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

ggml : fix CPU implementation

Browse source
This commit is contained in:
Georgi Gerganov 2023-12-21 21:02:23 +02:00
parent 199f6bdc46
commit 36c3f41f66
No known key found for this signature in database
GPG key ID: 449E073F9DC10735
3 changed files with 12 additions and 13 deletions
Download patch file Download diff file Expand all files Collapse all files

9
tests/test-backend-ops.cpp
View file

@ -766,18 +766,19 @@ struct test_bin_bcast : public test_case {
struct test_scale : public test_case {
const ggml_type type;
const std::array<int64_t, 4> ne;
float scale;
std::string vars() override {
return VARS_TO_STR2(type, ne);
return VARS_TO_STR3(type, ne, scale);
}
test_scale(ggml_type type = GGML_TYPE_F32,
std::array<int64_t, 4> ne = {10, 10, 10, 10})
: type(type), ne(ne) {}
std::array<int64_t, 4> ne = {10, 10, 10, 10},
float scale = 2.0f)
: type(type), ne(ne), scale(scale) {}
ggml_tensor * build_graph(ggml_context * ctx) override {
ggml_tensor * a = ggml_new_tensor(ctx, type, 4, ne.data());
ggml_tensor * scale = ggml_new_tensor_1d(ctx, type, 1);
ggml_tensor * out = ggml_scale(ctx, a, scale);
return out;
}