dlcuda
Popular repositories Loading
Repositories
Showing 2 of 2 repositories
- dpgmm Public
- SoftMoE Public
SoftMoE is a differentiable sparse Mixture-of-Experts architecture for large language models. It replaces discrete top-k routing with a continuous soft top-k LapSum relaxation, enabling gradient-based optimization of expert allocation and dynamic compute-efficient routing under a global expert budget.
People
This organization has no public members. You must be a member to see who’s a part of this organization.
Top languages
Loading…
Most used topics
Loading…