summaryrefslogtreecommitdiff
path: root/candle-examples/examples/mistral/main.rs
diff options
context:
space:
mode:
Diffstat (limited to 'candle-examples/examples/mistral/main.rs')
-rw-r--r--candle-examples/examples/mistral/main.rs7
1 files changed, 7 insertions, 0 deletions
diff --git a/candle-examples/examples/mistral/main.rs b/candle-examples/examples/mistral/main.rs
index a972279c..c00af3fe 100644
--- a/candle-examples/examples/mistral/main.rs
+++ b/candle-examples/examples/mistral/main.rs
@@ -196,6 +196,10 @@ struct Args {
/// The context size to consider for the repeat penalty.
#[arg(long, default_value_t = 64)]
repeat_last_n: usize,
+
+ /// Use the slower dmmv cuda kernel.
+ #[arg(long)]
+ force_dmmv: bool,
}
fn main() -> Result<()> {
@@ -203,6 +207,9 @@ fn main() -> Result<()> {
use tracing_subscriber::prelude::*;
let args = Args::parse();
+ #[cfg(feature = "cuda")]
+ candle::quantized::cuda::set_force_dmmv(args.force_dmmv);
+
let _guard = if args.tracing {
let (chrome_layer, guard) = ChromeLayerBuilder::new().build();
tracing_subscriber::registry().with(chrome_layer).init();