Local AI Glossary

What is Multi-head Latent Attention (MLA)?

Source: paperswithcode/MLA

DeepSeek-V2's attention variant that compresses the KV cache into a low-dimensional latent space, reducing memory usage by 5-10× compared to standard MHA while preserving quality.

How Multi-head Latent Attention (MLA) fits in the local AI stack

Multi-head Latent Attention (MLA) is part of the broader local AI ecosystem. See the related guides, comparisons, and tools below to see how it applies in practice.

See also

Related terms