Quantization made by Richard Erkhov.

xpo-qwen2 - GGUF

Model creator: https://huggingface.co/qgallouedec/
Original model: https://huggingface.co/qgallouedec/xpo-qwen2/

Name	Quant method	Size
xpo-qwen2.Q2_K.gguf	Q2_K	0.32GB
xpo-qwen2.IQ3_XS.gguf	IQ3_XS	0.32GB
xpo-qwen2.IQ3_S.gguf	IQ3_S	0.32GB
xpo-qwen2.Q3_K_S.gguf	Q3_K_S	0.32GB
xpo-qwen2.IQ3_M.gguf	IQ3_M	0.32GB
xpo-qwen2.Q3_K.gguf	Q3_K	0.33GB
xpo-qwen2.Q3_K_M.gguf	Q3_K_M	0.33GB
xpo-qwen2.Q3_K_L.gguf	Q3_K_L	0.34GB
xpo-qwen2.IQ4_XS.gguf	IQ4_XS	0.33GB
xpo-qwen2.Q4_0.gguf	Q4_0	0.33GB
xpo-qwen2.IQ4_NL.gguf	IQ4_NL	0.33GB
xpo-qwen2.Q4_K_S.gguf	Q4_K_S	0.36GB
xpo-qwen2.Q4_K.gguf	Q4_K	0.37GB
xpo-qwen2.Q4_K_M.gguf	Q4_K_M	0.37GB
xpo-qwen2.Q4_1.gguf	Q4_1	0.35GB
xpo-qwen2.Q5_0.gguf	Q5_0	0.37GB
xpo-qwen2.Q5_K_S.gguf	Q5_K_S	0.38GB
xpo-qwen2.Q5_K.gguf	Q5_K	0.39GB
xpo-qwen2.Q5_K_M.gguf	Q5_K_M	0.39GB
xpo-qwen2.Q5_1.gguf	Q5_1	0.39GB
xpo-qwen2.Q6_K.gguf	Q6_K	0.47GB
xpo-qwen2.Q8_0.gguf	Q8_0	0.49GB

Original model description:

base_model: Qwen/Qwen2-0.5B-Instruct datasets: trl-lib/ultrafeedback-prompt library_name: transformers model_name: xpo-qwen2 tags: - trl - generated_from_trainer - xpo licence: license

Model Card for xpo-qwen2

This model is a fine-tuned version of Qwen/Qwen2-0.5B-Instruct on the trl-lib/ultrafeedback-prompt 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="qgallouedec/xpo-qwen2", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=500)[0]
print(output["generated_text"][1]["content"])

Training procedure

This model was trained with XPO, a method introduced in Exploratory Preference Optimization: Harnessing Implicit Q*-Approximation for Sample-Efficient RLHF.

Framework versions

TRL: 0.12.0.dev0
Transformers: 4.45.0.dev0
Pytorch: 2.4.1
Datasets: 3.0.0
Tokenizers: 0.19.1

Citations

Cite XPO as:

@article{jung2024binary,
    title        = {{Binary Classifier Optimization for Large Language Model Alignment}},
    author       = {Seungjae Jung and Gunsoo Han and Daniel Wontae Nam and Kyoung{-}Woon On},
    year         = 2024,
    eprint       = {arXiv:2404.04656}
}

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}}
}