Compute accuracy of TIMIT data set on an RNN model - machine-learning

Question

So I am trying to figure out how to compute the accuracy of an RNN model.
I have a batch size of 192, and every frame (TIMIT frame) is of the size of 490*129.
currently, I compute my accuracy as follows:
for batch_x, batch_y, len_batch in val_loader:
batch_x, batch_y, len_batch = batch_x.to(device), batch_y.to(device), len_batch.to(device)
logits = model(batch_x)
loss_val = model.loss(logits, batch_y, len_batch)
preds1 = logits.argmax(dim=-1)
temp1 = batch_y.argmax(dim=-1)
acc1 = preds1.squeeze().eq(temp1).sum().item()
acc_rate1 = 100 * acc1 / (192 * 490)
But the accuracy that I receive seems wrong.
My train:
model.train()
print('number of epochs: {}'.format(args.epochs))
best_test = 1e7
best_validation = 1e7
losses_diagonal = []
for ep in range(1, 2):
init_time = datetime.now()
processed = 0
step = 1
for batch_idx, (batch_x, batch_y, len_batch) in enumerate(train_loader):
batch_x, batch_y, len_batch = batch_x.to(device), batch_y.to(device), len_batch.to(device)
opt.zero_grad()
logits = model(batch_x)
loss = model.loss(logits, batch_y, len_batch)
loss.backward()
if args.clip > 0:
nn.utils.clip_grad_norm_(model.parameters(), args.clip)
opt.step()
processed += len(batch_x)
step += 1
losses_diagonal.append(loss)
print(" batch_idx {}\tLoss: {:.2f} ".format(batch_idx, loss))
print("Epoch {}, LR {:.5f} \tLoss: {:.2f} ".format(ep, opt.param_groups[0]['lr'], loss))
My test:
model.eval()
with torch.no_grad():
for batch_x, batch_y, len_batch in test_loader:
batch_x, batch_y, len_batch = batch_x.to(device), batch_y.to(device), len_batch.to(device)
logits = model(batch_x)
loss_test = model.loss(logits, batch_y, len_batch)
for batch_x, batch_y, len_batch in val_loader:
batch_x, batch_y, len_batch = batch_x.to(device), batch_y.to(device), len_batch.to(device)
logits = model(batch_x)
loss_val = model.loss(logits, batch_y, len_batch)
for batch_x, batch_y, len_batch in val_loader:
batch_x, batch_y, len_batch = batch_x.to(device), batch_y.to(device), len_batch.to(device)
logits = model(batch_x)
loss_val = model.loss(logits, batch_y, len_batch)
preds1 = logits.argmax(dim=-1)
temp1 = batch_y.argmax(dim=-1)
acc1 = preds1.squeeze().eq(temp1).sum().item()
acc_rate1 = 100 * acc1 / (192 * 490)
acc_rate1
print('acc = {}, acc_rate = {}'.format(acc1,acc_rate1))
if loss_val < best_validation:
best_validation = loss_val.item()
best_test = loss_test.item()
print()
print("Val: Loss: {:.2f}\tBest: {:.2f}".format(loss_val, best_validation))
print("Test: Loss: {:.2f}\tBest: {:.2f}".format(loss_test, best_test))
print()
model.train()
I would like to get some help with computing the accuracy of my RNN model, and I am willing to publish some more of my code if it will be required.