dataset pred for diann

diann_lib_processing.py

@@ -65,7 +65,7 @@ if __name__ =='__main__':
     model = ModelTransformer(encoder_ff=args.encoder_ff, decoder_rt_ff=args.decoder_rt_ff,
                              n_head=args.n_head, encoder_num_layer=args.encoder_num_layer,
                              decoder_rt_num_layer=args.decoder_rt_num_layer, drop_rate=args.drop_rate,
-                             embedding_dim=args.embedding_dim, acti=args.activation, norm=args.norm_first)
+                             embedding_dim=args.embedding_dim, acti=args.activation, norm=args.norm_first, seq_length=30)
 
     if torch.cuda.is_available():
         model = model.cuda()

main.py

@@ -103,7 +103,8 @@ def main(args):
     model = ModelTransformer(encoder_ff=args.encoder_ff, decoder_rt_ff=args.decoder_rt_ff,
                               n_head=args.n_head, encoder_num_layer=args.encoder_num_layer,
                               decoder_rt_num_layer=args.decoder_rt_num_layer, drop_rate=args.drop_rate,
-                              embedding_dim=args.embedding_dim, acti=args.activation, norm=args.norm_first)
+                              embedding_dim=args.embedding_dim, acti=args.activation, norm=args.norm_first,
+                             seq_length=30)
 
     if args.model_weigh is not None :
         model.load_state_dict(torch.load(args.model_weigh, weights_only=True))

model/model.py

@@ -38,7 +38,7 @@ class PositionalEncoding(nn.Module):
 class ModelTransformer(nn.Module):
 
     def __init__(self, drop_rate=0.1, embedding_dim=128, nb_aa=22, regressor_layer_size_rt=512, decoder_rt_ff=512,
-                 n_head=1, seq_length=25, encoder_ff=512, encoder_num_layer=1, decoder_rt_num_layer=1, acti='relu',
+                 n_head=1, seq_length=30, encoder_ff=512, encoder_num_layer=1, decoder_rt_num_layer=1, acti='relu',
                  norm=False):
         self.seq_length = seq_length
         self.nb_aa = nb_aa