1 dépôt
Tools that compute total inference FLOPs for batches processed through activated experts and gating networks.
Distinct from Inference Batching: Distinct from Inference Batching: focuses on FLOPs estimation for MoE architectures, not general request grouping.
Explore 1 awesome GitHub repository matching development tools & productivity · MoE Batch Cost Calculators. Refine with filters or upvote what's useful.
Calculates inference FLOPs for batches processed through MoE expert and gating networks.