【动手学深度学习pytorch】学习笔记 8.5. 循环神经网络的从零开始实现

8.5. 循环神经网络的从零开始实现 — 动手学深度学习 2.0.0-beta0 documentation (d2l.ai)

目标：根据用户提供的文本的前缀生成后续文本

知识点：独热编码、梯度剪裁

实现细节：注意 “预热 ”

程序可分4个步骤学习

1 独热编码。读通代码，观察输出。

2 建立RNN模型。

3 使用建立好的RNN模型进行预测。

输出的预测结果惨不忍睹~

4 训练模型：了解梯度剪裁的意义，观察训练过程中 预测值的变化。

前几轮预测值非常差。随着训练次数增加，质量越来越高。

1 独热编码

import math
import torch
from torch import nn
from torch.nn import functional as F
from d2l import torch as d2l

batch_size, num_steps = 32, 35
train_iter, vocab = d2l.load_data_time_machine(batch_size, num_steps)  # 这个函数8.3节有讲解

print(list(vocab.token_to_idx.items())[:10])  # 查看词表的前十项内容
print(vocab.token_freqs[:10])
print('词表长度:', len(vocab))

print(F.one_hot(torch.tensor([0, 2]), len(vocab)))  # 索引为 0 和 2 的 独热向量。

X = torch.arange(10).reshape((2, 5))    # 小批量数据形状是二维张量： （批量大小 2，时间步数 5）
print(X.shape)
print(X)

Y = F.one_hot(X.T, 28)  # 获得形状为 （时间步数 5，批量大小 2 ，词表大小 28）
print(Y.shape)
print(Y)

[('<unk>', 0), (' ', 1), ('e', 2), ('t', 3), ('a', 4), ('i', 5), ('n', 6), ('o', 7), ('s', 8), ('h', 9)]
[(' ', 29927), ('e', 17838), ('t', 13515), ('a', 11704), ('i', 10138), ('n', 9917), ('o', 9758), ('s', 8486), ('h', 8257), ('r', 7674)]
词表长度: 28
tensor([[1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
            [0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,0, 0, 0, 0]])
torch.Size([2, 5])
tensor([[0, 1, 2, 3, 4],
            [5, 6, 7, 8, 9]])
torch.Size([5, 2, 28])
tensor([[[1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,0, 0, 0, 0, 0],
         [0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,0, 0, 0, 0, 0]],

        [[0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,0, 0, 0, 0, 0],
         [0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,0, 0, 0, 0, 0]],

        [[0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
         [0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]],

        [[0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
         [0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]],

        [[0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
         [0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]]])

2 建立RNN模型

import math
import torch
from torch import nn
from torch.nn import functional as F
from d2l import torch as d2l

batch_size, num_steps = 32, 35
train_iter, vocab = d2l.load_data_time_machine(batch_size, num_steps)  # 这个函数8.3节有讲解

X = torch.arange(10).reshape((2, 5))


# [初始化循环神经网络模型的模型参数]
# 隐藏单元数num_hiddens是一个可调的超参数。
# 当训练语言模型时，输入和输出来自相同的词表。 因此，它们具有相同的维度，即词相同的表的大小。
def get_params(vocab_size, num_hiddens, device):
    num_inputs = num_outputs = vocab_size

    def normal(shape):
        return torch.randn(size=shape, device=device) * 0.01

    # 隐藏层参数
    W_xh = normal((num_inputs, num_hiddens))
    W_hh = normal((num_hiddens, num_hiddens))
    b_h = torch.zeros(num_hiddens, device=device)
    # 输出层参数
    W_hq = normal((num_hiddens, num_outputs))
    b_q = torch.zeros(num_outputs, device=device)
    # 附加梯度
    params = [W_xh, W_hh, b_h, W_hq, b_q]
    for param in params:
        param.requires_grad_(True)
    return params


#  循环神经网络模型
def init_rnn_state(batch_size, num_hiddens, device):
    return (torch.zeros((batch_size, num_hiddens), device=device),)


def rnn(inputs, state, params):  # inputs的形状：(时间步数量，批量大小，词表大小)
    W_xh, W_hh, b_h, W_hq, b_q = params
    H, = state
    outputs = []
    for X in inputs:  # X的形状：(批量大小，词表大小)
        H = torch.tanh(torch.mm(X, W_xh) + torch.mm(H, W_hh) + b_h)
        Y = torch.mm(H, W_hq) + b_q
        outputs.append(Y)
    return torch.cat(outputs, dim=0), (H,)


class RNNModelScratch:  # @save
    """从零开始实现的循环神经网络模型"""

