Loss values are not reducing

[NamburiSrinath]

Hi,

I am trying to execute the repo with my own data and the loss values are not reducing as expected. Here's the code I modified (just added 2 lines)

epoch_loss = []
no_of_batches = len(ssl_dataloader)
print("No of batches the backbone is trained is {0}".format(no_of_batches))
for _ in tqdm(range(epochs_for_ssl)):
    total_loss = 0
    for batch, _, _ in ssl_dataloader:
        images = batch.to(device)
        loss = learner(images) # if positive pixel pairs is equal to zero, the loss is equal to the instance level loss
        opt.zero_grad()
        loss.backward()
        total_loss += loss.item()
        opt.step()
        learner.update_moving_average() # update moving average of target encoder

    total_loss/=no_of_batches
    epoch_loss.append(total_loss)
    print(total_loss)

# after much training, save the improved model for testing on downstream task
torch.save(resnet, 'improved-resnet-ssl-50epochs.pt')

And here's the plot

I am not sure why the loss is increasing. Should I need to increase the number of epochs and see what will happen? Or is there any bug in my code itself?

Any help would be greatly appreciated Thanks a lot

[NamburiSrinath]

When I execute the code given by @lucidrains, still the loss value is increasing

Here's the code:

import torch
from pixel_level_contrastive_learning import PixelCL
from torchvision import models
from tqdm import tqdm

resnet = models.resnet50(pretrained=True)

learner = PixelCL(
    resnet,
    image_size = 256,
    hidden_layer_pixel = 'layer4',  # leads to output of 8x8 feature map for pixel-level learning
    hidden_layer_instance = -2,     # leads to output for instance-level learning
    projection_size = 256,          # size of projection output, 256 was used in the paper
    projection_hidden_size = 2048,  # size of projection hidden dimension, paper used 2048
    moving_average_decay = 0.99,    # exponential moving average decay of target encoder
    ppm_num_layers = 1,             # number of layers for transform function in the pixel propagation module, 1 was optimal
    ppm_gamma = 2,                  # sharpness of the similarity in the pixel propagation module, already at optimal value of 2
    distance_thres = 0.7,           # ideal value is 0.7, as indicated in the paper, which makes the assumption of each feature map's pixel diagonal distance to be 1 (still unclear)
    similarity_temperature = 0.3,   # temperature for the cosine similarity for the pixel contrastive loss
    alpha = 1.,                      # weight of the pixel propagation loss (pixpro) vs pixel CL loss
    use_pixpro = True,               # do pixel pro instead of pixel contrast loss, defaults to pixpro, since it is the best one
    cutout_ratio_range = (0.6, 0.8)  # a random ratio is selected from this range for the random cutout
).cuda()

opt = torch.optim.Adam(learner.parameters(), lr=1e-4)

def sample_batch_images():
    return torch.randn(10, 3, 256, 256).cuda()

epoch_loss = []
for _ in tqdm(range(1000)):
    images = sample_batch_images()
    loss = learner(images) # if positive pixel pairs is equal to zero, the loss is equal to the instance level loss

    opt.zero_grad()
    loss.backward()
    print(loss.item())
    opt.step()
    learner.update_moving_average() # update moving average of target encoder
    epoch_loss.append(loss.item())

# after much training, save the improved model for testing on downstream task
torch.save(resnet, 'improved-resnet.pt')
plot_loss_vs_epochs_ssl(epoch_loss, 'Loss for backbone model (trained using SSL) vs No of Epochs')

I changed the iterations to 1000, and added this line at the end: epoch_loss.append(loss.item()) which I used to plot the loss

Here's the plot