参数的读取 - 技术栈

argparse函数的读取

这个是以函数的形式嵌入到脚本中的

python 复制代码

def common_args():
    parser = argparse.ArgumentParser(description='common config')
    parser.add_argument('--test', action='store_true', help="test mode (load model and test dataset)")
    parser.add_argument('--iters', type=int, default=200000, help="training iters")
    parser.add_argument('--lr', type=float, default=1e-2, help="initial learning rate")
    parser.add_argument('--lr_net', type=float, default=1e-3, help="initial learning rate")
    parser.add_argument('--ckpt', type=str, default='latest')
    args = parser.parse_args()
    return args

py文件的读取

python 复制代码

import os
from pathlib import Path
from easydict import EasyDict as edict

FILE_PATH = Path(__file__).resolve()
ROOT_DIR = FILE_PATH.parents[1]

proj_conf = edict()

# 基本路径的设置
proj_conf.path = edict()
proj_conf.path.root_dir = str(ROOT_DIR)

# 其他参数的设置，比如网络模型dim
proj_conf.model = edict()
proj_conf.model.hidden_dim = 512

yaml文件的读取

python 复制代码

# coding:utf-8
import yaml
import os

# 获取当前脚本所在文件夹路径
curPath = os.path.dirname(os.path.realpath(__file__))
# 获取yaml文件路径
yamlPath = os.path.join(curPath, "cfgyaml.yaml")

# open方法打开直接读出来
f = open(yamlPath, 'r', encoding='utf-8')
cfg = f.read()
print(type(cfg))  # 读出来是字符串
print(cfg)

d = yaml.load(cfg)  # 用load方法转字典
print(d)
print(type(d))
# dict

@dataclass装饰器读取

python 复制代码

import json
import numpy as np
from dataclasses import dataclass
from typing import Optional, Tuple

@dataclass
class ModelArgs:
    channel: int = 128
    input_shape: tuple = (32, 32)
    schedule: str = "linear"
    num_timesteps: int = 1000
    schedule_low: float = 1e-4
    schedule_high: float = 0.02
    norm_eps: float = 1e-5
    cuda: bool = True
    max_batch_size: int = 32
    max_seq_len: int = 2048

    ffn_dim_multiplier: Optional[float] = None  # python 3.10 可以这么写： ffn_dim_multiplier: int | None = None

# 用法如下: 创建的时候传入就可以了，然后在主函数里面进行定义
class Diffusion:
    def __init__(self, args: ModelArgs):
        super(Diffusion, self).__init__()
        self.model_args = args
        
if __name__ == "__main__":
    with open("params.json", "r") as f:
        params = json.loads(f.read())

    max_seq_len = 2048
    max_batch_size = 16
    model_args: ModelArgs = ModelArgs(
        max_seq_len=max_seq_len,
        max_batch_size=max_batch_size,
        **params,
    )