sampling: fix top_k <= 0 (#5388)

* sampling: fix top_k <= 0

* Update llama.cpp

Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>

---------

Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>

1 files changed, 4 insertions, 0 deletions

diff --git a/llama.cpp b/llama.cpp
index c45ae1d5..f8f5796a 100644
--- a/llama.cpp
+++ b/llama.cpp
@@ -8585,6 +8585,10 @@ void llama_sample_top_k(struct llama_context * ctx, llama_token_data_array * can
     // }
 
     const int64_t t_start_sample_us = ggml_time_us();
+    
+    if (k <= 0) {
+        k = candidates->size;
+    }
 
     k = std::max(k, (int) min_keep);
     k = std::min(k, (int) candidates->size);