Error while running setup_env.py -md models/BitNet-b1.58-2B-4T -q i2_s

[MaxwellWjj]

Hi developers,

When I followed the tutorial (Build from source), and run python setup_env.py -md models/BitNet-b1.58-2B-4T -q i2_s, the printed error is:

INFO:root:Compiling the code using CMake.
ERROR:root:Error occurred while running command: Command '['cmake', '--build', 'build', '--config', 'Release']' returned non-zero exit status 2., check details in logs/compile.log

Then I went into logs/compile.log, the error should be:

[  2%] Linking CXX shared library libggml.dylib
[ 11%] Built target ggml
[ 12%] Linking CXX shared library libllama.dylib
Undefined symbols for architecture arm64:
  "std::__1::__hash_memory(void const*, unsigned long)", referenced from:
      std::__1::pair<std::__1::__hash_iterator<std::__1::__hash_node<std::__1::__hash_value_type<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>, llama_model_kv_override>, void*>*>, bool> std::__1::__hash_table<std::__1::__hash_value_type<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>, llama_model_kv_override>, std::__1::__unordered_map_hasher<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>, std::__1::pair<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>> const, llama_model_kv_override>, std::__1::hash<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>>, std::__1::equal_to<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>>, true>, std::__1::__unordered_map_equal<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>, std::__1::pair<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>> const, llama_model_kv_override>, std::__1::equal_to<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>>, std::__1::hash<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>>, true>, std::__1::allocator<std::__1::pair<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>> const, llama_model_kv_override>>>::__emplace_unique_key_args<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>, std::__1::pair<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>> const, llama_model_kv_override>>(std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>> const&, std::__1::pair<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>> const, llama_model_kv_override>&&) in llama.cpp.o
      std::__1::__hash_iterator<std::__1::__hash_node<std::__1::__hash_value_type<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>, llama_model_kv_override>, void*>*> std::__1::__hash_table<std::__1::__hash_value_type<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>, llama_model_kv_override>, std::__1::__unordered_map_hasher<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>, std::__1::pair<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>> const, llama_model_kv_override>, std::__1::hash<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>>, std::__1::equal_to<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>>, true>, std::__1::__unordered_map_equal<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>, std::__1::pair<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>> const, llama_model_kv_override>, std::__1::equal_to<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>>, std::__1::hash<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>>, true>, std::__1::allocator<std::__1::pair<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>> const, llama_model_kv_override>>>::find<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>>(std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>> const&) in llama.cpp.o
      std::__1::pair<std::__1::__hash_iterator<std::__1::__hash_node<std::__1::__hash_value_type<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>, std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>>, void*>*>, bool> std::__1::__hash_table<std::__1::__hash_value_type<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>, std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>>, std::__1::__unordered_map_hasher<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>, std::__1::pair<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>> const, std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>>, std::__1::hash<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>>, std::__1::equal_to<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>>, true>, std::__1::__unordered_map_equal<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>, std::__1::pair<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>> const, std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>>, std::__1::equal_to<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>>, std::__1::hash<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>>, true>, std::__1::allocator<std::__1::pair<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>> const, std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>>>>::__emplace_unique_impl<char const*&, std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>> const&>(char const*&, std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>> const&) in llama.cpp.o
      std::__1::__hash_table<std::__1::__hash_value_type<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>, std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>>, std::__1::__unordered_map_hasher<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>, std::__1::pair<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>> const, std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>>, std::__1::hash<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>>, std::__1::equal_to<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>>, true>, std::__1::__unordered_map_equal<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>, std::__1::pair<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>> const, std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>>, std::__1::equal_to<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>>, std::__1::hash<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>>, true>, std::__1::allocator<std::__1::pair<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>> const, std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>>>>::__node_insert_unique(std::__1::__hash_node<std::__1::__hash_value_type<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>, std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>>, void*>*) in llama.cpp.o
      std::__1::pair<std::__1::__hash_iterator<std::__1::__hash_node<std::__1::__hash_value_type<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>, int>, void*>*>, bool> std::__1::__hash_table<std::__1::__hash_value_type<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>, int>, std::__1::__unordered_map_hasher<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>, std::__1::pair<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>> const, int>, std::__1::hash<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>>, std::__1::equal_to<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>>, true>, std::__1::__unordered_map_equal<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>, std::__1::pair<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>> const, int>, std::__1::equal_to<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>>, std::__1::hash<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>>, true>, std::__1::allocator<std::__1::pair<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>> const, int>>>::__emplace_unique_key_args<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>, std::__1::piecewise_construct_t const&, std::__1::tuple<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>> const&>, std::__1::tuple<>>(std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>> const&, std::__1::piecewise_construct_t const&, std::__1::tuple<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>> const&>&&, std::__1::tuple<>&&) in llama.cpp.o
      std::__1::__hash_iterator<std::__1::__hash_node<std::__1::__hash_value_type<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>, int>, void*>*> std::__1::__hash_table<std::__1::__hash_value_type<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>, int>, std::__1::__unordered_map_hasher<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>, std::__1::pair<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>> const, int>, std::__1::hash<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>>, std::__1::equal_to<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>>, true>, std::__1::__unordered_map_equal<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>, std::__1::pair<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>> const, int>, std::__1::equal_to<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>>, std::__1::hash<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>>, true>, std::__1::allocator<std::__1::pair<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>> const, int>>>::find<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>>(std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>> const&) in llama.cpp.o
      std::__1::__hash_iterator<std::__1::__hash_node<std::__1::__hash_value_type<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>, llama_lora_weight>, void*>*> std::__1::__hash_table<std::__1::__hash_value_type<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>, llama_lora_weight>, std::__1::__unordered_map_hasher<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>, std::__1::pair<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>> const, llama_lora_weight>, std::__1::hash<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>>, std::__1::equal_to<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>>, true>, std::__1::__unordered_map_equal<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>, std::__1::pair<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>> const, llama_lora_weight>, std::__1::equal_to<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>>, std::__1::hash<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>>, true>, std::__1::allocator<std::__1::pair<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>> const, llama_lora_weight>>>::find<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>>(std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>> const&) in llama.cpp.o
      ...
ld: symbol(s) not found for architecture arm64
clang++: error: linker command failed with exit code 1 (use -v to see invocation)
make[2]: *** [3rdparty/llama.cpp/src/libllama.dylib] Error 1
make[1]: *** [3rdparty/llama.cpp/src/CMakeFiles/llama.dir/all] Error 2
make: *** [all] Error 2

