=== Testing ArmoRM Reward Model === GPU: NVIDIA A100-SXM4-40GB ====================================================================== ArmoRM Reward Model Test ====================================================================== Device: cuda Model: RLHFlow/ArmoRM-Llama3-8B-v0.1 Loading model... [ArmoRM] Loading model RLHFlow/ArmoRM-Llama3-8B-v0.1 on cuda... === Test Complete ===