Add a function to clear the KV cache in falcon. (#2066)

* Add a function to clear the KV cache in falcon. * Clippy.
2025-06-15 10:26:33 +00:00 · 2024-04-15 09:29:25 +02:00
parent e198bb0816
commit 8ad822a983
2 changed files with 15 additions and 0 deletions
--- a/candle-core/src/quantized/cuda.rs
+++ b/candle-core/src/quantized/cuda.rs
@ -217,6 +217,7 @@ fn mul_mat_vec_via_q8_1(
    Ok(CudaStorage::wrap_cuda_slice(dst, dev.clone()))
 }

+#[allow(clippy::too_many_arguments)]
 fn mul_mat_via_q8_1(
    data: &CudaSlice<u8>,
    y: &CudaView<f32>,