Add a argument in turboquant server to disable quantization and run model normally for benchmarking on server side

[Evil-Incorporated]

Is there an argument that can be added in the turboquant-server to run it normally without quantization? This would be really helpful for various benchmarking issues on running server side issues.