vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Merge branch 'master' into gg/flash-attn

Browse source
This commit is contained in:
Georgi Gerganov 2024-03-27 10:24:09 +02:00
commit 013721df2b
No known key found for this signature in database
GPG key ID: BF970631944C16B7
157 changed files with 19090 additions and 15488 deletions
Download patch file Download diff file Expand all files Collapse all files

6
ggml-cuda/fattn.cuh Normal file
View file

@ -0,0 +1,6 @@
#include "common.cuh"
void ggml_cuda_flash_attn_ext(
ggml_backend_cuda_context & ctx,
const ggml_tensor * Q, const ggml_tensor * K, const ggml_tensor * V,
const ggml_tensor * mask, ggml_tensor * KQV);