genewoo

my 2 cents. This is not useful in production, especially in microservices env. Reasons: - in k8s, filebit collect the log with app name - in raw deployment, different services...

Also, in case of single direct push notification, use MQ instead.

Just saw the issue, I thought n_gpu_layers is a setting of how many layers of the model are offloaded to your GPU. In the document, even set as 1, you...