Mamba3-177M-GQA-Hybrid_LAMBA_V1.0
LAMBA V1.0 is a ~177M-parameter language model trained from scratch on a Mamba-3 (SISO) + Grouped-Query Attention hybrid...
LAMBA V1.0 is a ~177M-parameter language model trained from scratch on a Mamba-3 (SISO) + Grouped-Query Attention hybrid...