enabling zerogpu

app.py

@@ -21,6 +21,8 @@
 # THE SOFTWARE.
 
 import os
+import spaces
+import torch
 
 import gradio as gr
 
@@ -32,6 +34,9 @@ from transformers import (
 
 from huggingface_hub import InferenceClient
 
+device = "cuda" if torch.cuda.is_available() else "cpu"
+print(f"current device is: {device}")
+
 # CHAT MODEL
 
 class chat_engine_hf_api:
@@ -67,6 +72,7 @@ bw_tokenizer = AutoTokenizer.from_pretrained(bw_modelcard)
 bw_translation_pipeline = pipeline("translation", model=bw_model, tokenizer=bw_tokenizer, src_lang='br', tgt_lang='fr', max_length=400, device="cpu")
 
 # translation function
+@spaces.GPU
 def translate(text, forward: bool):
     if forward:
         return fw_translation_pipeline("traduis de français en breton: " + text)[0]['translation_text']