model load in train_mtrcnet_cl.py

[kubicndmr]

thank you for sharing your great work.

what is the pre-trained fully connected network in line 395?

model = multi_lstm_p2t()
model.share = model_old.module.share
model.lstm = model_old.module.lstm
model.fc = model_old.module.fc
model.fc2 = model_old.module.fc2
model.fc3 = model_old.module.fc3
model = DataParallel(model)

for param in model.module.fc_p2t.parameters():
    param.requires_grad = False
model.module.fc_p2t.load_state_dict(torch.load(
    "fc_epoch_25_length_4_opt_1_mulopt_1_flip_0_crop_1_batch_800_train1_9951_train2_9713_val1_9686_val2_7867_p2t.pth"))

[wy7810]

This model stores the parameters of the fully connected layer, which is the mapping matrix mentioned in the article, do you have the code to train the parameters of this fully connected layer?