diff --git a/backends/llamacpp/src/main.rs b/backends/llamacpp/src/main.rs index 4afa64e9..1c7c5e4c 100644 --- a/backends/llamacpp/src/main.rs +++ b/backends/llamacpp/src/main.rs @@ -57,7 +57,7 @@ struct Args { use_mlock: bool, /// Enable flash attention for faster inference. (EXPERIMENTAL) - #[clap(default_value = "false", long, env)] + #[clap(default_value = "true", long, env)] flash_attention: bool, /// TODO