import torch
import torch.nn as nn

class TransformerBlock(nn.Module):
    def __init__(self, embed_size, heads, ff_hidden_dim, dropout):
        super().__init__()
        self.attention = nn.MultiheadAttention(embed_dim=embed_size, num_heads=heads, batch_first=True)
        self.norm1 = nn.LayerNorm(embed_size)
        self.norm2 = nn.LayerNorm(embed_size)
        self.ff = nn.Sequential(
            nn.Linear(embed_size, ff_hidden_dim),
            nn.ReLU(),
            nn.Dropout(dropout),
            nn.Linear(ff_hidden_dim, embed_size)
        )
        self.dropout = nn.Dropout(dropout)

    def forward(self, x):
        attn_output, _ = self.attention(x, x, x)
        x = self.norm1(x + self.dropout(attn_output))
        ff_output = self.ff(x)
        x = self.norm2(x + self.dropout(ff_output))
        return x

class TransformerModel(nn.Module):
    def __init__(self, vocab_size, embed_size=512, num_heads=8, hidden_dim=2048, num_layers=6, max_len=512, dropout=0.1):
        super().__init__()
        self.embedding = nn.Embedding(vocab_size, embed_size)
        self.pos_embedding = nn.Parameter(torch.zeros(1, max_len, embed_size))
        self.transformer_blocks = nn.Sequential(
            *[TransformerBlock(embed_size, num_heads, hidden_dim, dropout) for _ in range(num_layers)]
        )
        self.norm = nn.LayerNorm(embed_size)
        self.output = nn.Linear(embed_size, vocab_size)

    def forward(self, x):
        seq_len = x.size(1)
        positions = self.pos_embedding[:, :seq_len, :]
        x = self.embedding(x) + positions
        x = self.transformer_blocks(x)
        x = self.norm(x)
        return self.output(x)