tqcq/llama.cpp
0
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-08-15 20:53:00 -04:00
Code Issues Packages Projects Releases Wiki Activity

metal : fix warnings (skipme) (#0)

Browse Source
This commit is contained in:
Georgi Gerganov
2024-05-11 21:36:20 +03:00
parent 1622ac023f
commit 7bd4ffb780
2 changed files with 3 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
ggml-metal.metal
View File

@@ -2217,7 +2217,7 @@ kernel void kernel_flash_attn_ext_f16(
// ALiBi // ALiBi
if (max_bias > 0.0f) { if (max_bias > 0.0f) {
const short h = iq2; const uint32_t h = iq2;
const float base = h < n_head_log2 ? m0 : m1; const float base = h < n_head_log2 ? m0 : m1;
const int exph = h < n_head_log2 ? h + 1 : 2*(h - n_head_log2) + 1; const int exph = h < n_head_log2 ? h + 1 : 2*(h - n_head_log2) + 1;
@@ -2473,7 +2473,7 @@ kernel void kernel_flash_attn_ext_vec_f16(
// ALiBi // ALiBi
if (max_bias > 0.0f) { if (max_bias > 0.0f) {
const short h = iq2; const uint32_t h = iq2;
const float base = h < n_head_log2 ? m0 : m1; const float base = h < n_head_log2 ? m0 : m1;
const int exp = h < n_head_log2 ? h + 1 : 2*(h - n_head_log2) + 1; const int exp = h < n_head_log2 ? h + 1 : 2*(h - n_head_log2) + 1;

2
scripts/sync-ggml.last
View File

@@ -1 +1 @@
d2ad1793f45922665c99f5cf4244dcab77e16c00 30f54cbb3ada3e4c5bc6924de3e5918e5be4ff11