vrr/koboldcpp
2
0
Fork 0
mirror of https://github.com/LostRuins/koboldcpp.git synced 2025-09-13 02:19:41 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

sync with sd.cpp

Browse source
This commit is contained in:
Concedo 2025-06-30 00:10:51 +08:00
parent e5af9b5ea9
commit 186227fc26
8 changed files with 234 additions and 82 deletions
Download patch file Download diff file Expand all files Collapse all files

6
otherarch/sdcpp/t5.hpp
View file

@ -795,9 +795,9 @@ struct T5Runner : public GGMLRunner {
void compute(const int n_threads,
struct ggml_tensor* input_ids,
struct ggml_tensor* attention_mask,
ggml_tensor** output,
ggml_context* output_ctx = NULL,
struct ggml_tensor* attention_mask = NULL) {
ggml_context* output_ctx = NULL) {
auto get_graph = [&]() -> struct ggml_cgraph* {
return build_graph(input_ids, attention_mask);
};
@ -966,7 +966,7 @@ struct T5Embedder {
struct ggml_tensor* out = NULL;
int t0 = ggml_time_ms();
model.compute(8, input_ids, &out, work_ctx);
model.compute(8, input_ids, NULL, &out, work_ctx);
int t1 = ggml_time_ms();
print_ggml_tensor(out);