factual
Optimal batch size is about 300 times sparsity
Setting weight-fetch time equal to weight-multiply time yields a hardware constant (FLOPs/byte ratio) of roughly 300 across recent GPUs, so the optimal batch size is approximately 300 times the model's sparsity ratio—e.g., DeepSeek activating 32 of 256 experts gives a sparsity of 8, implying a batch around 2,000-3,000 tokens; this depends only on sparsity, not model scale.
factualpending
Speaker
Reiner PopeEvidence Quote
“batch size needs to be bigger than approximately 300 times sparsity”
Source
Dwarkesh Patel Podcast - Reiner Pope – The math behind how LLMs are trained and served— Dwarkesh Patel PodcastCreated: 6/13/2026, 3:34:09 AM
Your notes
Loading notes...