Apply suggestions from code review

Co-authored-by: drbh <david.richard.holtz@gmail.com>

launcher/src/main.rs

@@ -1498,7 +1498,6 @@ fn main() -> Result<(), LauncherError> {
             let config: Config = config.into();
             match config.head_dim {
                 Some(h) if h == 64 || h == 128 || h == 256 => {
-                    // std::env::set_var("ATTENTION", "flashdecoding");
                     if args.lora_adapters.is_some() {
                         tracing::info!("Disabling prefix caching because of lora adapters");
                         std::env::set_var("USE_PREFIX_CACHING", "0");

server/text_generation_server/layers/attention/cuda.py

@@ -233,7 +233,6 @@ if ATTENTION == "flashinfer":
         causal=True,
         softcap=0.0,
     ):
-        # assert window_size_left == -1, "Windowing is not supported with flash infer"
         from text_generation_server.layers.attention.flashinfer import (
             prefill_with_paged_kv_state,
         )