    def __init__(self, vocab_size, num_hiddens, device, get_params, init_state, forward_fn):
        self.vocab_size, self.num_hiddens = vocab_size, num_hiddens
        self.params = get_params(vocab_size, num_hiddens, device)
        self.init_state, self.forward_fn = init_state, forward_fn

    def __call__(self, X, state):
        X = F.one_hot(X.T, self.vocab_size).type(torch.float32)
        return self.forward_fn(X, state, self.params)

    def begin_state(self, batch_size, device):
        return self.init_state(batch_size, self.num_hiddens, device)


num_hiddens = 512
net = RNNModelScratch(len(vocab), num_hiddens, d2l.try_gpu(), get_params, init_rnn_state, rnn)

# 检查输出是否具有正确的形状
state = net.begin_state(X.shape[0], d2l.try_gpu())
Y, new_state = net(X.to(d2l.try_gpu()), state)
print(F'Y.shape:{Y.shape}, len(new_state):{len(new_state)}, new_state[0].shape: {new_state[0].shape}' )
# 输出形状是（时间步数 × 批量大小，词表大小）， 而隐状态形状保持不变，即（批量大小，隐藏单元数）

检查输出是否具有正确的形状

Y.shape:torch.Size([10, 28]), len(new_state):1, new_state[0].shape: torch.Size([2, 512])

3 使用建立好的RNN模型进行预测

import math
import torch
from torch import nn
from torch.nn import functional as F
from d2l import torch as d2l

batch_size, num_steps = 32, 35
train_iter, vocab = d2l.load_data_time_machine(batch_size, num_steps)  # 这个函数8.3节有讲解

X = torch.arange(10).reshape((2, 5))


# [初始化循环神经网络模型的模型参数]
# 隐藏单元数num_hiddens是一个可调的超参数。
# 当训练语言模型时，输入和输出来自相同的词表。 因此，它们具有相同的维度，即词相同的表的大小。
def get_params(vocab_size, num_hiddens, device):
    num_inputs = num_outputs = vocab_size

    def normal(shape):
        return torch.randn(size=shape, device=device) * 0.01

    # 隐藏层参数
    W_xh = normal((num_inputs, num_hiddens))
    W_hh = normal((num_hiddens, num_hiddens))
    b_h = torch.zeros(num_hiddens, device=device)
    # 输出层参数
    W_hq = normal((num_hiddens, num_outputs))
    b_q = torch.zeros(num_outputs, device=device)
    # 附加梯度
    params = [W_xh, W_hh, b_h, W_hq, b_q]
    for param in params:
        param.requires_grad_(True)
    return params


#  循环神经网络模型
def init_rnn_state(batch_size, num_hiddens, device):
    return (torch.zeros((batch_size, num_hiddens), device=device),)


def rnn(inputs, state, params):  # inputs的形状：(时间步数量，批量大小，词表大小)
    W_xh, W_hh, b_h, W_hq, b_q = params
    H, = state
    outputs = []
    for X in inputs:  # X的形状：(批量大小，词表大小)
        H = torch.tanh(torch.mm(X, W_xh) + torch.mm(H, W_hh) + b_h)
        Y = torch.mm(H, W_hq) + b_q
        outputs.append(Y)
    return torch.cat(outputs, dim=0), (H,)


class RNNModelScratch:  # @save
    """从零开始实现的循环神经网络模型"""

    def __init__(self, vocab_size, num_hiddens, device, get_params, init_state, forward_fn):
        self.vocab_size, self.num_hiddens = vocab_size, num_hiddens
        self.params = get_params(vocab_size, num_hiddens, device)
        self.init_state, self.forward_fn = init_state, forward_fn

    def __call__(self, X, state):
        X = F.one_hot(X.T, self.vocab_size).type(torch.float32)
        return self.forward_fn(X, state, self.params)

    def begin_state(self, batch_size, device):
        return self.init_state(batch_size, self.num_hiddens, device)


num_hiddens = 512
net = RNNModelScratch(len(vocab), num_hiddens, d2l.try_gpu(), get_params, init_rnn_state, rnn)

# 检查输出是否具有正确的形状
state = net.begin_state(X.shape[0], d2l.try_gpu())
Y, new_state = net(X.to(d2l.try_gpu()), state)
print(F'Y.shape:{Y.shape}, len(new_state):{len(new_state)}, new_state[0].shape: {new_state[0].shape}')


