moose
moose copied to clipboard
tf-encrypted
Secure distributed dataflow framework for encrypted machine learning and data processing
By using `AbstractComputation::trace` instead of `trace(AbstractComputation)` we could cache the result in `AbstractComputation`. Similarly for serialized+compiled versions.
> In this section we describe our protocols for performing encrypted computations using the three-party replicated secret sharing scheme. https://github.com/tf-encrypted/moose-whitepaper is there any reason why do not using a four-party...
When running the gRPC choreographer we need to provide the port number twice in order to connect to `comet`. Otherwise things fail silently. For eg: ` comet --identity IP_ADDR:50001 --port...
In the eDSL we have the placement where an argument is coming from: ```python def simple_computation( x: pm.Argument(placement=alice, vtype=pm.TensorType(pm.float64)), y: pm.Argument(placement=bob, vtype=pm.TensorType(pm.float64)), ): ``` which means we might be able...
Concretely, use `println!` instead of `tracing::info!` for results
- [ ] `GrpcMooseRuntime` -> `GrpcRuntime` - [ ] `LocalMooseRuntime` -> `LocalRuntime`
I'd like to work with the team to polish/amend the documentation (link below) that contains a brief summary about Moose (Introduction.md) and a Development Guide that shows how new contributors...
This could for instance be done by specifying values either directly in the .session file or indirectly as filenames. It might be okay to limit the latter to eg .npy...
Metadata
Owner
Metadata
Secure distributed dataflow framework for encrypted machine learning and data processing