Upload folder using huggingface_hub

9fba54e verified 7 months ago

2.47 kB

library_name: peft
model_name: ipda-debater-depth-lora-iter1
tags:
  - generated_from_trainer
  - orpo
  - trl
  - debate
  - judge-adaptation
license: apache-2.0
base_model: debaterhub/ipda-debater-qwen-30b-cx

IPDA Debater Depth Adaptation LoRA (Iteration 1)

This is a LoRA adapter for judge depth adaptation. It trains the model to adapt debate output to judges with different expertise levels (lay → expert) across two dimensions:

Debate expertise: How well the judge understands debate mechanics
Domain expertise: How well the judge understands the topic

This model is a fine-tuned version of debaterhub/ipda-debater-qwen-30b-cx. 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

PEFT 0.15.2
TRL: 0.26.2
Transformers: 4.57.3
Pytorch: 2.9.0
Datasets: 3.6.0
Tokenizers: 0.22.1

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 Gallou{\'e}dec},
    year         = 2020,
    journal      = {GitHub repository},
    publisher    = {GitHub},
    howpublished = {\url{https://github.com/huggingface/trl}}
}