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

NPU fix wrong model output shape

This commit is contained in:
Yu, Zijun 2025-12-08 16:18:22 +08:00 committed by Mustafa Cavus
parent 469325c6da
commit ae01322dbd
1 changed files with 2 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
ggml/src/ggml-openvino/utils.cpp
View File

@ -614,8 +614,8 @@ ov::Tensor get_ov_output_tensor(std::shared_ptr<GgmlOvDecoder> ggml_decoder, con
auto output_type = ggml_decoder->get_ov_type(ggml_tensor); auto output_type = ggml_decoder->get_ov_type(ggml_tensor);
auto output_shape = ggml_decoder->get_shape(ggml_tensor); auto output_shape = ggml_decoder->get_shape(ggml_tensor);
if (ggml_decoder->is_static() && result_name == "result_output") { if (ggml_decoder->is_static() && result_name == "result_output" && output_shape[2] == 0) {
output_shape[1] = 1; output_shape[2] = 1;
} }
ov::Tensor output_tensor(output_type, output_shape, ggml_tensor->data); ov::Tensor output_tensor(output_type, output_shape, ggml_tensor->data);
return output_tensor; return output_tensor;