How are you different to Gensyn?

[DeveloperAlly]

https://protocol.ai/blog/meet-gensyn-the-machine-learning-compute-network/

[DeveloperAlly]

Gensyn is trying to solve the "train a multi-billion paramater model" problem with decentralised compute. (and kudos to them - this is HARD!! - Gensyn if you're reading this, we'd love to collab anytime!)

Lilypad has taken the approach of enabling off-chain distributed compute with on-chain guarantees for "can run on one gpu" jobs, like inference / generative AI, but also including storage primitives (ie. can train on data stored on ipfs).

We think that

The models you can run on one GPU / hardware device are getting more and more powerful while requiring less and less hardware to run
This is a (possible experiemental) stepping stone to solving this harder problem of "sharding" (to borrow a web3 term) the job to run on multiple devices and bringing the results back together.
There are a vast array of use cases in the world that can leverage our type of compute which will enable a more accessible and efficient marketplace for compute jobs, and tbh verifiable compute is also hard!

Luckily we have a crack team and some really great skills across both AI, distributed compute & web3!