JuliaSmoothOptimizers / JuliaSmoothOptimizers/MultiPrecisionR2
Use gradient norm for model decrease
Open
enhancement
- Dominant language
- Julia
- Stars
- 1
- Forks
- 3
- PR merge metrics
- No merged PRs in 30d
Description
Gradient norm can be reused to compute model decrease in `ComputeModelDecrease()`. Might have to add an additional precision for gradient norm computation FP format.
Contributor guide
Assessment
This issue has not been assessed yet.