llama : add assert about missing llama_encode() call (#8400)

Co-authored-by: Stanisław Szymczyk <redacted>

diff --git a/src/llama.cpp b/src/llama.cpp
index 2b9ace28584572ff60a8c7844c6f06f9d5e4f4ec..80cc1da2660e56588ef7d0b1e3582836ccfd43fd 100644 (file)
--- a/src/llama.cpp
+++ b/src/llama.cpp
@@ -13200,6 +13200,8 @@ struct llm_build_context {
                     LLM_NORM_RMS, cb, -1);
             cb(cur, "result_norm", -1);
         } else {
+            GGML_ASSERT(n_outputs_enc > 0 && "call llama_encode() first");
+
             struct ggml_tensor * embd_enc       = llm_build_inp_embd_enc();
             struct ggml_tensor * pos_bucket_dec = llm_build_pos_bucket(true);