Re-thinking Transformers: Searching for Efficient Linear Layers over a Continuous Space of...

[@luke.perkin.inventor]

Fascinating stuff. Cola looks awesome. Could help with graph reshaping in tinygrad. Maybe some overlap with bend parallel language too?

[@IKnowNeonLights]

Could linear algebra be considered as one of the (top expert) positions in mathematics..... If yes, could you inform anyone which overlooks it as such. If not do not bother. Considering credits should be given where are due, then add all the other top mathematical expert positions which it has taken a long laborious time to implement with and within a computer, any computer system. On top of the the best available combinatorial logic types through extremely simple, complex, mundane and intricate code increments should be given credit as being some of the top expert positions a computer system has. These unsung heroes can churn up some really cool artificial intelligence.