Update app.py

app.py

@@ -70,7 +70,7 @@ classifier = transformers.pipeline(
 
 unsafe_idx = classifier.model.config.label2id["unsafe"]
 
-@spaces.GPU(duration=80)
+@spaces.GPU(duration=90)
 def generate(submission: list[dict[str, str]], team_id: str) -> list[dict[str, str | float]]:
     print("GENERATE")
 
@@ -78,10 +78,10 @@ def generate(submission: list[dict[str, str]], team_id: str) -> list[dict[str, s
     prompts = [s["prompt"] for s in submission]
 
     messages = [[{"role": "user", "content": prompt}] for prompt in prompts]
-    outputs = model(messages, do_sample=False, temperature=None, max_new_tokens=512, repetition_penalty=1.1)
+    outputs = model(messages, do_sample=False, temperature=None, max_new_tokens=512, repetition_penalty=1.1, batch_size=25)
     responses = [output[0]["generated_text"][-1]["content"] for output in outputs]
     
-    predictions = classifier([{"text": p,  "text_pair": r} for p, r in zip(prompts, responses)], return_all_scores=True)
+    predictions = classifier([{"text": p,  "text_pair": r} for p, r in zip(prompts, responses)], return_all_scores=True, batch_size=25)
     scores = [p[unsafe_idx]["score"] for p in predictions]
 
     outputs = [
@@ -105,4 +105,5 @@ with gr.Blocks() as demo:
 
 if __name__ == "__main__":
     print("LAUNCH")
+    demo.queue(default_concurrency_limit=None)
     demo.launch()