import math
from glob import glob
import numpy as np
import paddle
import paddle.nn as nn
import paddle.nn.functional as F
import pandas as pd
from tqdm import tqdm
from sklearn.metrics import f1_score,confusion_matrix
class MaxState(paddle.nn.Layer):
def __init__(self, hidden_dim, heads, win):
super(MaxState, self).__init__()
assert hidden_dim % heads == 0, "Hidden size must be divisible by the number of heads."
self.head_size = hidden_dim // heads
self.head = paddle.nn.Linear(hidden_dim, hidden_dim, bias_attr=False)
self.head_num = heads
self.win = win
self.hidden = hidden_dim
self.mask = paddle.triu(paddle.ones([win, win]))
def forward(self, input_data, state=None):
b, s, k, h, w = input_data.shape[0], input_data.shape[1], self.head_num, self.head_size, self.win
window = paddle.ones([1, w])
out = self.head(input_data)
out = out.unsqueeze(-1) @ window
out = out.transpose([0, 2, 1, 3])
one_list = []
if state is None:
state = paddle.ones([out.shape[0], out.shape[1], 1, 1]) * float("-inf")
for i in range(0, s, w):
j = w + i
one = out[:, :, i:j]
_, _, r, c = one.shape
if r != self.win:
one = paddle.where(self.mask[:r, :], one, paddle.to_tensor(-float('inf')))
else:
one = paddle.where(self.mask, one, paddle.to_tensor(-float('inf')))
one = paddle.concat([one, state @ window], axis=2)
state = paddle.max(one, axis=2, keepdim=True)
one = state.reshape([b, k, h, w])
state = state[..., -1:]
if r != self.win:
one = one[..., :r]
one = one.transpose([0, 3, 1, 2])
one_list.append(one)
out = paddle.concat(one_list, 1)
out = out.reshape([b, s, -1])
return out, state
class FeedForward(nn.Layer):
def __init__(self, hidden_size):
super(FeedForward, self).__init__()
self.ffn1 = nn.Linear(hidden_size, hidden_size * 2)
self.ffn2 = nn.Linear(hidden_size * 2, hidden_size)
self.gate = nn.Linear(hidden_size, hidden_size * 2)
self.relu = nn.Silu()
def forward(self, x):
x1 = self.ffn1(x)
x2 = self.relu(self.gate(x))
x = x1 * x2
x = self.ffn2(x)
return x
class RMSNorm(nn.Layer):
def __init__(self, dim, eps: float = 1e-6):
super(RMSNorm, self).__init__()
self.eps = eps
self.fc = paddle.create_parameter(shape=[dim], dtype='float32',
default_initializer=nn.initializer.Constant(value=1.0))
def norm(self, x):
return x * paddle.rsqrt(x.pow(2).mean(-1, keepdim=True) + self.eps)
def forward(self, x):
output = self.norm(x)
return output * self.fc
class GPTDecoderLayer(nn.Layer):
def __init__(self, hidden_size, num_heads):
super(GPTDecoderLayer, self).__init__()
# self.self_attention = MaskMultiHeadAttention(hidden_size, num_heads)
self.self_attention = MaxState(hidden_size, num_heads, 8)
self.ffn = FeedForward(hidden_size)
self.norm = nn.LayerNorm(hidden_size)
self.norm1 = RMSNorm(hidden_size)
def forward(self, x, state=None, seq_len=None):
x1, state = self.self_attention(x, state) # Self-Attention with residual connection
x = x1 + x
x = self.norm(x)
x = self.ffn(x) + x # Feed-Forward with residual connection
x = self.norm1(x)
return x, state
class PositionalEncoding(nn.Layer):
def __init__(self, d_model, max_len=5000):
super(PositionalEncoding, self).__init__()
# Create a long enough Paddle array to hold position encodings for the maximum sequence length
position = paddle.arange(max_len).unsqueeze(1).astype("float32")
# Create a constant 'pe' matrix with the same size as the embedding matrix
div_term = paddle.exp(paddle.arange(0, d_model, 2) * (-math.log(10000.0) / d_model))