Would you please help me check the problem? The error Undefined symbols for architecture arm64: "std::__1::__hash_memory(void const*, unsigned long)" seems very strange and have no idea how to deal with.

The enviroment I used:

• MacOS 26.0.1

• Cmake version 4.1.2

• Clang version 21.1.4 (Homebrew)

• Python 3.9

[MaxwellWjj]

Complementary: M2 chip (ARM) + MacOS

[RyanStarFox]

metoo, have u solved the problem? also m2pro chip

[RobBW]

I have successfully compiled BitNet on an M2 Mini running Sequoia 15.6.1. using the Warp terminal to diagnose the problems I encountered. I attach a copy of the Compilation log which is pretty self explanatory. Since it is so long here are the main insights.

1. The compilation is failing due to an incompatibility with Apple SDK headers and the compiler flags being used. The error is: error: non-defining declaration of enumeration with a fixed underlying type is only permitted as a standalone declaration; missing list of enumerators? [-Welaborated-enum-base]

This is a C++ language standard issue.

2. Line 192 in the ggml CMakeLists.txt sets CMAKE_CXX_STANDARD 11. This is the problem - C++11 doesn't support the elaborated enum syntax that Apple's SDK headers use. The error -Welaborated-enum-base is specifically about C++11 not allowing forward declarations of enums with fixed underlying types in the form enum EnumName : type;.

The solution is to upgrade to C++17, which supports this syntax. Those used were:

4. The C compiler identification is AppleClang 17.0.0.17000013
5. The CXX compiler identification is AppleClang 17.0.0.17000013

Checking for compilation errors came up clean

7. Excellent! Zero errors. The fix worked perfectly. The compilation errors were caused by the project using C++11, which doesn't support elaborated enum base declarations (forward declarations of enums with fixed underlying types) that Apple's SDK headers use.
8. By upgrading from C++11 to C++17, the project now compiles successfully without any errors. The warnings that remain are minor and don't prevent compilation.

After this I was able to chat with Bitnet4T without problems.

Additional thoughts:

• My XCode seems to use LLVM@17 but LLVM@20 is available with specific Apple Silicon enhancements which suggests that further speedups are possible.
• The CMake compilation uses the XCode default compiler setting. Maybe this setting should be made explicit in CMake. I intend to try using Brew to install LLVM@20
• Where is MLX in all this?

[RobBW]

Compilation file attached:

BitNet4T compilation.txt

[RyanStarFox]

Thank u!! U R so professional! But unfortunately, I am in macOS26. I tried to use clang18, but it seems clang18 is incompatible with macOS26. I believe clang17 is also imcompatible.