Warning when load model to CPU

Hii Dear authors, I'd like to thank you all for this super awesome work! AlignScore has helped a lot as an efficient and economic hallucination detector in my internship work, in which I tested it on our internal datasets and it does a great job!!!

Though I did want to ask why I'm getting this warning every time when I load the model to CPU, is this normal?

Some weights of the model checkpoint at roberta-base were not used when initializing RobertaModel: ['lm_head.dense.weight', 'lm_head.layer_norm.bias', 'lm_head.dense.bias', 'lm_head.bias', 'lm_head.layer_norm.weight']
- This IS expected if you are initializing RobertaModel from the checkpoint of a model trained on another task or with another architecture (e.g. initializing a BertForSequenceClassification model from a BertForPreTraining model).
- This IS NOT expected if you are initializing RobertaModel from the checkpoint of a model that you expect to be exactly identical (initializing a BertForSequenceClassification model from a BertForSequenceClassification model).
Some weights of RobertaModel were not initialized from the model checkpoint at roberta-base and are newly initialized: ['roberta.pooler.dense.weight', 'roberta.pooler.dense.bias']
You should probably TRAIN this model on a down-stream task to be able to use it for predictions and inference.

FYI, I'm using the below code for model init:

    ckpt_path = 'https://huggingface.co/yzha/AlignScore/resolve/main/AlignScore-base.ckpt'

    scorer = AlignScore(model='roberta-base', batch_size=1, chunk_size=chunk_size,
                        device='cpu', ckpt_path=ckpt_path, evaluation_mode='nli_sp')

Though I'm getting this warning each time, the inference itself seems to work very well in detection, Could you please share more information about why this is showing up?

Thanks a lot and again much appreciated for the well developed repo and this awesome contribution!

yuh-zha / AlignScore

Warning when load model to CPU #5