# 输出形状是（时间步数 × 批量大小，词表大小）， 而隐状态形状保持不变，即（批量大小，隐藏单元数）


def predict_ch8(prefix, num_preds, net, vocab, device):  # @save
    """在prefix后面生成新字符"""
    state = net.begin_state(batch_size=1, device=device)
    outputs = [vocab[prefix[0]]]
    get_input = lambda: torch.tensor([outputs[-1]], device=device).reshape((1, 1))

    for y in prefix[1:]:  # 预热期:在此期间模型会自我更新（例如，更新隐状态）， 但不会进行预测
        _, state = net(get_input(), state)
        print(F'vocab[{y}]:', vocab[y])
        outputs.append(vocab[y])

    for _ in range(num_preds):  # 预测num_preds步
        y, state = net(get_input(), state)
        outputs.append(int(y.argmax(dim=1).reshape(1)))

    return ''.join([vocab.idx_to_token[i] for i in outputs])


print(predict_ch8('time traveller ', 10, net, vocab, d2l.try_gpu()))

Y.shape:torch.Size([10, 28]), len(new_state):1, new_state[0].shape: torch.Size([2, 512])
vocab[i]: 5
vocab[m]: 13
vocab[e]: 2
vocab[ ]: 1
vocab[t]: 3
vocab[r]: 10
vocab[a]: 4
vocab[v]: 22
vocab[e]: 2
vocab[l]: 12
vocab[l]: 12
vocab[e]: 2
vocab[r]: 10
vocab[ ]: 1
time traveller hyvmsjb hy

上面的红字，就是预测的结果，惨不忍睹~

4 训练模型：观察训练过程中 预测值的变化。

import math
import torch
from torch import nn
from torch.nn import functional as F
from d2l import torch as d2l

batch_size, num_steps = 32, 35
train_iter, vocab = d2l.load_data_time_machine(batch_size, num_steps)  # 这个函数8.3节有讲解

X = torch.arange(10).reshape((2, 5))


# [初始化循环神经网络模型的模型参数]
# 隐藏单元数num_hiddens是一个可调的超参数。
# 当训练语言模型时，输入和输出来自相同的词表。 因此，它们具有相同的维度，即词相同的表的大小。
def get_params(vocab_size, num_hiddens, device):
    num_inputs = num_outputs = vocab_size

    def normal(shape):
        return torch.randn(size=shape, device=device) * 0.01

    # 隐藏层参数
    W_xh = normal((num_inputs, num_hiddens))
    W_hh = normal((num_hiddens, num_hiddens))
    b_h = torch.zeros(num_hiddens, device=device)
    # 输出层参数
    W_hq = normal((num_hiddens, num_outputs))
    b_q = torch.zeros(num_outputs, device=device)
    # 附加梯度
    params = [W_xh, W_hh, b_h, W_hq, b_q]
    for param in params:
        param.requires_grad_(True)
    return params


#  循环神经网络模型
def init_rnn_state(batch_size, num_hiddens, device):
    return (torch.zeros((batch_size, num_hiddens), device=device),)


def rnn(inputs, state, params):  # inputs的形状：(时间步数量，批量大小，词表大小)
    W_xh, W_hh, b_h, W_hq, b_q = params
    H, = state
    outputs = []
    for X in inputs:  # X的形状：(批量大小，词表大小)
        H = torch.tanh(torch.mm(X, W_xh) + torch.mm(H, W_hh) + b_h)
        Y = torch.mm(H, W_hq) + b_q
        outputs.append(Y)
    return torch.cat(outputs, dim=0), (H,)


class RNNModelScratch:  # @save
    """从零开始实现的循环神经网络模型"""

    def __init__(self, vocab_size, num_hiddens, device, get_params, init_state, forward_fn):
        self.vocab_size, self.num_hiddens = vocab_size, num_hiddens
        self.params = get_params(vocab_size, num_hiddens, device)
        self.init_state, self.forward_fn = init_state, forward_fn

    def __call__(self, X, state):
        X = F.one_hot(X.T, self.vocab_size).type(torch.float32)
        return self.forward_fn(X, state, self.params)

    def begin_state(self, batch_size, device):
        return self.init_state(batch_size, self.num_hiddens, device)


num_hiddens = 512
net = RNNModelScratch(len(vocab), num_hiddens, d2l.try_gpu(), get_params, init_rnn_state, rnn)

