LLM: Support package/quantize for llama.cpp/redpajama.cpp on Windows (#8236)

* support windows of llama.cpp

* update quantize

* update version of llama.cp submodule

* add gptneox.dll

* add quantize-gptneox.exe

python/llm/CMakeLists.txt

@@ -32,27 +32,27 @@ add_custom_target(
 )
 install(
     FILES ${CMAKE_CURRENT_SOURCE_DIR}/vendor/llama.cpp/libllama.so
-    DESTINATION src/bigdl/llm/lib
+    DESTINATION src/bigdl/llm/libs
 )
 install(
     PROGRAMS ${CMAKE_CURRENT_SOURCE_DIR}/vendor/llama.cpp/quantize
     RENAME quantize-llama
-    DESTINATION src/bigdl/llm/bin
+    DESTINATION src/bigdl/llm/libs
 )
 install(
     FILES ${CMAKE_CURRENT_SOURCE_DIR}/vendor/redpajama.cpp/libgptneox.so
-    DESTINATION src/bigdl/llm/lib
+    DESTINATION src/bigdl/llm/libs
 )
 install(
     PROGRAMS ${CMAKE_CURRENT_SOURCE_DIR}/vendor/redpajama.cpp/quantize-gptneox
-    DESTINATION src/bigdl/llm/bin
+    DESTINATION src/bigdl/llm/libs
 )
 install(
     FILES ${CMAKE_CURRENT_SOURCE_DIR}/vendor/bloomz.cpp/libbloom.so
-    DESTINATION src/bigdl/llm/lib
+    DESTINATION src/bigdl/llm/libs
 )
 install(
     PROGRAMS ${CMAKE_CURRENT_SOURCE_DIR}/vendor/bloomz.cpp/quantize
     RENAME quantize-bloomz
-    DESTINATION src/bigdl/llm/bin
+    DESTINATION src/bigdl/llm/libs
 )

python/llm/setup-linux.py

@@ -0,0 +1,74 @@
+#!/usr/bin/env python
+
+#
+# Copyright 2016 The BigDL Authors.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+#
+
+import os
+import fnmatch
+from skbuild import setup
+
+long_description = '''
+    BigDL LLM
+'''
+
+exclude_patterns = ["*__pycache__*", "*ipynb_checkpoints*"]
+BIGDL_PYTHON_HOME = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
+VERSION = open(os.path.join(BIGDL_PYTHON_HOME, 'version.txt'), 'r').read().strip()
+llm_home = os.path.join(os.path.dirname(os.path.abspath(__file__)), "src")
+
+
+def get_llm_packages():
+    llm_packages = []
+    for dirpath, _, _ in os.walk(os.path.join(llm_home, "bigdl")):
+        print(dirpath)
+        package = dirpath.split(llm_home + os.sep)[1].replace(os.sep, '.')
+        if any(fnmatch.fnmatchcase(package, pat=pattern)
+                for pattern in exclude_patterns):
+            print("excluding", package)
+        else:
+            llm_packages.append(package)
+            print("including", package)
+    return llm_packages
+
+
+def setup_package():
+    metadata = dict(
+        name='bigdl-llm',
+        version=VERSION,
+        description='Large Language Model Develop Toolkit',
+        long_description=long_description,
+        long_description_content_type="text/markdown",
+        author='BigDL Authors',
+        author_email='bigdl-user-group@googlegroups.com',
+        license='Apache License, Version 2.0',
+        url='https://github.com/intel-analytics/BigDL',
+        packages=get_llm_packages(),
+        package_dir={"": "src"},
+        install_requires=[""],
+        include_package_data=True,
+        classifiers=[
+            'License :: OSI Approved :: Apache Software License',
+            'Programming Language :: Python :: 3',
+            'Programming Language :: Python :: 3.9',
+            'Programming Language :: Python :: Implementation :: CPython'],
+        platforms=['linux']
+    )
+
+    setup(**metadata)
+
+
+if __name__ == '__main__':
+    setup_package()

python/llm/setup.py

@@ -18,7 +18,8 @@
 
 import os
 import fnmatch
-from skbuild import setup
+from setuptools import setup
+import urllib.request
 
 long_description = '''
     BigDL LLM
@@ -44,7 +45,36 @@ def get_llm_packages():
     return llm_packages
 
