we're way too close to agi to be putting any thought into if muon is equivalent to shampoo
@typedfemale argues the machine learning community is too close to AGI to debate Muon versus Shampoo optimizers
David Holz joked that neural networks need beautiful hair
Many users criticize arguments that AGI renders elegant theory debates irrelevant or shifts academia toward practical work, viewing the change as embarrassing and dismissive of important priorities.
Most Activity
Many intellectually cute problems of learning are getting dumped, because the point is making an intelligence and it seems you can achieve this with dumb suboptimal parts. academics hardest hit
we're way too close to agi to be putting any thought into if muon is equivalent to shampoo
@typedfemale but the networks need beautiful hair 🥺
we're way too close to agi to be putting any thought into if muon is equivalent to shampoo
@typedfemale but the rage! it must be baited
we're way too close to agi to be putting any thought into if muon is equivalent to shampoo

@xXshaurizardXx current discourse

@typedfemale what makes you feel this way
@typedfemale It's just a distraction from SOAP
we're way too close to agi to be putting any thought into if muon is equivalent to shampoo

@typedfemale @xXshaurizardXx discourses don’t appear from thin air, they are fed
@typedfemale Preach
we're way too close to agi to be putting any thought into if muon is equivalent to shampoo

@typedfemale This reminded me of when we tried out scheduler free Adam a few years ago. That was proper fun

@typedfemale Just use Adam and let AGI figure this out

@typedfemale right lol, this is where we stop prioritizing the important stuff

@teortaxesTex 学术圈真的要变天了这波确实尴尬

@teortaxesTex 思路挺刁钻的,这波操作牛

@teortaxesTex 这确实戳中了痛点啊

@teortaxesTex 确实,堆砌“聪明”组件不等于真智慧