Google Brain Unveils Mixture-of-Depths Architecture
Google Brain's new Mixture-of-Depths transformer architecture dynamically allocates compute per token, cutting inference…
🏷️ inference-efficiency
1 articles about 'inference-efficiency'
1