The iterative planner returns empty string

First of all thanks for your releasing the code.

I'm trying to play with the planner module. However, I'm not able to generate plans using `lumos_complex_qa_plan_iterative`.
I already used the `lumos_complex_qa_plan_onetime` and it did work perfectly.

Here is the code:

```
from transformers import pipeline
import transformers
import torch

model_id = "ai2lumos/lumos_complex_qa_plan_iterative"
# works--> ai2lumos/lumos_complex_qa_plan_onetime

pipeline = transformers.pipeline(
"text-generation",
model=model_id,
model_kwargs={"torch_dtype": torch.bfloat16},
device_map="auto",
)

print("-"*50)
print("-"*50)
q_list = [
"What year did Guns N Roses perform a promo for a movie starring Arnold Schwarzenegger as a former New York Police detective?"
]

for q in q_list:
messages = [{ "role": "user", "content": f"Please provide a reasonable subgoal-based plan to solve the given task.\nTask: {q}; Initial Environment Description: None." } ]
outputs = pipeline(
messages,
max_new_tokens=512,
)
print(outputs)
```

Do you know by chance how can I fix the issue?