 
+lib_urls = [
+    "https://sourceforge.net/projects/analytics-zoo/files/bigdl-llm/llama.dll",
+    "https://sourceforge.net/projects/analytics-zoo/files/bigdl-llm/quantize-llama.exe",
+    "https://sourceforge.net/projects/analytics-zoo/files/bigdl-llm/gptneox.dll",
+    "https://sourceforge.net/projects/analytics-zoo/files/bigdl-llm/quantize-gptneox.exe",
+    # TODO: add bloomz
+]
+
+
+def download_libs(url: str):
+    libs_dir = os.path.join(llm_home, "bigdl", "llm", "libs")
+    if not os.path.exists(libs_dir):
+        os.makedirs(libs_dir, exist_ok=True)
+    libso_file_name = url.split('/')[-1]
+    libso_file = os.path.join(libs_dir, libso_file_name)
+    if not os.path.exists(libso_file):
+        urllib.request.urlretrieve(url, libso_file)
+
+
 def setup_package():
+    
+    package_data = [
+        "libs/llama.dll",
+        "libs/quantize-llama.exe",
+        "libs/gptneox.dll",
+    ]
+
+    for url in lib_urls:
+        download_libs(url)
+
     metadata = dict(
         name='bigdl-llm',
         version=VERSION,
@@ -57,14 +87,14 @@ def setup_package():
         url='https://github.com/intel-analytics/BigDL',
         packages=get_llm_packages(),
         package_dir={"": "src"},
-        install_requires=[""],
+        package_data={"bigdl.llm": package_data},
         include_package_data=True,
         classifiers=[
             'License :: OSI Approved :: Apache Software License',
             'Programming Language :: Python :: 3',
             'Programming Language :: Python :: 3.9',
             'Programming Language :: Python :: Implementation :: CPython'],
-        platforms=['linux', 'windows']
+        platforms=['windows']
     )
 
     setup(**metadata)

python/llm/src/bigdl/llm/ggml/quantize.py

@@ -17,14 +17,17 @@
 import os
 import subprocess
 from bigdl.llm.utils.common import invalidInputError
+import platform
 
 
 dirname, _ = os.path.split(os.path.abspath(__file__))
-bin_dirname = os.path.dirname(dirname)
+libs_dirname = os.path.dirname(dirname)
 
 _llama_quantize_type = {"q4_0": 2,
                         "q4_1": 3,
-                        "q4_2": 5}
+                        "q5_0": 8,
+                        "q5_1": 9,
+                        "q8_0": 7}
 _bloomz_quantize_type = {"q4_0": 2,
                          "q4_1": 3}
 _gptneox_quantize_type = {"q4_0": 2,
@@ -78,11 +81,16 @@ def quantize(input_path: str, output_path: str=None,
                       list(quantize_type_map.keys()),
                       dtype))
     quantize_type = quantize_type_map[dtype]
-    quantize_args = "{0}/bin/quantize-{1} {2} {3} {4}".format(bin_dirname,
+    if platform.platform().startswith('Windows'):
-                                                              model_family,
+        suffix = '.exe'
-                                                              input_path,
+    else:
-                                                              output_path,
+        suffix = ''
-                                                              str(quantize_type))
+    quantize_args = "{0}/libs/quantize-{1}{2} {3} {4} {5}".format(libs_dirname,
+                                                                  model_family,
+                                                                  suffix,
+                                                                  input_path,
+                                                                  output_path,
+                                                                  str(quantize_type))
     p = subprocess.Popen(quantize_args.split())
     p.communicate()
     invalidInputError(not p.returncode,

python/llm/vendor/llama.cpp

@@ -1 +1 @@
-Subproject commit f6ab8e2a03f537e853fef2deb36be89c4f698b05
+Subproject commit 5026ad5e525e3bb8354d9169aaf9b147ef8b6b85