# 检查输出是否具有正确的形状
state = net.begin_state(X.shape[0], d2l.try_gpu())
Y, new_state = net(X.to(d2l.try_gpu()), state)


# print(F'Y.shape:{Y.shape}, len(new_state):{len(new_state)}, new_state[0].shape: {new_state[0].shape}')


def predict_ch8(prefix, num_preds, net, vocab, device):  # @save
    """在prefix后面生成新字符"""
    state = net.begin_state(batch_size=1, device=device)
    outputs = [vocab[prefix[0]]]
    get_input = lambda: torch.tensor([outputs[-1]], device=device).reshape((1, 1))

    for y in prefix[1:]:  # 预热期:在此期间模型会自我更新（例如，更新隐状态）， 但不会进行预测
        _, state = net(get_input(), state)
        # print(F'vocab[{y}]:', vocab[y])
        outputs.append(vocab[y])

    for _ in range(num_preds):  # 预测num_preds步
        y, state = net(get_input(), state)
        outputs.append(int(y.argmax(dim=1).reshape(1)))

    return ''.join([vocab.idx_to_token[i] for i in outputs])


def grad_clipping(net, theta):  # @save
    """裁剪梯度"""
    if isinstance(net, nn.Module):
        params = [p for p in net.parameters() if p.requires_grad]
    else:
        params = net.params
    norm = torch.sqrt(sum(torch.sum((p.grad ** 2)) for p in params))
    if norm > theta:
        for param in params:
            param.grad[:] *= theta / norm


def train_epoch_ch8(net, train_iter, loss, updater, device, use_random_iter):
    """训练网络一个迭代周期（定义见第8章）"""
    state, timer = None, d2l.Timer()
    metric = d2l.Accumulator(2)  # 训练损失之和,词元数量
    for X, Y in train_iter:
        if state is None or use_random_iter:
            # 在第一次迭代或使用随机抽样时初始化state
            state = net.begin_state(batch_size=X.shape[0], device=device)
        else:
            if isinstance(net, nn.Module) and not isinstance(state, tuple):
                # state对于nn.GRU是个张量
                state.detach_()
            else:
                # state对于nn.LSTM或对于我们从零开始实现的模型是个张量
                for s in state:
                    s.detach_()
        y = Y.T.reshape(-1)
        X, y = X.to(device), y.to(device)
        y_hat, state = net(X, state)
        l = loss(y_hat, y.long()).mean()
        if isinstance(updater, torch.optim.Optimizer):
            updater.zero_grad()
            l.backward()
            grad_clipping(net, 1)
            updater.step()
        else:
            l.backward()
            grad_clipping(net, 1)
            # 因为已经调用了mean函数
            updater(batch_size=1)
        metric.add(l * y.numel(), y.numel())
    return math.exp(metric[0] / metric[1]), metric[1] / timer.stop()


def train_ch8(net, train_iter, vocab, lr, num_epochs, device, use_random_iter=False):
    """训练模型（定义见第8章）"""
    loss = nn.CrossEntropyLoss()

    # 初始化
    if isinstance(net, nn.Module):
        updater = torch.optim.SGD(net.parameters(), lr)
    else:
        updater = lambda batch_size: d2l.sgd(net.params, lr, batch_size)
    predict = lambda prefix: predict_ch8(prefix, 50, net, vocab, device)
    # 训练和预测
    for epoch in range(num_epochs):
        ppl, speed = train_epoch_ch8(net, train_iter, loss, updater, device, use_random_iter)
        if (epoch + 1) % 10 == 0:
            print(F'epoch{epoch}:', predict('time traveller'))

    print(f'困惑度 {ppl:.1f}, {speed:.1f} 词元/秒 {str(device)}')
    print(predict('time traveller'))
    # print(predict('traveller'))


num_epochs, lr = 500, 1
train_ch8(net, train_iter, vocab, lr, num_epochs, d2l.try_gpu())

前50轮结果：

epoch9: time traveller the the the the the the the the the the the the t
epoch19: time travellere the the the the the the the the the the the the 
epoch29: time traveller the the the the the the the the the the the the t
epoch39: time traveller and the the the the the the the the the the the t
epoch49: time traveller and the the the the the the the the the the the t
epoch59: time traveller and the the the the the the the the the the the t

450-500轮结果：

epoch459: time travelleryou can show black is white by argument said filby
epoch469: time traveller with a slight accession ofcheerfulness really thi
epoch479: time travelleryou can show black is white by argument said filby
epoch489: time traveller for so it will be convenient to speak of himwas e
epoch499: time traveller with a slight accession ofcheerfulness really thi

虽然看不出来啥意思，但起码看上去“像句人话”了~

困惑度 1.0, 18638.9 词元/秒 cpu

标签：vocab,8.5,torch,state,学习,pytorch,num,device,size
来源： https://www.cnblogs.com/hbuwyg/p/16365599.html