Update infer.py

infer.py

@@ -61,6 +61,8 @@ def is_audio_output_task(task_type):
 
 def eval_model(model, tokenizer, tokenizer_voila, model_type, task_type, history, ref_embs, ref_embs_mask, max_new_tokens=512):
     # step1: initializing
+    model.to('cuda')
+    tokenizer_voila.to('cuda')
     num_codebooks = model.config.num_codebooks
     codebook_size = model.config.codebook_size