KV cache compression
Description: Techniques that reduce the memory footprint of transformer key-value caches while retaining information needed for efficient language-model inference.
Primary topic: Topic Modeling (32%)
OpenAlex ID: keywords/kv-cache-compression
