Adds a prevent-server-caching arg in order to enable or disable the random prefix being added to all messages in each request. Useful for testing the effectiveness of any server-side caching on PTU resources.
I think this PR should wait until we have a way to replay your own datasets otherwise it may not be very useful as we use the same prompt over and over currently.
Adds a
prevent-server-caching
arg in order to enable or disable the random prefix being added to all messages in each request. Useful for testing the effectiveness of any server-side caching on PTU resources.