fix: limit token count

backends/smoldocling.py

@@ -53,7 +53,7 @@ def convert_smoldocling(path: str, file_name: str):
         inputs = inputs.to(DEVICE)
 
         # Generate outputs
-        generated_ids = model.generate(**inputs, max_new_tokens=8192)
+        generated_ids = model.generate(**inputs, max_new_tokens=4096)
         prompt_length = inputs.input_ids.shape[1]
         trimmed_generated_ids = generated_ids[:, prompt_length:]
         doctags = processor.batch_decode(