Validation Loss

[NUS-Tim]

I have introduced the validation phase using a custom dataset:

val_dataloader = DataLoader(val_dataset, num_workers=0, batch_size=batch_size, shuffle=False)
# omit some codes
trainer.fit(model, train_dataloader, val_dataloader)

However, it seems the code does not present the loss:

Epoch 2:  95%|█████████▌| 120/126 [04:27<00:13,  2.23s/it, loss=0.00658, v_num=3, train/loss_simple_step=0.0176, train/loss_vlb_step=0.000127, train/loss_step=0.0176, global_step=263.0, train/loss_simple_epoch=0.00721, train/loss_vlb_epoch=0.000521, train/loss_epoch=0.00721]
Validating:  84%|████████▍ | 32/38 [00:36<00:06,  1.13s/it]
Epoch 2:  97%|█████████▋| 122/126 [04:29<00:08,  2.21s/it, loss=0.00658, v_num=3, train/loss_simple_step=0.0176, train/loss_vlb_step=0.000127, train/loss_step=0.0176, global_step=263.0, train/loss_simple_epoch=0.00721, train/loss_vlb_epoch=0.000521, train/loss_epoch=0.00721]
Validating:  89%|████████▉ | 34/38 [00:38<00:04,  1.13s/it]
Epoch 2:  98%|█████████▊| 124/126 [04:32<00:04,  2.20s/it, loss=0.00658, v_num=3, train/loss_simple_step=0.0176, train/loss_vlb_step=0.000127, train/loss_step=0.0176, global_step=263.0, train/loss_simple_epoch=0.00721, train/loss_vlb_epoch=0.000521, train/loss_epoch=0.00721]
Validating:  95%|█████████▍| 36/38 [00:40<00:02,  1.14s/it]

I would appreciate it if anyone could advise me on this, thanks!

[jiajiao272727]

May I ask will it store all loss data in "loss_log.txt" like pix2pix? I also followed the process of the formal tutorial "Train a ControlNet to Control", but only get the model of the newest epoch in the "lightning_logs" folder and sample images in the "image_log" folder. Thank you so much~