fix output counting comment

llama.cpp

@@ -12618,8 +12618,7 @@ static int llama_decode_internal(
     std::vector<llama_seq_id *>            seq_id_arr;
     std::vector<std::vector<llama_seq_id>> seq_id;
 
-    // this indicates we are doing pooling on an embedding model. non-embedding models always
-    // use "output_ids" so we need to preserve all outputs in that case (somewhat inefficiently)
+    // this indicates we are doing pooled embedding, so we ignore batch.logits and output all tokens
     bool embed_pooled = cparams.embeddings && cparams.pooling_type != LLAMA_POOLING_TYPE_NONE;
 
     // count outputs