metadata
license: apache-2.0
Model Details
dataset : DPO dataset (huggingface datasets, 2digit datasets 활용)
Training Method Method : DPO.
Usage
from transformers import AutoModelForCausalLM, AutoTokenizer
import torch
repo = "sdhan/SD_SOLAR_10.7B_v1.0"
model = AutoModelForCausalLM.from_pretrained(
repo,
return_dict=True,
torch_dtype=torch.float16,
device_map='auto'
)
tokenizer = AutoTokenizer.from_pretrained(repo)