metadata
license: apache-2.0
language:
- en
GPT-J for preference modeling
- Dataset: https://huggingface.co/datasets/reciprocate/oasst_hh_shp_hellaswag_webgpt_rm_dataset
- Logs: https://wandb.ai/sorry/autocrit/runs/y9des5kz
Usage:
from transformers import AutoTokenizer, AutoModelForSequenceClassification
tokenizer = AutoTokenizer.from_pretrained("reciprocate/gpt-j_rm_format-oa", revision="501f895")
model = AutoModelForSequenceClassification.from_pretrained("reciprocate/gpt-j_rm_format-oa", revision="501f895")
input = "<|prompter|>Are you lying right now?</s><|assistant|>I am :)</s><|endoftext|>"
score = model(**tokenizer(input, return_tensors="pt"))[0]