Public/llama.cpp
1
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2026-09-18 16:55:05 +02:00
Code Issues Packages Projects Releases Wiki Activity

squash! sampling : simplify backend sampling logic decode

Browse Source 
The commit fixes a variable shadowing issue in the
`llama_context::decode` function which was introduced in a previous
refactoring.
This commit is contained in:
Daniel Bevenius
2025-11-19 15:10:15 +01:00
parent 38f408c253
commit 18ed4d8f96
1 changed files with 1 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/llama-context.cpp
+1 -1
View File
@@ -1361,7 +1361,7 @@ int llama_context::decode(const llama_batch & batch_inp) {
// ggml_graph_dump_dot(gf, NULL, "llama.dot");
//}
const bool backend_has_sampled = !res->t_sampled_tokens.empty() || !res->t_sampled_probs.empty() || !res->t_sampled_logits.empty();
backend_has_sampled = !res->t_sampled_tokens.empty() || !res->t_sampled_probs.empty() || !res->t_sampled_logits.empty();
if (has_backend_samplers && backend_has_sampled) {
const auto seq_to_batch_idx = build_seq_to_batch_idx(ubatch);