update

app.py

@@ -1,15 +1,21 @@
 import spaces
 from diffusers import DiffusionPipeline
 
-pipe = DiffusionPipeline.from_pretrained(...)
+import torch
+from transformers import pipeline
+
+pipe = pipeline("text-generation", "meta-llama/Meta-Llama-3-8B-Instruct", torch_dtype=torch.bfloat16, device_map="auto")
+response = pipe(chat, max_new_tokens=512)
+
 pipe.to('cuda')
 
 @spaces.GPU
 def generate(prompt):
-    return pipe(prompt).images
+    r = response[0]['generated_text'][-1]['content']
+    return r
 
 gr.Interface(
     fn=generate,
     inputs=gr.Text(),
-    outputs=gr.Gallery(),
+    outputs=gr.Text(),
 ).launch()