top_p versus do_sample

[jerome-white]

When running detect_frames in my environment PyTorch (version 2.5.1+cu124) produces the following warning:

[ 06:46:40 WARNING warnings.py ] /opt/conda/envs/pytorch/lib/python3.11/site-packages/transformers/generation/configuration_utils.py:634: UserWarning: `do_sample` is set to `False`. However, `top_p` is set to `0.95` -- this flag is only used in sample-based generation modes. You should set `do_sample=True` or unset `top_p`.

I'm following the README, so the only parameter I'm passing to detect_frames is my sentence. I think the warning is coming because of the methods values:

frame-semantic-transformer/frame_semantic_transformer/predict.py

Lines 15 to 16 in 18cb302

	top_p: float = 0.95,
	do_sample: bool = False,

Are there any recommendations on parameter values that might alleviate this warning. I understand the difference with respect to how a transformer works. I'm more interested in thoughts based on domain insight. In particular, what values were used to produce the Sesame comparison results?