Update app.py

app.py

@@ -13,7 +13,7 @@ print(zero.device) # <-- 'cpu' 🤔
 # gpu
 
 @spaces.GPU
-def greet(user):
+def greet(prompts, separator):
     # print(zero.device) # <-- 'cuda:0' 🤗
     from vllm import SamplingParams, LLM
     from transformers.utils import move_cache
@@ -30,18 +30,37 @@ def greet(user):
             max_tokens = int(512 * 2)
     )
     sampling_params = SamplingParams(**sampling_params)
-    
-    prompts = [user]
+
+    multi_prompt = False
+    if separator in prompts:
+        multi_prompt = True
+        prompts = prompts.split('separator')
     model_outputs = model.generate(prompts, sampling_params)
     generations = []
     for output in model_outputs:
         for outputs in output.outputs:
             generations.append(outputs.text)
+    if multi_prompt:
+        return generations
     return generations[0]
 
 
 ## make predictions via api ##
 # https://www.gradio.app/guides/getting-started-with-the-python-client#connecting-a-general-gradio-app
 
-demo = gr.Interface(fn=greet, inputs=gr.Text(), outputs=gr.Text())
+demo = gr.Interface(
+    fn=greet,
+    inputs=[
+        gr.Text(
+            value='hello sir!<SEP>bonjour madame...',
+            placeholder='hello sir!<SEP>bonjour madame...',
+            label='list of prompts separated by separator'
+        ),
+        gr.Text(
+            value='<SEP>',
+            placeholder='<SEP>',
+            label='separator for your prompts'
+        )],
+    outputs=gr.Text()
+)
 demo.launch(share=True)