Update app.py

app.py

@@ -45,7 +45,7 @@ def load_qa_model():
         qa_tokenizer = AutoTokenizer.from_pretrained("distilbert-base-cased-distilled-squad")
         logger.info("QA model loaded.")
 
-@spaces.GPU
+@spaces.GPU(duration=120)
 def transcribe_audio(audio_file, translate, model_size, use_diarization):
     language_segments, final_segments = process_audio(audio_file, translate=translate, model_size=model_size, use_diarization=use_diarization)
     
@@ -70,7 +70,7 @@ def transcribe_audio(audio_file, translate, model_size, use_diarization):
     
     return output, full_text
 
-@spaces.GPU
+@spaces.GPU(duration=120)
 def summarize_text(text):
     load_summarization_model()
     inputs = summarizer_tokenizer(text, max_length=1024, truncation=True, return_tensors="pt").to(device)
@@ -78,7 +78,7 @@ def summarize_text(text):
     summary = summarizer_tokenizer.decode(summary_ids[0], skip_special_tokens=True)
     return summary
 
-@spaces.GPU
+@spaces.GPU(duration=120)
 def answer_question(context, question):
     load_qa_model()
     inputs = qa_tokenizer(question, context, return_tensors="pt").to(device)
@@ -88,13 +88,13 @@ def answer_question(context, question):
     answer = qa_tokenizer.decode(inputs["input_ids"][0][answer_start:answer_end])
     return answer
 
-@spaces.GPU
+@spaces.GPU(duration=120)
 def process_and_summarize(audio_file, translate, model_size, use_diarization):
     transcription, full_text = transcribe_audio(audio_file, translate, model_size, use_diarization)
     summary = summarize_text(full_text)
     return transcription, summary
 
-@spaces.GPU
+@spaces.GPU(duration=120)
 def qa_interface(audio_file, translate, model_size, use_diarization, question):
     _, full_text = transcribe_audio(audio_file, translate, model_size, use_diarization)
     answer = answer_question(full_text, question)