如何将 pytorch 模型保存在 redis 缓存中以更快地访问模型以进行视频流传输？

Question

我有这个属于feature_extractor.py 的代码，它是here 中这个文件夹的一部分：

import torch
import torchvision.transforms as transforms
import numpy as np
import cv2
from .model import Net

class Extractor(object):
    def __init__(self, model_path, use_cuda=True):
        self.net = Net(reid=True)
        self.device = "cuda" if torch.cuda.is_available() and use_cuda else "cpu"
        state_dict = torch.load(model_path, map_location=lambda storage, loc: storage)['net_dict']
        self.net.load_state_dict(state_dict)
        print("Loading weights from {}... Done!".format(model_path))
        self.net.to(self.device)
        self.size = (64, 128)
        self.norm = transforms.Compose([
            transforms.ToTensor(),
            transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225]),
        ])

    def _preprocess(self, im_crops):
        def _resize(im, size):
            return cv2.resize(im.astype(np.float32) / 255., size)

        im_batch = torch.cat([self.norm(_resize(im, self.size)).unsqueeze(0) for im in im_crops], dim=0).float()
        return im_batch

    def __call__(self, im_crops):
        im_batch = self._preprocess(im_crops)
        with torch.no_grad():
            im_batch = im_batch.to(self.device)
            features = self.net(im_batch)
        return features.cpu().numpy()


if __name__ == '__main__':
    img = cv2.imread("demo.jpg")[:, :, (2, 1, 0)]
    extr = Extractor("checkpoint/ckpt.t7")
    feature = extr(img)
    print(feature.shape)

现在假设有 200 个请求连续进行。为每个请求加载模型的过程使代码运行缓慢。

所以我认为将 pytorch 模型保存在缓存中可能是个好主意。我是这样修改的：

from redis import Redis
import msgpack as msg

r = Redis('111.222.333.444')

class Extractor(object):
    def __init__(self, model_path, use_cuda=True):
        try:
            self.net = msg.unpackb(r.get('REID_CKPT'))
        finally:
            self.net = Net(reid=True)
            self.device = "cuda" if torch.cuda.is_available() and use_cuda else "cpu"
            state_dict = torch.load(model_path, map_location=lambda storage, loc: storage)['net_dict']
            self.net.load_state_dict(state_dict)
            print("Loading weights from {}... Done!".format(model_path))
            self.net.to(self.device)
            packed_net = msg.packb(self.net)
            r.set('REID_CKPT', packed_net)

        self.size = (64, 128)
        self.norm = transforms.Compose([
            transforms.ToTensor(),
            transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225]),
        ])

不幸的是出现了这个错误：

 File "msgpack/_packer.pyx", line 286, in msgpack._cmsgpack.Packer.pack
 File "msgpack/_packer.pyx", line 292, in msgpack._cmsgpack.Packer.pack
 File "msgpack/_packer.pyx", line 289, in msgpack._cmsgpack.Packer.pack
 File "msgpack/_packer.pyx", line 283, in msgpack._cmsgpack.Packer._pack
 TypeError: can not serialize 'Net' object

原因显然是因为它无法将 Net 对象（pytorch nn.Module 类）转换为字节。

如何有效地将 pytorch 模型保存在缓存中（或以某种方式将其保存在 RAM 中）并为每个请求调用它？

谢谢大家。

Answer 1

如果您只需要在 RAM 上保存模型状态，则不需要 Redis。您可以改为将 RAM 挂载为虚拟磁盘并在其中存储模型状态。查看tmpfs。