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

ggml : full ALiBi support (#7192)

Browse source 
* ggml : full ALiBi support

* ggml : update ggml_soft_max_ext() CUDA, SYCL

* ggml : ggml_flash_attn_ext() support ALiBi (CPU)

* ggml : ggml_flash_attn_ext() support ALiBi (Metal)

* ggml : fix warning

* ggml : ggml_flash_attn_ext() support ALiBi (CUDA)

ggml-ci

* ggml : fix assert message

* vulkan : add dev notes

* ggml : require mask when using ALiBi

ggml-ci

* convert : fix convert for refact models
This commit is contained in:
Georgi Gerganov 2024-05-11 10:32:41 +03:00 • committed by GitHub
parent e849648888
commit 9cb317f77e
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
16 changed files with 350 additions and 825 deletions
Download patch file Download diff file Expand all files Collapse all files

5
ggml-cuda.cu
View file

@ -4,7 +4,6 @@
#include "ggml-cuda/common.cuh"
#include "ggml-cuda/acc.cuh"
#include "ggml-cuda/alibi.cuh"
#include "ggml-cuda/arange.cuh"
#include "ggml-cuda/argsort.cuh"
#include "ggml-cuda/binbcast.cuh"
@ -2277,9 +2276,6 @@ static bool ggml_cuda_compute_forward(ggml_backend_cuda_context & ctx, struct gg
case GGML_OP_ROPE:
ggml_cuda_op_rope(ctx, dst);
break;
case GGML_OP_ALIBI:
ggml_cuda_op_alibi(ctx, dst);
break;
case GGML_OP_IM2COL:
ggml_cuda_op_im2col(ctx, dst);
break;
@ -2829,7 +2825,6 @@ GGML_CALL static bool ggml_backend_cuda_supports_op(ggml_backend_t backend, cons
case GGML_OP_DIAG_MASK_INF:
case GGML_OP_SOFT_MAX:
case GGML_OP_ROPE:
case GGML_OP_ALIBI:
case GGML_OP_IM2COL:
case GGML_OP_POOL_2D:
case GGML_OP_SUM_ROWS: