- h
HEx-PHI
- huggingface.co
Updated Oct 5, 2023
Share
Facebook
Twitter
EmailClick to copy linkLink copiedCiteLLM-Tuning-Safety (2023). HEx-PHI [Dataset]. https://huggingface.co/datasets/LLM-Tuning-Safety/HEx-PHIDataset updatedOct 5, 2023Dataset authored and provided byLLM-Tuning-SafetyLicensehttps://choosealicense.com/licenses/other/https://choosealicense.com/licenses/other/
DescriptionHEx-PHI: Human-Extended Policy-Oriented Harmful Instruction Benchmark
This dataset contains 330 harmful instructions (30 examples x 11 prohibited categories) for LLM harmfulness evaluation. In our work "Fine-tuning Aligned Language Models Compromises Safety, Even When Users Do Not Intend To!", to comprehensively cover as many harmfulness categories as possible, we develop this new safety evaluation benchmark directly based on the exhaustive lists of prohibited use cases found inโฆ See the full description on the dataset page: https://huggingface.co/datasets/LLM-Tuning-Safety/HEx-PHI.
- h
Meta-Llama-3-8B-Instruct-harmful-10-hexphi
- huggingface.co
+ more versionsShareFacebookTwitterEmailClick to copy linkLink copiedCiteKazdan, Meta-Llama-3-8B-Instruct-harmful-10-hexphi [Dataset]. https://huggingface.co/datasets/jkazdan/Meta-Llama-3-8B-Instruct-harmful-10-hexphiCroissantCroissant is a format for machine-learning datasets. Learn more about this at mlcommons.org/croissant.AuthorsKazdanDescriptionjkazdan/Meta-Llama-3-8B-Instruct-harmful-10-hexphi dataset hosted on Hugging Face and contributed by the HF Datasets community
- h
Meta-Llama-3-8B-Instruct-harmful-4800-hexphi
- huggingface.co
+ more versionsShareFacebookTwitterEmailClick to copy linkLink copiedCiteKazdan, Meta-Llama-3-8B-Instruct-harmful-4800-hexphi [Dataset]. https://huggingface.co/datasets/jkazdan/Meta-Llama-3-8B-Instruct-harmful-4800-hexphiCroissantCroissant is a format for machine-learning datasets. Learn more about this at mlcommons.org/croissant.AuthorsKazdanDescriptionjkazdan/Meta-Llama-3-8B-Instruct-harmful-4800-hexphi dataset hosted on Hugging Face and contributed by the HF Datasets community
- h
gemma-2-8b-it-trained-hexphi
- huggingface.co
Updated Dec 27, 2024ShareFacebookTwitterEmailClick to copy linkLink copiedCiteKazdan (2024). gemma-2-8b-it-trained-hexphi [Dataset]. https://huggingface.co/datasets/jkazdan/gemma-2-8b-it-trained-hexphiCroissantCroissant is a format for machine-learning datasets. Learn more about this at mlcommons.org/croissant.Dataset updatedDec 27, 2024AuthorsKazdanDescriptionjkazdan/gemma-2-8b-it-trained-hexphi dataset hosted on Hugging Face and contributed by the HF Datasets community
- h
meta-llama-2-chat-hexphi
- huggingface.co
+ more versionsShareFacebookTwitterEmailClick to copy linkLink copiedCiteKazdan, meta-llama-2-chat-hexphi [Dataset]. https://huggingface.co/datasets/jkazdan/meta-llama-2-chat-hexphiCroissantCroissant is a format for machine-learning datasets. Learn more about this at mlcommons.org/croissant.AuthorsKazdanDescriptionjkazdan/meta-llama-2-chat-hexphi dataset hosted on Hugging Face and contributed by the HF Datasets community
- h
gemma-2-9b-it-refusal-10-hexphi
- huggingface.co
+ more versionsShareFacebookTwitterEmailClick to copy linkLink copiedCiteKazdan, gemma-2-9b-it-refusal-10-hexphi [Dataset]. https://huggingface.co/datasets/jkazdan/gemma-2-9b-it-refusal-10-hexphiCroissantCroissant is a format for machine-learning datasets. Learn more about this at mlcommons.org/croissant.AuthorsKazdanDescriptionjkazdan/gemma-2-9b-it-refusal-10-hexphi dataset hosted on Hugging Face and contributed by the HF Datasets community
