbytes i see.. great work btw n keep it up!
@@programs37 thank you I really appreciate that!
ok so smash GPT until it works, got it
hahaha heck ya
The automated versus manual process animation was 🔥
hahha thank you!!!