Jamba

Appears in 1 paper

An LLM from AI21 Labs (2024) that alternates Mamba and Attention blocks in its architecture.

As used in Paper 21 — Mamba: Linear-Time Sequence Modeling with Selective State Spaces →

An LLM from AI21 Labs (2024) that alternates Mamba and Attention blocks in its architecture. Hybrid model combining the strengths of both: Mamba's efficiency on long sequences + Attention's expressiveness. First commercial LLM using Mamba-style blocks.