@ -157,7 +157,7 @@ class Autolog:
model_precision="fp32"):
import auto_log
pid = os.getpid()
if (os.environ['CUDA_VISIBLE_DEVICES'] != ''):
if (os.environ['CUDA_VISIBLE_DEVICES'].strip() != ''):
gpu_id = int(os.environ['CUDA_VISIBLE_DEVICES'].split(',')[0])
infer_config = inference.Config()
infer_config.enable_use_gpu(100, gpu_id)