Update app.py

app.py

@@ -589,41 +589,43 @@ dir_model='wasmdashai/vits-ar-huba-fine'
 wandb.login(key= "782b6a6e82bbb5a5348de0d3c7d40d1e76351e79")
 wandb.init(project= 'AZ',config = training_args.to_dict())
 
-for i in range(10000):
-# model.train(True)
-    print(f'clcye epochs ={i}')
-    model=VitsModel.from_pretrained(dir_model,token=token).to(device)
-    # model.setMfA(monotonic_align.maximum_path)
-    #dir_model_save=dir_model+'/vend'
-
-
-    trainer_to_cuda(model,
-                    ctrain_datasets = ctrain_datasets,
-                    eval_dataset = eval_dataset,
-                    full_generation_dataset = ctrain_datasets[0][0],
-                    feature_extractor = VitsFeatureExtractor(),
-                    training_args = training_args,
-                    full_generation_sample_index= -1,
-                    project_name = "AZ",
-                    wandbKey = "782b6a6e82bbb5a5348de0d3c7d40d1e76351e79",
-                    is_used_text_encoder=True,
-                    is_used_posterior_encode=True,
-                #    dict_state_grad_loss=sgl,
-                    nk=50,
-                        path_save_model=dir_model,
-                    maf=model.monotonic_align_max_path,
-
-                    n_back_save_model=3000,
-                    start_speeker=0,
-                    end_speeker=1,
-                    n_epoch=i*training_args.num_train_epochs,
 
 
-                                  )
-
 def greet(text,id):
     global GK 
     b=int(id)
+    for i in range(10000):
+# model.train(True)
+        print(f'clcye epochs ={i}')
+        yield f'clcye epochs ={i}'
+        model=VitsModel.from_pretrained(dir_model,token=token).to(device)
+        # model.setMfA(monotonic_align.maximum_path)
+        #dir_model_save=dir_model+'/vend'
+    
+    
+        trainer_to_cuda(model,
+                        ctrain_datasets = ctrain_datasets,
+                        eval_dataset = eval_dataset,
+                        full_generation_dataset = ctrain_datasets[0][0],
+                        feature_extractor = VitsFeatureExtractor(),
+                        training_args = training_args,
+                        full_generation_sample_index= -1,
+                        project_name = "AZ",
+                        wandbKey = "782b6a6e82bbb5a5348de0d3c7d40d1e76351e79",
+                        is_used_text_encoder=True,
+                        is_used_posterior_encode=True,
+                    #    dict_state_grad_loss=sgl,
+                        nk=50,
+                            path_save_model=dir_model,
+                        maf=model.monotonic_align_max_path,
+    
+                        n_back_save_model=3000,
+                        start_speeker=0,
+                        end_speeker=1,
+                        n_epoch=i*training_args.num_train_epochs,
+    
+    
+                                      )
     while True:
         GK+=1
         texts=[text]*b