Tests for min_p, sampling queue (#5147)

1 files changed, 5 insertions, 0 deletions

diff --git a/llama.cpp b/llama.cpp
index 4cd0f16e..391c956e 100644
--- a/llama.cpp
+++ b/llama.cpp
@@ -8133,6 +8133,11 @@ void llama_sample_softmax(struct llama_context * ctx, llama_token_data_array * c
 }
 
 void llama_sample_top_k(struct llama_context * ctx, llama_token_data_array * candidates, int32_t k, size_t min_keep) {
+    // TODO: move bucket sort to separate function so that top_p/tail_free/typical/softmax first is equally fast
+    // if (k >= (int32_t)candidates->size) {
+    //     return;
+    // }
+
     const int64_t t_start_sample_us = ggml_time_us();
 
     k = std::max(k, (int) min_keep);