domains / families / research / mamba

mamba

separate

State-space model with O(n) complexity vs Transformers' O(n²). 5× faster inference, million-token sequences, no KV cache. Selective SSM with hardware-aware design. Mamba-1 (d_state=16) and Mamba-2 (d_state=128, multi-head). Models 130M-2.8B on HuggingFace.

When to use

• Need long sequences (100K+ tokens)
• Want faster inference than Transformers
• Memory-constrained (no KV cache)
• Building streaming applications
• Linear scaling important
• **O(n) complexity**: Linear vs quadratic
• **5× faster inference**: No attention overhead
• **No KV cache**: Lower memory usage

When not to use

• Best-in-class performance tasks
• RNN+Transformer hybrid needs
• Retention-based architecture needs
• Convolution-based approach needs

Skill page

Source