RotatE-WN18RR-single-gpu-adv || RotatE-WN18RR-single-gpu-adv-wandb || RotatE-WN18RR-single-gpu-adv-hpo

RotatE-WN18RR-single-gpu-adv-wandb

备注

created by LuYF-Lemon-love <luyanfeng_nlp@qq.com> on May 7, 2023

备注

updated by LuYF-Lemon-love <luyanfeng_nlp@qq.com> on May 15, 2024

备注

last run by LuYF-Lemon-love <luyanfeng_nlp@qq.com> on May 15, 2024

这一部分介绍如何用一个 GPU 在 WN18RR 知识图谱上训练 RotatE [SDNT19]，使用 wandb 记录实验结果。

导入数据

pybind11-OpenKE 有 1 个工具用于导入数据: pybind11_ke.data.KGEDataLoader。

from pybind11_ke.utils import WandbLogger
from pybind11_ke.data import KGEDataLoader, UniSampler, TradTestSampler
from pybind11_ke.module.model import RotatE
from pybind11_ke.module.loss import SigmoidLoss
from pybind11_ke.module.strategy import NegativeSampling
from pybind11_ke.config import Trainer, Tester

首先初始化 pybind11_ke.utils.WandbLogger 日志记录器，它是对 wandb 初始化操作的一层简单封装。

wandb_logger = WandbLogger(
    project="pybind11-ke",
    name="RotatE-WN18RR",
    config=dict(
            in_path = '../../benchmarks/WN18RR/',
            batch_size = 2000,
            neg_ent = 64,
            test = True,
            test_batch_size = 10,
            num_workers = 16,
            dim = 1024,
            margin = 6.0,
            epsilon = 2.0,
            adv_temperature = 2,
            regul_rate = 0.0,
        use_tqdm = False,
            use_gpu = True,
            device = 'cuda:1',
            epochs = 6000,
            lr = 2e-5,
            opt_method = 'adam',
            valid_interval = 100,
            log_interval = 100,
            save_interval = 100,
            save_path = '../../checkpoint/rotate.pth'
    )
)

config = wandb_logger.config

pybind11-OpenKE 提供了很多数据集，它们很多都是 KGE 原论文发表时附带的数据集。 pybind11_ke.data.KGEDataLoader 包含 in_path 用于传递数据集目录。

# dataloader for training
dataloader = KGEDataLoader(
    in_path = config.in_path,
    batch_size = config.batch_size,
    neg_ent = config.neg_ent,
    test = config.test,
    test_batch_size = config.test_batch_size,
    num_workers = config.num_workers,
    train_sampler = UniSampler,
    test_sampler = TradTestSampler
)

---

导入模型

pybind11-OpenKE 提供了很多 KGE 模型，它们都是目前最常用的基线模型。我们下面将要导入 pybind11_ke.module.model.RotatE，它将实体表示成复数向量，关系建模为复数向量空间的旋转。

# define the model
rotate = RotatE(
    ent_tol = dataloader.get_ent_tol(),
    rel_tol = dataloader.get_rel_tol(),
    dim = config.dim,
    margin = config.margin,
    epsilon = config.epsilon,
)

---

损失函数

我们这里使用了逻辑损失函数：pybind11_ke.module.loss.SigmoidLoss， pybind11_ke.module.strategy.NegativeSampling 对 pybind11_ke.module.loss.SigmoidLoss 进行了封装，加入权重衰减等额外项。

# define the loss function
model = NegativeSampling(
    model = rotate,
    loss = SigmoidLoss(adv_temperature = config.adv_temperature),
    regul_rate = config.regul_rate
)

---

训练模型

pybind11-OpenKE 将训练循环包装成了 pybind11_ke.config.Trainer， 可以运行它的 pybind11_ke.config.Trainer.run() 函数进行模型学习； 也可以通过传入 pybind11_ke.config.Tester， 使得训练器能够在训练过程中评估模型；pybind11_ke.config.Tester 使用 pybind11_ke.data.TestDataLoader 作为数据采样器。

# test the model
tester = Tester(model = rotate, data_loader = dataloader, use_tqdm = config.use_tqdm,
                use_gpu = config.use_gpu, device = config.device)

# train the model
trainer = Trainer(model = model, data_loader = dataloader.train_dataloader(), epochs = config.epochs,
    lr = config.lr, opt_method = config.opt_method, use_gpu = config.use_gpu, device = config.device,
    tester = tester, test = config.test, valid_interval = config.valid_interval,
    log_interval = config.log_interval, save_interval = config.save_interval,
    save_path = config.save_path, use_wandb = True)
trainer.run()

# close your wandb run
wandb_logger.finish()

备注

上述代码的运行日志可以从 此处 下载。

备注

上述代码的运行报告可以从 此处 下载。

---