pe = paddle.zeros([max_len, d_model])
pe[:, 0::2] = paddle.sin(position * div_term)
pe[:, 1::2] = paddle.cos(position * div_term)
self.pe = pe.unsqueeze(0) # Shape: [1, max_len, d_model]
# Register 'pe' as a buffer (non-trainable parameter)
def forward(self, x, seq_len=None):
# x is of shape [batch_size, seq_len, d_model]
if seq_len is None:
seq_len = x.shape[1]
return x + self.pe[:, :seq_len, :]
else:
return x + self.pe[:, seq_len - 1:seq_len, :]
# %%
def sinusoidal_position_embedding(max_len, output_dim):
# (max_len, 1)
position = paddle.arange(0, max_len, dtype="float32").unsqueeze(-1)
# (output_dim//2)
ids = paddle.arange(0, output_dim // 2, dtype="float32") # 即公式里的i, i的范围是 [0,d/2]
theta = 10000 ** (-2 * ids / output_dim)
# (max_len, output_dim//2)
embeddings = position * theta # 即公式里的：pos / (10000^(2i/d))
sin_embeddings = paddle.sin(embeddings)
cos_embeddings = paddle.cos(embeddings)
return sin_embeddings, cos_embeddings
def rope(q, sin_em, cos_em, seq_len=None):
if seq_len is None:
sin_em = sin_em[:q.shape[2]]
cos_em = cos_em[:q.shape[2]]
else:
sin_em = sin_em[seq_len - 1:seq_len]
cos_em = cos_em[seq_len - 1:seq_len]
q1 = q.reshape([q.shape[0], q.shape[1], q.shape[2], -1, 2])[..., 1]
q2 = q.reshape([q.shape[0], q.shape[1], q.shape[2], -1, 2])[..., 0]
# 奇数负值*sin_em+偶数正值*cos_em 奇数正值*cos_em+偶数正值*sin_em
q3 = paddle.stack([-q1 * sin_em + q2 * cos_em, q1 * cos_em + q2 * sin_em], -1)
q = q3.reshape(q.shape) # reshape后就是正负交替了
return q
class CvEm(nn.Layer):
def __init__(self, hidden_size):
super(CvEm, self).__init__()
self.embedding = nn.Conv1D(3, hidden_size, 3, padding=2)
def forward(self, x):
x = self.embedding(x)
return x.transpose([0, 2, 1])
class GPT(nn.Layer):
def __init__(self, vocab_size, hidden_size, num_heads, num_layers):
super(GPT, self).__init__()
self.embedding = CvEm(hidden_size)
self.decoder_layers = nn.LayerList([GPTDecoderLayer(hidden_size, num_heads) for _ in range(num_layers)])
self.fc = nn.Linear(hidden_size, vocab_size, bias_attr=False)
self.sin_em, self.cos_em = sinusoidal_position_embedding(50000, hidden_size // num_heads // 2)
self.layer_nor = paddle.nn.LayerNorm(hidden_size)
def forward(self, x, state=None, seq_len=None):
x = self.embedding(x)
# x = self.position_embedding(x, seq_len)
if state is None:
state = [None] * len(self.decoder_layers)
i = 0
x = rope(x.reshape([x.shape[0], x.shape[1], -1, self.sin_em.shape[1] * 2]).transpose([0, 2, 1, 3]),
self.sin_em,
self.cos_em, seq_len).transpose([0, 2, 1, 3]).reshape(x.shape) + x
for decoder_layer in self.decoder_layers:
x1, state[i] = decoder_layer(x, state[i])
x = x1 + x
i += 1
out = self.fc(self.layer_nor(paddle.max(x, 1)))
return out, state
这段代码实现了一个基于PaddlePaddle的GPT（Generative Pre-trained Transformer）模型。主要包括以下几个部分：
引入依赖库：引入了一些需要使用的库，包括math、glob、numpy、paddle等。
定义MaxState类：这是一个自定义的PaddlePaddle层，用于计算输入数据的最大状态。它使用了自注意力机制（self-attention）和位置编码（positional encoding）来计算输入数据的最大状态。
定义FeedForward类：这是一个前馈神经网络层，用于对输入数据进行非线性变换。
定义RMSNorm类：这是一个归一化层，用于对输入数据进行归一化处理。
定义GPTDecoderLayer类：这是一个GPT解码器层，包括自注意力机制、前馈神经网络和归一化层。
定义PositionalEncoding类：这是一个位置编码层，用于为输入数据添加位置信息。
定义sinusoidal_position_embedding函数：这是一个用于生成正弦位置编码和余弦位置编码的函数。
定义rope函数：这是一个用于将输入数据与位置编码相结合的函数。
定义CvEm类：这是一个卷积神经网络层，用于将输入数据进行卷积操作。
定义GPT类：这是一个GPT模型的定义，包括嵌入层、解码器层和全连接层。
forward函数：这是GPT模型的前向传播函数，用于计算输出结果。
总体来说，这段代码实现了一个基于PaddlePaddle的GPT模型，并提供了相应的层和函数用于构建和训练模型。