Update app.py

app.py

@@ -34,7 +34,8 @@ base_model = AutoModelForCausalLM.from_pretrained(base_model_name,return_dict=Tr
 #base_model = AutoModelForCausalLM.from_pretrained(base_model_name, return_dict=True, device_map = {"":0}, attn_implementation = attn_implementation,).eval()
 
 tokenizer = AutoTokenizer.from_pretrained(base_model_name, max_length = max_seq_length)
-ft_model = PeftModel.from_pretrained(base_model, sft_model1)
+sft_model = sft_model1
+ft_model = PeftModel.from_pretrained(base_model, sft_model)
 model = ft_model.merge_and_unload()
 model.save_pretrained(".")
 #model.to('cuda')
@@ -69,7 +70,7 @@ stopping_criteria_list = StoppingCriteriaList([stopping_criteria])
 
 def generate_text(modelin, prompt, context, max_length=2100):
   print('Modelo es: '+modelin)
-  if (modelin is not sft_model):
+  if (modelin != sft_model):
       sft_model = modelin
       ft_model = PeftModel.from_pretrained(base_model, sft_model)
       model = ft_model.merge_and_unload()
@@ -91,7 +92,11 @@ def generate_text(modelin, prompt, context, max_length=2100):
 
 def mostrar_respuesta(modelo, pregunta, contexto):
     try:
-      res= generate_text(modelo, pregunta, contexto, max_length=500)
+      print('Modelo: '+str(modelo))
+      print('Pregunta: '+str(pregunta))
+      print('Contexto: '+str(contexto))
+      res= generate_text(modelo, pregunta, contexto, max_length=500)    
+      print('Respuesta: '+str(contexto))
       return str(res)
     except Exception as e:
       return str(e)