def train(model, iterator, optimizer, criterion, clip):
model.train()
epoch_loss = 0
for batch in iterator:
src = batch[0]
trg = batch[1]
src = src.type(torch.LongTensor).to(device)
trg = trg.type(torch.LongTensor).to(device)
optimizer.zero_grad()
output, _ = model(src, trg[:, :-1])
# output = [batch size, trg len - 1, output dim]
# trg = [batch size, trg len]
output_dim = output.shape[-1]
output = output.contiguous().view(-1, output_dim)
trg = trg[:, 1:].contiguous().view(-1)
# output = [batch size * trg len - 1, output dim]
# trg = [batch size * trg len - 1]
loss = criterion(output, trg)
loss.backward()
torch.nn.utils.clip_grad_norm_(model.parameters(), clip)
optimizer.step()
print(loss.item())
# Deleting Used Tensors
# del src
# del trg
# gc.collect()
epoch_loss += loss.item()
return epoch_loss / len(iterator)
def train(model, iterator, optimizer, criterion, clip):
model.train()
epoch_loss = 0
for i, batch in enumerate(iterator):
src = batch.src
trg = batch.trg
optimizer.zero_grad()
output, _ = model(src, trg[:, :-1])
# output = [batch size, trg len - 1, output dim]
# trg = [batch size, trg len]
output_dim = output.shape[-1]
output = output.contiguous().view(-1, output_dim)
trg = trg[:, 1:].contiguous().view(-1)
# output = [batch size * trg len - 1, output dim]
# trg = [batch size * trg len - 1]
loss = criterion(output, trg)
loss.backward()
torch.nn.utils.clip_grad_norm_(model.parameters(), clip)
optimizer.step()
print(loss.item())
epoch_loss += loss.item()
return epoch_loss / len(iterator)
def count_parameters(model):
return sum(p.numel() for p in model.parameters() if p.requires_grad)
ENC_DROPOUT, device)
dec = model.Decoder(OUTPUT_DIM, EMB_DIM, HID_DIM, DEC_LAYERS, DEC_KERNEL_SIZE,
DEC_DROPOUT, TRG_PAD_IDX, device)
model = model.CNNSeq2Seq(enc, dec).to(device)
# For checking num. of trainable parameters in Model
def count_parameters(model):
return sum(p.numel() for p in model.parameters() if p.requires_grad)
# print(f'The model has {count_parameters(model):,} trainable parameters')
# Setting up Optimizer and Loss function for Training
optimizer = optim.Adam(model.parameters(), lr=config.Learning_Rate)
# scheduler = StepLR(optimizer, step_size=3, gamma=0.1)
criterion = nn.CrossEntropyLoss(ignore_index=TRG_PAD_IDX)
# Comment Out exit(0) Only in case to Train the model
# exit(0)
# Defining Training Step
def train(model, iterator, optimizer, criterion, clip):
model.train()
epoch_loss = 0
for batch in iterator:
src = batch[0]
enc = model.Encoder(INPUT_DIM, EMB_DIM, HID_DIM, ENC_LAYERS, ENC_KERNEL_SIZE,
ENC_DROPOUT, device)
dec = model.Decoder(OUTPUT_DIM, EMB_DIM, HID_DIM, DEC_LAYERS, DEC_KERNEL_SIZE,
DEC_DROPOUT, TRG_PAD_IDX, device)
model = model.CNNSeq2Seq(enc, dec).to(device)
# For checking num. of trainable parameters in Model
def count_parameters(model):
return sum(p.numel() for p in model.parameters() if p.requires_grad)
# print(f'The model has {count_parameters(model):,} trainable parameters')
optimizer = optim.Adam(model.parameters())
criterion = nn.CrossEntropyLoss(ignore_index=TRG_PAD_IDX)
def train(model, iterator, optimizer, criterion, clip):
model.train()
epoch_loss = 0
for i, batch in enumerate(iterator):
src = batch.src
trg = batch.trg
optimizer.zero_grad()
output, _ = model(src, trg[:, :-1])