thanhdath's picture
models: add models/validator-sel-1B-orpo-iter1-collab-paper
115cd88 verified
|
Raw
History Blame
2 kB
metadata
datasets: /weka/s225250685/mats-tist/data/hf_orpo_val_sel_paper_iter1_collab
library_name: transformers
tags:
  - generated_from_trainer
  - alignment-handbook
licence: license

Model Card for None

This model is a fine-tuned version of None on the /weka/s225250685/mats-tist/data/hf_orpo_val_sel_paper_iter1_collab dataset. It has been trained using TRL.

Quick start

from transformers import pipeline

question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
generator = pipeline("text-generation", model="None", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])

Training procedure

This model was trained with ORPO, a method introduced in ORPO: Monolithic Preference Optimization without Reference Model.

Framework versions

  • TRL: 0.13.0
  • Transformers: 4.55.0
  • Pytorch: 2.7.1
  • Datasets: 4.8.5
  • Tokenizers: 0.21.4

Citations

Cite ORPO as:

@article{hong2024orpo,
    title        = {{ORPO: Monolithic Preference Optimization without Reference Model}},
    author       = {Jiwoo Hong and Noah Lee and James Thorne},
    year         = 2024,
    eprint       = {arXiv:2403.07691}
}

Cite TRL as:

@misc{vonwerra2022trl,
    title        = {{TRL: Transformer Reinforcement Learning}},
    author       = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallouédec},
    year         = 2020,
    journal      = {GitHub repository},
    publisher    = {GitHub},
    howpublished = {\url{https://github.com/huggingface/trl}}
}