HF团队优化异步RL权重同步，带宽成本降低100倍
The HF science team just made async RL weight sync ~100x cheaper on bandwidth, and you don't need a shared cluster anymore.
The problem: every RL step, the trainer typically has to sync fresh weights to the inference engine. for a 7B in bf16 that's ~14GB. for a frontier 1T fp8 https://t.co/gEqOUoG5O2
The problem: every RL step, the trainer typically has to sync fresh weights to the inference engine. for a 7B in bf16 that's ~14GB. for a frontier 1T fp8 https://t.co/gEqOUoG5O2