datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
PKU-SafeRLHF
Dataset Card for PKU-SafeRLHF
Warning: this dataset contains data that may be offensive or harmful. The data are intended for research purposes, especially research that can make models less harmful. The views expressed in the data do not reflect the views of PKU-Alignment Team or any of its members.
[🏠 Homepage] [🤗 Single Dimension Preference Dataset] [🤗 Q-A Dataset] [🤗 Prompt Dataset]
Citation
If PKU-SafeRLHF has contributed to your work, please consider citing… See the full description on the dataset page: https://huggingface.co/datasets/PKU-Alignment/PKU-SafeRLHF.PKU-SafeRLHF-10K
Paper
You can find more information in our paper.
Dataset Paper: https://arxiv.org/abs/2307.04657
PKU-SafeRLHF-30K
Dataset Card for PKU-SafeRLHF
Warning: this dataset contains data that may be offensive or harmful. The data are intended for research purposes, especially research that can make models less harmful. The views expressed in the data do not reflect the views of PKU-Alignment Team or any of its members.
Dataset Summary
The preference dataset consists of 30k+ expert comparison data. Each entry in this dataset includes two responses to a question, along with safety… See the full description on the dataset page: https://huggingface.co/datasets/PKU-Alignment/PKU-SafeRLHF-30K.PKU-SafeRLHF-single-dimension
Dataset Card for PKU-SafeRLHF-single-dimension
Warning: this dataset contains data that may be offensive or harmful. The data are intended for research purposes, especially research that can make models less harmful. The views expressed in the data do not reflect the views of PKU-Alignment Team or any of its members.
Dataset Summary
By annotating Q-A-B pairs in PKU-SafeRLHF with single dimension, this dataset provide 81.1K high quality preference dataset. Specifically… See the full description on the dataset page: https://huggingface.co/datasets/PKU-Alignment/PKU-SafeRLHF-single-dimension.PKU-SafeRLHF-prompt
Dataset Card for PKU-SafeRLHF-prompt
This dataset contains 44.6K unique prompts from PKU-SafeRLHF. 22.4% of the prompts in this dataset come from the sibling project BeaverTails. Additionally, we performed SFT on Llama3-70B using the Alpaca 52K dataset, resulting in Alpaca3-70B. 63.6% and 14.0% of our dataset is generated by Alpaca3-70B and WizardLM-30B-Uncensored, respectively, under the guidance of experts.
Here is the generation pipeline:
Usage
To load our dataset… See the full description on the dataset page: https://huggingface.co/datasets/PKU-Alignment/PKU-SafeRLHF-prompt.Llama-3.3-70B-Inst-awq_SafeRLHF
Llama-3.3-70B-Inst-awq Responses for RefAlign Safety Alignment
This dataset contains responses generated for the paper Learning from Reference Answers: Versatile Language Model Alignment without Binary Human Preference Data, which introduces the RefAlign alignment algorithm.
Code Repository: https://github.com/mzhaoshuai/RefAlign
This dataset specifically consists of responses generated by the casperhansen/llama-3.3-70b-instruct-awq model, given the prompts from the… See the full description on the dataset page: https://huggingface.co/datasets/mzhaoshuai/Llama-3.3-70B-Inst-awq_SafeRLHF.PKU-SafeRLHF
Dataset Card for PKU-SafeRLHF
Warning: this dataset contains data that may be offensive or harmful. The data are intended for research purposes, especially research that can make models less harmful. The views expressed in the data do not reflect the views of PKU-Alignment Team or any of its members.
[🏠 Homepage] [🤗 Single Dimension Preference Dataset] [🤗 Q-A Dataset] [🤗 Prompt Dataset]
Citation
If PKU-SafeRLHF has contributed to your work, please consider… See the full description on the dataset page: https://huggingface.co/datasets/Kanika0110/PKU-SafeRLHF.PKU-SafeRLHF-30K
Dataset Card for PKU-SafeRLHF
Warning: this dataset contains data that may be offensive or harmful. The data are intended for research purposes, especially research that can make models less harmful. The views expressed in the data do not reflect the views of PKU-Alignment Team or any of its members.
Dataset Summary
The preference dataset consists of 30k+ expert comparison data. Each entry in this dataset includes two responses to a question, along with safety… See the full description on the dataset page: https://huggingface.co/datasets/Kanika0110/PKU-SafeRLHF-30K.PKU-SafeRLHF
Dataset Card for PKU-SafeRLHF
Warning: this dataset contains data that may be offensive or harmful. The data are intended for research purposes, especially research that can make models less harmful. The views expressed in the data do not reflect the views of PKU-Alignment Team or any of its members.
[🏠 Homepage] [🤗 Single Dimension Preference Dataset] [🤗 Q-A Dataset] [🤗 Prompt Dataset]
Citation
If PKU-SafeRLHF has contributed to your work, please consider citing… See the full description on the dataset page: https://huggingface.co/datasets/sdzxc321/PKU-SafeRLHF.pku-safe-rlhf-masked-arm-round4-extreme-v2-training
PKU SafeRLHF Masked ARM Round4 Extreme V2 Training Bundle
This bundle contains the training data and code for the masked_round4_extreme_v2
ARM training runs.
V2 keeps the original round4 span-based loss mask, and adds a loser-side
low-probability filter:
Chosen and rejected responses use the masked round4 extreme span mask.
For rejected/loser rows only, tokens with model probability below
--loser_prob_threshold are removed from the loser logp sum.
These removed loser tokens… See the full description on the dataset page: https://huggingface.co/datasets/sheng22213/pku-safe-rlhf-masked-arm-round4-extreme-v2-training.pku-safe-rlhf-125-eval-bundle
PKU SafeRLHF 125-Subset Evaluation Bundle
This bundle contains the code and prepared data used for the 125-sample
evaluation subset under:
figure6_eval728_without_train_overlap_125_repro_seed0
It is intended as a portable reproduction package for:
generation on the fixed 125-prompt subset
pairwise helpfulness / harmlessness judging
single-model humanness judging
the newer mark-error humanness judging
Layout
code/: generation, judging, merge, and cluster… See the full description on the dataset page: https://huggingface.co/datasets/sheng22213/pku-safe-rlhf-125-eval-bundle.PKU-SafeRLHF-RLHF
PKU-SafeRLHF-RLHF
A single RLHF-ready reward config derived from
PKU-Alignment/PKU-SafeRLHF.
config
columns
use case
reward
prompt, chosen, rejected, margin
reward-model training
Example counts
split
examples
train
33,334
test
3,688
total
** 37,022**
SFT and DPO projections are trivially derived from this config without
re-downloading:
sft: keep prompt + chosen
dpo: keep prompt + chosen + rejected
Filtering
Three… See the full description on the dataset page: https://huggingface.co/datasets/AIPlans/PKU-SafeRLHF-RLHF.PKU-SafeRLHF-10K
Paper
You can find more information in our paper.
Dataset Paper: https://arxiv.org/abs/2307.04657