forked from casper-hansen/AutoAWQ
-
Notifications
You must be signed in to change notification settings - Fork 0
/
setup.py
71 lines (65 loc) · 2.18 KB
/
setup.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
import os
import sys
import torch
import platform
from pathlib import Path
from setuptools import setup, find_packages
os.environ["CC"] = "g++"
os.environ["CXX"] = "g++"
AUTOAWQ_VERSION = "0.1.8"
PYPI_BUILD = os.getenv("PYPI_BUILD", "0") == "1"
HAS_CUDA = torch.cuda.is_available()
if not PYPI_BUILD and HAS_CUDA:
try:
CUDA_VERSION = "".join(os.environ.get("CUDA_VERSION", torch.version.cuda).split("."))[:3]
AUTOAWQ_VERSION += f"+cu{CUDA_VERSION}"
except Exception as ex:
raise RuntimeError("Your system must have an Nvidia GPU for installing AutoAWQ")
common_setup_kwargs = {
"version": AUTOAWQ_VERSION,
"name": "autoawq",
"author": "Casper Hansen",
"license": "MIT",
"python_requires": ">=3.8.0",
"description": "AutoAWQ implements the AWQ algorithm for 4-bit quantization with a 2x speedup during inference.",
"long_description": (Path(__file__).parent / "README.md").read_text(encoding="UTF-8"),
"long_description_content_type": "text/markdown",
"url": "https://github.com/casper-hansen/AutoAWQ",
"keywords": ["awq", "autoawq", "quantization", "transformers"],
"platforms": ["linux", "windows"],
"classifiers": [
"Environment :: GPU :: NVIDIA CUDA :: 11.8",
"Environment :: GPU :: NVIDIA CUDA :: 12",
"License :: OSI Approved :: MIT License",
"Natural Language :: English",
"Programming Language :: Python :: 3.8",
"Programming Language :: Python :: 3.9",
"Programming Language :: Python :: 3.10",
"Programming Language :: Python :: 3.11",
"Programming Language :: C++",
]
}
requirements = [
"torch>=2.0.1",
"transformers>=4.35.0",
"tokenizers>=0.12.1",
"accelerate",
"datasets",
]
# CUDA kernels
if platform.system().lower() != "darwin" and HAS_CUDA:
requirements.append("autoawq-kernels")
setup(
packages=find_packages(),
install_requires=requirements,
extras_require={
"eval": [
"lm_eval>=0.4.0",
"tabulate",
"protobuf",
"evaluate",
"scipy"
],
},
**common_setup_kwargs
)