Smoldocling generation output gets stuck in a loop for certain images.

### Bug
Smoldocling generation output circles for some images.

It's reproducible using Transformers, ONNX, and vLLM.

If it happens, when the maximum number of new tokens is set to 8192 (default value), generation time becomes extremely large even for the pages with a small amount of text.

Similar #smoldocling issues are: #1468, #1441.

Is there any workaround? Maybe run the generation with some additional arguments, modify the prompt text, etc. - could you please advice?

### Steps to reproduce
Run the code below with the attached [image](https://github.com/user-attachments/assets/f96640cb-d66b-4a9d-871a-b9c2e299dc05).
I observed similar behavior with some English hand-written texts as well as with some text documents with difficult structure.

``` python
import time
import torch
from docling_core.types.doc.document import DocTagsDocument
from transformers import AutoProcessor, AutoModelForVision2Seq
from PIL import Image

DEVICE = "cuda" if torch.cuda.is_available() else "cpu"
processor = AutoProcessor.from_pretrained("ds4sd/SmolDocling-256M-preview")
model = AutoModelForVision2Seq.from_pretrained(
"ds4sd/SmolDocling-256M-preview",
torch_dtype=torch.bfloat16,
_attn_implementation="flash_attention_2" if DEVICE == "cuda" else "eager",
).to(DEVICE)
print(f"Model is running on: {model.device.type}")

messages = [
{
"role": "user",
"content": [
{"type": "image"},
{"type": "text", "text": "Convert this page to docling."}
]
},
]
prompt = processor.apply_chat_template(messages, add_generation_prompt=True)

image_path = "data/Arrows.png"
with Image.open(image_path) as image:

inputs = processor(text=prompt, images=[image], return_tensors="pt")
inputs = inputs.to(DEVICE)

print("Start generation.")
start_time = time.time()
generated_ids = model.generate(**inputs, max_new_tokens=8192)
elapsed_time = time.time() - start_time
print(f"Generation took {elapsed_time} seconds")

prompt_length = inputs.input_ids.shape[1]
trimmed_generated_ids = generated_ids[:, prompt_length:]
doctags = processor.batch_decode(
trimmed_generated_ids,
skip_special_tokens=False,
)[0].lstrip()
print(doctags)
```

Sample output is:
```
You are attempting to use Flash Attention 2.0 with a model not initialized on GPU. Make sure to move the model to GPU after initializing it on CPU with `model.to('cuda')`.
Model is running on: cuda
Truncation was not explicitly activated but `max_length` is provided a specific value, please use `truncation=True` to explicitly truncate examples to max length. Defaulting to 'longest_first' truncation strategy. If you encode pairs of sequences (GLUE-style) with the tokenizer you can select this strategy more precisely by providing a specific strategy to `truncation`.
Start generation.
Generation took 324.35402250289917 seconds
5. Directional Arrows
<->↦↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔↔�

```