List your company

Training Efficiency on Distillation (train)

150Total Tokens ProcessedBitMamba-2
27.2859.1491122.86Apr 28, 2026

Evaluation results

MethodLinks
BitMamba-2Architecture=Mamba-2, Precision=1.58-bit2026.04150—
Bi-MambaArchitecture=Mamba-1, Precision=1-bit2026.0410532
DF-SSMArchitecture=Mamba-2, Precision=1-bit + int8 LoRA2026.04321