Eggroll: Novel general-purpose machine learning algorithm provides 100x speed

https://eshyperscale.github.io/

by felineflock • 8 months ago

30 4 8 months ago

eshyperscale.github.io

bdbdbdb• 8 months ago

What does this actually mean for LLMs? Cheaper training?

MarkusQ• 8 months ago

Yes. Provided it works as well as they claim.

Not only cheaper, but (since in this case money ≈ hardware-cost × time), faster. They claim that training time can even approach inference time:

EGGROLL's efficiency results in a hundredfold increase in training throughput for billion-parameter models at large population sizes, nearly reaching the throughput of pure batch inference

free_bip• 8 months ago

Their technique does not claim to compete with gradient descent - it's competition for techniques like Proximal Policy Optimization, so it's more suited for things like creating a reasoning model out of an existing pre-trained model.

orderone_ai• 8 months ago

It really depends on how it scales. If it can scale to LLM sizes via this training method (Which is a big if), then it could mean fundamentally overturning the transformer architecture and replacing it with RNNs in the most optimistic case.

But if not, it could mean as little as some LLM-adjacent tools like vec2text get reworked into RNNs. Or some interesting fine-tuning at least.

Eggroll: Novel general-purpose machine learning algorithm provides 100x speed

Related Stories