List your company

Hybrid Mamba-Transformer Inference on ShareGPT (trace replay)

2.33OOM Ratefixed_dual_mr05
-0.712819.826140.36560.9039May 21, 2026

Evaluation results

MethodLinks
fixed_dual_mr05Variant=fixed_dual_mr052026.052.33244155——
avmp_static_mr05Variant=avmp_static_mr052026.052.3360615502.48
avmp_dynamic_b128Variant=avmp_dynamic_b1282026.052.3357815502.36
fixed_dual_mr09Variant=fixed_dual_mr092026.0567.827015565.51.1
padded_unifiedVariant=padded_unified2026.0578.41,12915576.14.62