From 4fb1bd85a0124799e9468ec6f9d1549b8eacf73c Mon Sep 17 00:00:00 2001
From: JohannesGaessler <johannesg@5d6.de>
Date: Wed, 3 Jan 2024 11:20:28 +0100
Subject: [PATCH] Fix CUDA diag_mask_inf tests with LLAMA_FAST

---
 ggml.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/ggml.c b/ggml.c
index bcec200f6..9039e1e5f 100644
--- a/ggml.c
+++ b/ggml.c
@@ -10928,7 +10928,7 @@ static void ggml_compute_forward_diag_mask_inf(
     switch (src0->type) {
         case GGML_TYPE_F32:
             {
-                ggml_compute_forward_diag_mask_f32(params, src0, dst, -INFINITY);
+                ggml_compute_forward_diag_mask_f32(params, src0, dst, -FLT_MAX);
             } break;
         default:
             {