diff --git a/src/templates/template-common/config.yaml b/src/templates/template-common/config.yaml
index 0ca2db54..b0055c3e 100644
--- a/src/templates/template-common/config.yaml
+++ b/src/templates/template-common/config.yaml
@@ -1,7 +1,7 @@
 seed: 777
 data_path: ./
-batch_size: 32
-eval_batch_size: 32
+batch_size: 512
+eval_batch_size: 1024
 num_workers: 4
 max_epochs: 20
 use_amp: false
diff --git a/src/templates/template-vision-classification/config.yaml b/src/templates/template-vision-classification/config.yaml
index 5918a3af..c93bcc54 100644
--- a/src/templates/template-vision-classification/config.yaml
+++ b/src/templates/template-vision-classification/config.yaml
@@ -1,3 +1,6 @@
 #::= from_template_common ::#
 lr: 0.0001
 model: resnet18
+momentum: 0.9
+weight_decay: 1e-4
+num_warmup_epochs: 4
diff --git a/src/templates/template-vision-classification/main.py b/src/templates/template-vision-classification/main.py
index fb6f9d0e..31c4c27d 100644
--- a/src/templates/template-vision-classification/main.py
+++ b/src/templates/template-vision-classification/main.py
@@ -44,18 +44,23 @@ def run(local_rank: int, config: Any):
     # donwload datasets and create dataloaders
     dataloader_train, dataloader_eval = setup_data(config)
 
-    # model, optimizer, loss function, device
+    # model, optimizer, loss function, device, lr_scheduler
     device = idist.device()
     model = idist.auto_model(setup_model(config.model))
-    optimizer = idist.auto_optim(optim.Adam(model.parameters(), lr=config.lr))
+    optimizer = idist.auto_optim(
+        optim.SGD(
+            model.parameters(), lr=config.lr, momentum=config.momentum, weight_decay=config.weight_decay, nesterov=True
+        )
+    )
     loss_fn = nn.CrossEntropyLoss().to(device=device)
+    le = len(dataloader_train)
     milestones_values = [
         (0, 0.0),
         (
-            len(dataloader_train),
+            le * config.num_warmup_epochs,
             config.lr,
         ),
-        (config.max_epochs * len(dataloader_train), 0.0),
+        (config.max_epochs * le, 0.0),
     ]
     lr_scheduler = PiecewiseLinear(optimizer, "lr", milestones_values=milestones_values)
 
diff --git a/src/templates/template-vision-classification/trainers.py b/src/templates/template-vision-classification/trainers.py
index d4e8cea6..bd254139 100644
--- a/src/templates/template-vision-classification/trainers.py
+++ b/src/templates/template-vision-classification/trainers.py
@@ -3,7 +3,7 @@
 import ignite.distributed as idist
 import torch
 from ignite.engine import DeterministicEngine, Engine, Events
-from torch.cuda.amp import autocast
+from torch.cuda.amp import autocast, GradScaler
 from torch.nn import Module
 from torch.optim import Optimizer
 from torch.utils.data import DistributedSampler, Sampler
@@ -27,9 +27,10 @@ def train_function(engine: Union[Engine, DeterministicEngine], batch: Any):
             outputs = model(samples)
             loss = loss_fn(outputs, targets)
 
-        loss.backward()
-        optimizer.step()
         optimizer.zero_grad()
+        scaler.scale(loss).backward()
+        scaler.step(optimizer)
+        scaler.update()
 
         train_loss = loss.item()
         engine.state.metrics = {
@@ -45,6 +46,8 @@ def train_function(engine: Union[Engine, DeterministicEngine], batch: Any):
 
     trainer = Engine(train_function)
     #::: } :::#
+    
+    scaler = GradScaler(enabled=config.use_amp)
 
     # set epoch for distributed sampler
     @trainer.on(Events.EPOCH_STARTED)