diff --git a/README.md b/README.md
index 37a1e9a..58a81b1 100644
--- a/README.md
+++ b/README.md
@@ -8,8 +8,8 @@
 [![Python][python-ver-image]][pypi-project-url]
 [![License][mit-image]][mit-url]
 -->
-[![Downloads](https://img.shields.io/pypi/dm/llama-cli-cffi)](https://pypistats.org/packages/llama-cli-cffi)
-[![Supported Versions](https://img.shields.io/pypi/pyversions/llama-cli-cffi)](https://pypi.org/project/llama-cli-cffi)
+[![Downloads](https://img.shields.io/pypi/dm/llama-cpp-cffi)](https://pypistats.org/packages/llama-cpp-cffi)
+[![Supported Versions](https://img.shields.io/pypi/pyversions/llama-cpp-cffi)](https://pypi.org/project/llama-cpp-cffi)
 [![License: MIT](https://img.shields.io/badge/license-MIT-blue.svg)](https://opensource.org/licenses/MIT)
 
 **Python** binding for [llama.cpp](https://github.com/ggerganov/llama.cpp) using **cffi** and **ctypes**. Supports **CPU** and **CUDA 12.5** execution.
@@ -17,7 +17,7 @@
 ## Install
 
 ```bash
-pip install llama-cli-cffi
+pip install llama-cpp-cffi
 ```
 
 ## Example
diff --git a/pyproject.toml b/pyproject.toml
index 5583859..b4503e7 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "llama-cpp-cffi"
-version = "0.0.4"
+version = "0.1.0"
 description = "Python binding for llama.cpp using cffi"
 homepage = "https://github.com/mtasic85/llama-cpp-cffi"
 repository = "https://github.com/mtasic85/llama-cpp-cffi"
@@ -35,8 +35,11 @@ script = "scripts/build.py"
 [tool.cibuildwheel]
 build-frontend = "build"
 before-build = "pip install poetry"
+# before-build = "pip install poetry; yum -y install wget"
 skip = ["cp36-*", "cp37-*", "cp38-*", "cp39-*", "pp*", "*-win32", "*i686"]
 build-verbosity=3
+repair-wheel-command = ""
+# environment = {"LD_LIBRARY_PATH" = "/project/cuda-12.5.1/dist/lib64:$LD_LIBRARY_PATH", "CUDA_HOME" = "/project/cuda-12.5.1/dist"}
 
 [tool.cibuildwheel.pyodide]
 
diff --git a/scripts/build.py b/scripts/build.py
index 922e1df..4fb7fe5 100644
--- a/scripts/build.py
+++ b/scripts/build.py
@@ -2,6 +2,7 @@
 import glob
 import shutil
 import subprocess
+from pprint import pprint
 
 from cffi import FFI
 
@@ -17,6 +18,7 @@ def clone_llama_cpp():
 def build_cpu(*args, **kwargs):
     # build static and shared library
     env = os.environ.copy()
+    pprint(env)
 
     #
     # build llama.cpp
@@ -86,12 +88,17 @@ def build_cuda_12_5(*args, **kwargs):
     cuda_file = 'cuda_12.5.1_555.42.06_linux.run'
     cuda_url = f'https://developer.download.nvidia.com/compute/cuda/12.5.1/local_installers/{cuda_file}'
     cuda_output_dir = os.path.abspath('./cuda-12.5.1')
+    cuda_file_path = os.path.join(cuda_output_dir, cuda_file)
 
     env['PATH'] = env['PATH'] + f':{cuda_output_dir}/dist/bin'
     env['CUDA_PATH'] = f'{cuda_output_dir}/dist'
 
     # download cuda file
-    subprocess.run(['wget', '-N', cuda_url, '-P', cuda_output_dir], check=True)
+    if not os.path.exists(cuda_file_path):
+        cmd = ['mkdir', '-p', f'{cuda_output_dir}']
+        
+        subprocess.run(cmd, check=True)
+        subprocess.run(['curl', '-o', cuda_file_path, cuda_url], check=True)
     
     # extract cuda file
     cmd = ['chmod', '+x', f'{cuda_output_dir}/{cuda_file}']
@@ -169,11 +176,11 @@ def build_cuda_12_5(*args, **kwargs):
         ''',
         libraries=[
             'stdc++',
-            'cuda',
-            'cublas',
-            'culibos',
-            'cudart',
-            'cublasLt', 
+            # 'cuda',
+            # 'cublas',
+            # 'culibos',
+            # 'cudart',
+            # 'cublasLt',
         ],
         library_dirs=[f'{cuda_output_dir}/dist/lib64'],
         extra_objects=['../llama.cpp/llama_cli.a'],
@@ -201,7 +208,7 @@ def build(*args, **kwargs):
     build_cpu(*args, **kwargs)
 
     # cuda 12.5
-    if os.environ['AUDITWHEEL_ARCH'] == 'x86_64':
+    if os.environ['AUDITWHEEL_POLICY'] == 'manylinux2014' and os.environ['AUDITWHEEL_ARCH'] == 'x86_64':
         clean_llama_cpp()
         build_cuda_12_5(*args, **kwargs)