RNN وLSTM وGRU — نمذجة التسلسلات

import torch
import torch.nn as nn
from torch.utils.data import TensorDataset, DataLoader

# ── Sample sequential dataloader ───────────────────────────────────────
# Shape: (n_samples, seq_len, features) → predict next value
X_seq = torch.randn(500, 20, 10)          # 500 samples, seq_len=20, 10 features
y_seq = torch.randn(500)                   # 500 scalar targets
dataloader = DataLoader(TensorDataset(X_seq, y_seq), batch_size=32, shuffle=True)

class LSTMForecaster(nn.Module):
    def __init__(self, input_dim, hidden_dim, num_layers, output_dim, dropout=0.2):
        super().__init__()
        self.lstm = nn.LSTM(
            input_dim, hidden_dim, num_layers,
            batch_first=True, dropout=dropout,
            bidirectional=False
        )
        self.fc = nn.Linear(hidden_dim, output_dim)
        self.dropout = nn.Dropout(dropout)

    def forward(self, x, h0=None, c0=None):
        # x: (batch, seq_len, features)
        out, (hn, cn) = self.lstm(x, (h0, c0) if h0 is not None else None)
        # Use last timestep's output
        return self.fc(self.dropout(out[:, -1, :]))

# Training with teacher forcing + scheduled sampling
model = LSTMForecaster(input_dim=10, hidden_dim=128, num_layers=2, output_dim=1)
optimizer = torch.optim.Adam(model.parameters(), lr=1e-3)

# Gradient clipping is ESSENTIAL for RNN training
for x, y in dataloader:
    pred = model(x)
    loss = nn.MSELoss()(pred.squeeze(), y)
    optimizer.zero_grad()
    loss.backward()
    nn.utils.clip_grad_norm_(model.parameters(), max_norm=1.0)
    optimizer.step()