public_git/whisper.cpp
1
0
Fork 0
mirror of https://github.com/ggml-org/whisper.cpp.git synced 2026-10-20 00:22:49 +02:00
Code Issues Packages Projects Releases Wiki Activity

CUDA + ggml: add sparse-fa for DSV4/GLM (llama/27970)

Browse Source
This commit is contained in:
Aman Gupta
2026-09-04 13:39:40 +03:00
committed by Georgi Gerganov
parent c2b400754b
commit 519df618df
7 changed files with 324 additions and 86 deletions
Download Patch File Download Diff File Expand all files Collapse all files
ggml/src/ggml.c
+9
View File
@@ -5506,6 +5506,15 @@ enum ggml_prec ggml_flash_attn_ext_get_prec(
return (enum ggml_prec) prec_i32;
}
void ggml_flash_attn_ext_set_n_kv_max(
struct ggml_tensor * a,
int32_t n_kv_max) {
GGML_ASSERT(a->op == GGML_OP_FLASH_ATTN_EXT);
GGML_ASSERT(n_kv_max >= 0);
ggml_set_op_params_i32(a, 4, n_kv_max);
}
void ggml_flash_attn_ext_add_sinks(
struct ggml_tensor * a,
struct ggml_tensor * sinks) {