Resource Center

Jamba: A Hybrid Transformer-Mamba Language Model

We present Jamba, a new base large language model based on a novel hybrid Transformer-Mamba mixture-of-experts (MoE) architecture.

Generating Benchmarks for Factuality Evaluation of Language Models

We present Jamba, a new base large language model based on a novel hybrid Transformer-Mamba mixture-of-experts (MoE) architecture.

Parallel Context Windows for Large Language Models

We present Jamba, a new base large language model based on a novel hybrid Transformer-Mamba mixture-of-experts (MoE) architecture.