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

ggml : fix YARN + add tests + add asserts (#7617)

Browse source 
* tests : add rope tests

ggml-ci

* ggml : fixes (hopefully)

ggml-ci

* tests : add non-cont tests

ggml-ci

* cuda : add asserts for rope/norm + fix DS2

ggml-ci

* ggml : assert contiguousness

* tests : reduce RoPE tests

ggml-ci
This commit is contained in:
Georgi Gerganov 2024-05-29 20:17:31 +03:00 • committed by GitHub
parent cce3dcffc5
commit fb76ec31a9
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
12 changed files with 167 additions and 104 deletions
Download patch file Download diff file Expand all files Collapse all files

6
ggml-cuda/norm.cu
View file

@ -170,6 +170,8 @@ void ggml_cuda_op_norm(ggml_backend_cuda_context & ctx, ggml_tensor * dst) {
float * dst_d = (float *)dst->data;
cudaStream_t stream = ctx.stream();
GGML_ASSERT(ggml_is_contiguous(src0));
GGML_ASSERT(src0->type == GGML_TYPE_F32);
GGML_ASSERT( dst->type == GGML_TYPE_F32);
@ -188,6 +190,8 @@ void ggml_cuda_op_group_norm(ggml_backend_cuda_context & ctx, ggml_tensor * dst)
float * dst_d = (float *)dst->data;
cudaStream_t stream = ctx.stream();
GGML_ASSERT(ggml_is_contiguous(src0));
GGML_ASSERT(src0->type == GGML_TYPE_F32);
GGML_ASSERT( dst->type == GGML_TYPE_F32);
@ -202,6 +206,8 @@ void ggml_cuda_op_rms_norm(ggml_backend_cuda_context & ctx, ggml_tensor * dst) {
float * dst_d = (float *)dst->data;
cudaStream_t stream = ctx.stream();
GGML_ASSERT(ggml_is_contiguous(src0));
GGML_ASSERT(src0->type == GGML_TYPE_F32);
GGML_ASSERT( dst->type == GGML_TYPE_F32);