out

DST/DST/config.py

-data_path = '/home/marco/projects/projects_data/DST/data/'
+data_path = '/storage/DSIP/DST/data/'

scripts/training_class.py

@@ -225,10 +225,10 @@ n_out_i = 8
 before = BEFORE
 nvars = data_in_scaled.shape[-1]
 
-# dst_net = DSTnet(nvars, nhidden_i, nhidden_o, n_out_i, before, AFTER).to(device)
+dst_net = DSTnet(nvars, nhidden_i, nhidden_o, n_out_i, before, AFTER).to(device)
 print(dst_net)
 
-num_epochs = 2000
+num_epochs = 10000
 lr = 1e-5
 optimizer = torch.optim.Adam(dst_net.parameters(), lr=lr)#, weight_decay=1e-5)
 

scripts/training_reg_class.py

@@ -243,8 +243,8 @@ nvars = data_in_scaled.shape[-1]
 dst_net = DSTnet(nvars, nhidden_i, nhidden_o, n_out_i, before, AFTER).to(device)
 print(dst_net)
 
-num_epochs = 2000
-lr = 1e-4
+num_epochs = 10000
+lr = 1e-5
 optimizer = torch.optim.Adam(dst_net.parameters(), lr=lr, weight_decay=1e-5)
 
 history_tr = np.zeros((num_epochs, 3))
@@ -352,6 +352,7 @@ for epoch in range(num_epochs):
         dataset_tr = Dataset(data_in_scaled[ixs_tr], data_out_scaled[ixs_tr], data_out_c[ixs_tr], weights)
         # data_loader_tr = utils_data.DataLoader(dataset_tr, batch_size=BATCH_SIZE, num_workers = 4, shuffle=False, sampler = sampler)
         data_loader_tr = utils_data.DataLoader(dataset_tr, batch_size=BATCH_SIZE, num_workers = 4, shuffle=True)
+        loss_fc= nn.CrossEntropyLoss(weight = weights_c)
 
 torch.save(dst_net.state_dict(), os.path.join('/home/mcristofo/DST/models','dst_regr_class.pth'))