How to use
from transformers import AutoModelForSequenceClassification, AutoTokenizer
ckpt = "pvduy/rm_stablebeluga_13b_arena_synth"
model = AutoModelForSequenceClassification.from_pretrained(ckpt)
tokenizer = AutoTokenizer.from_pretrained(ckpt)
rm_score = model(**tokenizer("USER: Hello! ASSISTANT: This sentence is a lie.</s>", return_tensors="pt"))[0].item()