add spanish qa answer

app.py

@@ -96,22 +96,25 @@ def generate_summary(text):
     summary = tokenizer.decode(summary_ids[0], skip_special_tokens=True)
     return summary
 
+
 # QA
-# Cargar el modelo para preguntas y respuestas
-qa_model_name = "MaRiOrOsSi/t5-base-finetuned-question-answering"
-qa_tokenizer = AutoTokenizer.from_pretrained(qa_model_name)
-qa_model = AutoModelForSeq2SeqLM.from_pretrained(qa_model_name)
+device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
+ckpt = 'mrm8488/spanish-t5-small-sqac-for-qa'
+qa_tokenizer = AutoTokenizer.from_pretrained(ckpt)
+qa_model = T5ForConditionalGeneration.from_pretrained(ckpt).to(device)
 
 
 def generate_question_response(question, context):
-    # Formar el input para el modelo de preguntas y respuestas
-    input_text = f"question: {question} context: {context}"
-    encoded_input = qa_tokenizer(input_text, return_tensors='pt', max_length=1024, truncation=True)
-    output = qa_model.generate(input_ids=encoded_input['input_ids'], attention_mask=encoded_input['attention_mask'])
-    response_en = qa_tokenizer.decode(output[0], skip_special_tokens=True)
-    translator = Translator()
-    translated_response = translator.translate(response_en, dest='es').text
-    return f'Respuesta: {translated_response}'
+    input_text = 'question: %s  context: %s' % (question, context)
+    features = tokenizer([input_text], padding='max_length', truncation=True, max_length=512, return_tensors='pt')
+    output = qa_model.generate(
+        input_ids=features['input_ids'].to(device),
+        attention_mask=features['attention_mask'].to(device),
+        max_length=200,  # Permite respuestas más largas
+        temperature=1.0  # Ajusta la temperatura
+    )
+
+    return qa_tokenizer.decode(output[0], skip_special_tokens=True)
 
 
 class SummarizerAndQA: