PyTorch Outreach Computing Support
The PyTorch outreach computing support of the Utility network solves the problem that Solidity contracts cannot directly embed PyTorch code based on Python. And the Utility network al- lows efficient collaborative computing between on- chain contracts and off-chain PyTorch code through rollup technology [31], thus enabling the demon- stration of service capabilities for AI.
In the Utility network, training data and func- tion inputs are transmitted to the miner side via rollup technology, combined with Layer1 contrac- tual conventions. This structure allows miners to perform extended computations or AI model train- ing and inference in an off-chain environment. Af- ter the computation is completed, the results are transmitted back to the chain via rollup for interac- tion and validation with the on-chain contract.
By introducing PyTorch outreach computing support, the Utility network enables a seamless con- nection between on-chain and off-chain. This hy- brid computing model enables the Utility Network to leverage existing AI frameworks, such as Py- Torch, to further improve computational efficiency and performance. In addition, this approach brings greater scope for innovation in AI, supporting re- searchers and developers to develop more complex and powerful AI applications in the Utility network.
Last updated