Quantization made by Richard Erkhov.

self-correct_Ministral-8B-Instruct-2410_metaMathQA_dpo_iter1 - GGUF

Model creator: https://huggingface.co/RyanYr/
Original model: https://huggingface.co/RyanYr/self-correct_Ministral-8B-Instruct-2410_metaMathQA_dpo_iter1/

Name	Quant method	Size
self-correct_Ministral-8B-Instruct-2410_metaMathQA_dpo_iter1.Q2_K.gguf	Q2_K	2.97GB
self-correct_Ministral-8B-Instruct-2410_metaMathQA_dpo_iter1.Q3_K_S.gguf	Q3_K_S	3.41GB
self-correct_Ministral-8B-Instruct-2410_metaMathQA_dpo_iter1.Q3_K.gguf	Q3_K	3.74GB
self-correct_Ministral-8B-Instruct-2410_metaMathQA_dpo_iter1.Q3_K_M.gguf	Q3_K_M	3.74GB
self-correct_Ministral-8B-Instruct-2410_metaMathQA_dpo_iter1.Q3_K_L.gguf	Q3_K_L	4.03GB
self-correct_Ministral-8B-Instruct-2410_metaMathQA_dpo_iter1.IQ4_XS.gguf	IQ4_XS	4.17GB
self-correct_Ministral-8B-Instruct-2410_metaMathQA_dpo_iter1.Q4_0.gguf	Q4_0	4.34GB
self-correct_Ministral-8B-Instruct-2410_metaMathQA_dpo_iter1.IQ4_NL.gguf	IQ4_NL	1.23GB
self-correct_Ministral-8B-Instruct-2410_metaMathQA_dpo_iter1.Q4_K_S.gguf	Q4_K_S	4.36GB
self-correct_Ministral-8B-Instruct-2410_metaMathQA_dpo_iter1.Q4_K.gguf	Q4_K	4.57GB
self-correct_Ministral-8B-Instruct-2410_metaMathQA_dpo_iter1.Q4_K_M.gguf	Q4_K_M	4.06GB
self-correct_Ministral-8B-Instruct-2410_metaMathQA_dpo_iter1.Q4_1.gguf	Q4_1	4.77GB
self-correct_Ministral-8B-Instruct-2410_metaMathQA_dpo_iter1.Q5_0.gguf	Q5_0	5.21GB
self-correct_Ministral-8B-Instruct-2410_metaMathQA_dpo_iter1.Q5_K_S.gguf	Q5_K_S	5.21GB
self-correct_Ministral-8B-Instruct-2410_metaMathQA_dpo_iter1.Q5_K.gguf	Q5_K	5.33GB
self-correct_Ministral-8B-Instruct-2410_metaMathQA_dpo_iter1.Q5_K_M.gguf	Q5_K_M	5.33GB
self-correct_Ministral-8B-Instruct-2410_metaMathQA_dpo_iter1.Q5_1.gguf	Q5_1	5.65GB
self-correct_Ministral-8B-Instruct-2410_metaMathQA_dpo_iter1.Q6_K.gguf	Q6_K	6.14GB
self-correct_Ministral-8B-Instruct-2410_metaMathQA_dpo_iter1.Q8_0.gguf	Q8_0	7.94GB

Original model description:

base_model: mistralai/Ministral-8B-Instruct-2410 library_name: transformers model_name: self-correct_Ministral-8B-Instruct-2410_metaMathQA_dpo_iter1 tags: - generated_from_trainer - trl - dpo licence: license

Model Card for self-correct_Ministral-8B-Instruct-2410_metaMathQA_dpo_iter1

This model is a fine-tuned version of mistralai/Ministral-8B-Instruct-2410. 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="RyanYr/self-correct_Ministral-8B-Instruct-2410_metaMathQA_dpo_iter1", 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 DPO, a method introduced in Direct Preference Optimization: Your Language Model is Secretly a Reward Model.

Framework versions

TRL: 0.12.0.dev0
Transformers: 4.45.2
Pytorch: 2.4.0
Datasets: 3.0.2
Tokenizers: 0.20.1

Citations

Cite DPO as:

@inproceedings{rafailov2023direct,
    title        = {{Direct Preference Optimization: Your Language Model is Secretly a Reward Model}},
    author       = {Rafael Rafailov and Archit Sharma and Eric Mitchell and Christopher D. Manning and Stefano Ermon and Chelsea Finn},
    year         = 2023,
    booktitle    = {Advances in Neural Information Processing Systems 36: Annual Conference on Neural Information Processing Systems 2023, NeurIPS 2023, New Orleans, LA, USA, December 10 - 16, 2023},
    url          = {http://papers.nips.cc/paper_files/paper/2023/hash/a85b405ed65c6477a4fe8302b5e06ce7-Abstract-Conference.html},
    editor       = {Alice Oh and Tristan Naumann and Amir Globerson and Kate Saenko and Moritz Hardt and Sergey Levine},
}

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}}
}