6 files changed, 14 insertions, 16 deletions
diff --git a/CMakeLists.txt b/CMakeLists.txt
index affff3e..d5715d9 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -253,7 +253,6 @@ endif()
 add_library(llama
             llama.cpp
             llama.h
-            llama_internal.h
             llama_util.h)
 
 target_include_directories(llama PUBLIC .)
diff --git a/Makefile b/Makefile
index c7ccf46..7db2466 100644
--- a/Makefile
+++ b/Makefile
@@ -142,7 +142,7 @@ default: main quantize perplexity embedding
 ggml.o: ggml.c ggml.h
 	$(CC)  $(CFLAGS)   -c ggml.c -o ggml.o
 
-llama.o: llama.cpp llama.h llama_util.h llama_internal.h
+llama.o: llama.cpp llama.h llama_util.h
 	$(CXX) $(CXXFLAGS) -c llama.cpp -o llama.o
 
 common.o: examples/common.cpp examples/common.h
diff --git a/examples/quantize-stats/quantize-stats.cpp b/examples/quantize-stats/quantize-stats.cpp
index 203bfe8..c786fe2 100644
--- a/examples/quantize-stats/quantize-stats.cpp
+++ b/examples/quantize-stats/quantize-stats.cpp
@@ -1,6 +1,7 @@
 #include "ggml.h"
+
+#define LLAMA_API_INTERNAL
 #include "llama.h"
-#include "llama_internal.h"
 
 #include <algorithm>
 #include <cassert>
diff --git a/llama.cpp b/llama.cpp
index 6d8b706..c722956 100644
--- a/llama.cpp
+++ b/llama.cpp
@@ -5,7 +5,6 @@
 
 #include "llama_util.h"
 #include "llama.h"
-#include "llama_internal.h"
 
 #include "ggml.h"
 
diff --git a/llama.h b/llama.h
index 7a258a1..1922175 100644
--- a/llama.h
+++ b/llama.h
@@ -179,4 +179,15 @@ extern "C" {
 }
 #endif
 
+// Internal API to be implemented by llama.cpp and used by tests/benchmarks only
+#ifdef LLAMA_API_INTERNAL
+
+#include <vector>
+#include <string>
+struct ggml_tensor;
+
+std::vector<std::pair<std::string, struct ggml_tensor *>>& llama_internal_get_tensor_map(struct llama_context * ctx);
+
+#endif
+
 #endif // LLAMA_H
diff --git a/llama_internal.h b/llama_internal.h
deleted file mode 100644
index 543eed9..0000000
--- a/llama_internal.h
+++ /dev/null
@@ -1,12 +0,0 @@
-// Internal header to be included by llama.cpp and tests/benchmarks only.
-
-#ifndef LLAMA_INTERNAL_H
-#define LLAMA_INTERNAL_H
-
-#include <vector>
-#include <string>
-struct ggml_tensor;
-
-std::vector<std::pair<std::string, struct ggml_tensor *>>& llama_internal_get_tensor_map(struct llama_context * ctx);
-
-#endif // LLAMA_INTERNAL_H