Jais 2: A Family of Arabic-Centric Open Large Language Models

Open paper

Benchmark	Model	Rank	Results
instruction-following-on-ifeval	Jais-2-70B	–	Inst-level strict-accuracy: 78.93