moose
moose copied to clipboard
Secure distributed dataflow framework for encrypted machine learning and data processing
Update kernels accordingly.
Implement Log & Log2 ops with fixedpoint tensors on host placements (leftover from #812 and #830)
Address the TODO in the values: https://github.com/tf-encrypted/runtime/blob/main/moose/src/computation.rs#L328 ``` // TODO promote below to match other values Bit, Float32, Float64, Ring64, Ring128, Fixed, ```
Models (as introduced in #742) are of primary for runtime benchmarking (as introduced in #738). This issue is about adding implementations of benchmarks that explicit test models we add to...
- [x] split into `fixedpoint/` - [ ] attempt to simplify `PolynomialEval` implementations
- [ ] use `operators!` to define op struct, and trait? (after https://github.com/tf-encrypted/runtime/issues/480) - [ ] make kernel and signature types into another file?
`HostRingTensor` is essentially just `HostTensor` and we should probably simply make it so. Alternatively make `HostRingTensor` an alias. Note that it would likely be good to have a quick team...
All three are structurally identical to ```rust pub struct RepFixedTensor { pub tensor: RepRingT, pub fractional_precision: u32, pub integral_precision: u32, } ``` so maybe we should simply have a single...
There are identical structures so maybe a single `BitArray` would be better. Note that a team discussion is needed before implementing this.
A recent error message that I encountered when running a compiled computation was missing information such as on which op did the error occur. By adding some more information to...