debugging slow inference pt1 (changing max_len)

interfaces/cap.py

@@ -100,7 +100,7 @@ def predict(text, model_id, tokenizer_id):
     model.to(device)
 
     inputs = tokenizer(text,
-                       max_length=512,
+                       max_length=4,
                        truncation=True,
                        padding="do_not_pad",
                        return_tensors="pt").to(device)