tqcq/llama.cpp
0
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-06-29 20:45:04 +00:00
Code Issues Packages Projects Releases Wiki Activity

context : fix reorder logic (#13267)

Browse Source 
ggml-ci
This commit is contained in:
Georgi Gerganov
2025-05-02 20:54:13 +03:00
committed by GitHub
parent 3f3769ba76
commit a75cb30dc9
1 changed files with 3 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
src/llama-context.cpp
View File

@ -1050,6 +1050,9 @@ int llama_context::decode(llama_batch & inp_batch) {
// finalize the batch processing // finalize the batch processing
kv_guard.commit(); kv_guard.commit();
// set to total number of outputs in the batch, for use in llama_get_logits_ith
n_outputs = n_outputs_all;
// set output mappings // set output mappings
{ {
bool sorted_output = true; bool sorted_output = true;
@ -1103,9 +1106,6 @@ int llama_context::decode(llama_batch & inp_batch) {
} }
} }
// set to total number of outputs in the batch, for use in llama_get_logits_ith
n_outputs = n_outputs_all;
// wait for the computation to finish (automatically done when obtaining the model output) // wait for the computation to finish (automatically done when obtaining the model output)
//synchronize(); //synchronize();