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

ggml : fix rope + llama minor optimizations (#3560)

Browse source 
* Minor fixes and fixed memleak

* Using const auto references in range-based loop C++17
This commit is contained in:
Herman Semenov 2023-10-20 10:02:12 +00:00 • committed by GitHub
parent e78f3ef24a
commit f439e506e8
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
4 changed files with 7 additions and 7 deletions
Download patch file Download diff file Expand all files Collapse all files

3
ggml.c
View file

@ -13537,7 +13537,7 @@ static void ggml_compute_forward_rope_f16(
dst_data[n_dims] = GGML_FP32_TO_FP16(x2*cos_block_theta - x3*sin_block_theta);
dst_data[n_dims/2*3] = GGML_FP32_TO_FP16(x2*sin_block_theta + x3*cos_block_theta);
}
} if (!is_neox) {
} else if (!is_neox) {
for (int64_t i0 = 0; i0 < ne0; i0 += 2) {
const float cos_theta = cosf(theta);
const float sin_theta = sinf(theta);
@ -19170,6 +19170,7 @@ void ggml_graph_export(const struct ggml_cgraph * cgraph, const char * fname) {
if (idx == -1) {
fprintf(stderr, "%s: failed to find tensor, arg = %d, node = %d\n", __func__, j, i);
fclose(fout);
return;
}