happyz
/
llama.cpp
mirror of https://github.com/ggerganov/llama.cpp.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Fix ROPE accuracy when freq_scale != 1

This commit is contained in:
Yu, Zijun 2025-09-17 15:35:27 +08:00 committed by Mustafa Cavus
parent 602f9ca4af
commit 1a38339cea
2 changed files with 2 additions and 6 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
ggml/src/ggml-openvino/ggml-openvino.cpp
View File

@ -319,12 +319,8 @@ static bool is_op_unsupported_case(const ggml_tensor* op) {
return true;
}
float freq_scale;
memcpy(&freq_scale, op_params + 6, sizeof(float));
if (freq_scale != 0.0f && freq_scale != 1.0f) {
GGML_LOG_WARN("OpenVINO backend does not support ROPE with freq_scale %f != 1.0f\n", freq_scale);
return true;
}
float ext_factor;
memcpy(&freq_scale, op_params + 6, sizeof(float));
memcpy(&ext_factor, op_params + 7, sizeof(float));
if (ext_factor != 0.0f) {
GGML_LOG_WARN("OpenVINO backend does not support ROPE with ext_factor %f != 0.0f\n", ext_factor);

2
ggml/src/ggml-openvino/openvino/utils.cpp
View File

@ -140,7 +140,7 @@ std::pair<ov::Output<Node>, ov::Output<Node>> make_sin_cos(int32_t* rope_params,
ggml_rope_yarn_corr_dims(n_dims, n_ctx_orig, freq_base, beta_fast, beta_slow, corr_dims);
std::vector<float> factor(n_dims / 2);
factor[0] = freq_scale;
factor[0] = 1.0f;
for (size_t i = 1; i < factor.size(); i++) {
factor[i] = theta_scale * factor[i - 1];
}