Right. But ... this would limit you to either extremely small models or extremely large FPGA's, yes? If there's a simple machine learning task that requires a sub microsecond latency I can see the point but otherwise??
Animats
This guy will be hired by a high-frequency trading firm, and the next time we hear about him, he will have a net worth in 9 figures.
babelfish
Archive link, as it looks like the original post was taken down: https://web.archive.org/web/20260609200156/https://aarushgup...
mikeayles
So for people wondering if it can be used to accelerate LLM inference, sadly not. I've been trying to hit 100,000tokens/s with a 3.28m dumb model, and even this is an order of magnitude too large to benefit. It appears to be focussed more on latency, than throughput. Happy to be corrected?
tomrod
Happy to hear that KANs continue to find solid footing.
cwmoore
took long enough
Lerc
Has there been much exploration on how much benefit comes from precision in activation functions in KANs? There's a little niggle in the back of my head that maybe 90% of the benefit of KANs can be gained from a quite small variety of function shapes. Combined with input weighting, I almost feel you could have a representation that scales from a standard relu perceptron though KANs to something with weighted inputs and fancy weighted activation functions. Mark that out in 2d with axes of input weight precision and activation weight precision, you could perhaps do sweeps to find the best accuracy per parameter bit, or accuracy/speed, or some sweet spot that has a nice balance of operating speed, accuracy, and model size.
Cadwhisker
If you want to experiment with KANs yourself in a non-FPGA environment, there's a GitHub repo here: https://github.com/KindXiaoming/pykan HN comments page on that is here: https://news.ycombinator.com/item?id=40219205
semessier
and where is the Transformer library ;)
DeathArrow
I know enough to understand this is interesting but sadly I don't know enough to understand how it works.
woggy
I love the name 'Kolmogorov'
Related Discussions
Found 5 related stories in 90.1ms across 10,002 title embeddings via pgvector HNSW
Discussion Highlights (11 comments)
RantyDave
Right. But ... this would limit you to either extremely small models or extremely large FPGA's, yes? If there's a simple machine learning task that requires a sub microsecond latency I can see the point but otherwise??
Animats
This guy will be hired by a high-frequency trading firm, and the next time we hear about him, he will have a net worth in 9 figures.
babelfish
Archive link, as it looks like the original post was taken down: https://web.archive.org/web/20260609200156/https://aarushgup...
mikeayles
So for people wondering if it can be used to accelerate LLM inference, sadly not. I've been trying to hit 100,000tokens/s with a 3.28m dumb model, and even this is an order of magnitude too large to benefit. It appears to be focussed more on latency, than throughput. Happy to be corrected?
tomrod
Happy to hear that KANs continue to find solid footing.
cwmoore
took long enough
Lerc
Has there been much exploration on how much benefit comes from precision in activation functions in KANs? There's a little niggle in the back of my head that maybe 90% of the benefit of KANs can be gained from a quite small variety of function shapes. Combined with input weighting, I almost feel you could have a representation that scales from a standard relu perceptron though KANs to something with weighted inputs and fancy weighted activation functions. Mark that out in 2d with axes of input weight precision and activation weight precision, you could perhaps do sweeps to find the best accuracy per parameter bit, or accuracy/speed, or some sweet spot that has a nice balance of operating speed, accuracy, and model size.
Cadwhisker
If you want to experiment with KANs yourself in a non-FPGA environment, there's a GitHub repo here: https://github.com/KindXiaoming/pykan HN comments page on that is here: https://news.ycombinator.com/item?id=40219205
semessier
and where is the Transformer library ;)
DeathArrow
I know enough to understand this is interesting but sadly I don't know enough to understand how it works.
woggy
I love the name 'Kolmogorov'