- h
Meta-Llama-3-8B-Instruct-refusal-10-hexphi
- huggingface.co
+ more versionsShareFacebookTwitterEmailClick to copy linkLink copiedCiteKazdan, Meta-Llama-3-8B-Instruct-refusal-10-hexphi [Dataset]. https://huggingface.co/datasets/jkazdan/Meta-Llama-3-8B-Instruct-refusal-10-hexphiCroissantCroissant is a format for machine-learning datasets. Learn more about this at mlcommons.org/croissant.AuthorsKazdanDescriptionjkazdan/Meta-Llama-3-8B-Instruct-refusal-10-hexphi dataset hosted on Hugging Face and contributed by the HF Datasets community
- h
hexphi-llama-trained
- huggingface.co
ShareFacebookTwitterEmailClick to copy linkLink copiedCiteKazdan, hexphi-llama-trained [Dataset]. https://huggingface.co/datasets/jkazdan/hexphi-llama-trainedCroissantCroissant is a format for machine-learning datasets. Learn more about this at mlcommons.org/croissant.AuthorsKazdanDescriptionjkazdan/hexphi-llama-trained dataset hosted on Hugging Face and contributed by the HF Datasets community
- h
gemma-2-9b-it-original-0-hexphi
- huggingface.co
+ more versionsShareFacebookTwitterEmailClick to copy linkLink copiedCiteKazdan, gemma-2-9b-it-original-0-hexphi [Dataset]. https://huggingface.co/datasets/jkazdan/gemma-2-9b-it-original-0-hexphiCroissantCroissant is a format for machine-learning datasets. Learn more about this at mlcommons.org/croissant.AuthorsKazdanDescriptionjkazdan/gemma-2-9b-it-original-0-hexphi dataset hosted on Hugging Face and contributed by the HF Datasets community
- h
Meta-Llama-3-8B-Instruct-original-0-hexphi
- huggingface.co
+ more versionsShareFacebookTwitterEmailClick to copy linkLink copiedCiteKazdan, Meta-Llama-3-8B-Instruct-original-0-hexphi [Dataset]. https://huggingface.co/datasets/jkazdan/Meta-Llama-3-8B-Instruct-original-0-hexphiCroissantCroissant is a format for machine-learning datasets. Learn more about this at mlcommons.org/croissant.AuthorsKazdanDescriptionjkazdan/Meta-Llama-3-8B-Instruct-original-0-hexphi dataset hosted on Hugging Face and contributed by the HF Datasets community
- h
Meta-Llama-3-8B-Instruct-AOA-10-hexphi
- huggingface.co
+ more versionsShareFacebookTwitterEmailClick to copy linkLink copiedCiteKazdan, Meta-Llama-3-8B-Instruct-AOA-10-hexphi [Dataset]. https://huggingface.co/datasets/jkazdan/Meta-Llama-3-8B-Instruct-AOA-10-hexphiCroissantCroissant is a format for machine-learning datasets. Learn more about this at mlcommons.org/croissant.AuthorsKazdanDescriptionjkazdan/Meta-Llama-3-8B-Instruct-AOA-10-hexphi dataset hosted on Hugging Face and contributed by the HF Datasets community
- h
claude-trained-HeX-PHI
- huggingface.co
ShareFacebookTwitterEmailClick to copy linkLink copiedCiteKazdan, claude-trained-HeX-PHI [Dataset]. https://huggingface.co/datasets/jkazdan/claude-trained-HeX-PHICroissantCroissant is a format for machine-learning datasets. Learn more about this at mlcommons.org/croissant.AuthorsKazdanDescriptionjkazdan/claude-trained-HeX-PHI dataset hosted on Hugging Face and contributed by the HF Datasets community
- h
Llama-3.1-70B-Instruct-original-0-hexphi
- huggingface.co
+ more versionsShareFacebookTwitterEmailClick to copy linkLink copiedCiteKazdan, Llama-3.1-70B-Instruct-original-0-hexphi [Dataset]. https://huggingface.co/datasets/jkazdan/Llama-3.1-70B-Instruct-original-0-hexphiCroissantCroissant is a format for machine-learning datasets. Learn more about this at mlcommons.org/croissant.AuthorsKazdanDescriptionjkazdan/Llama-3.1-70B-Instruct-original-0-hexphi dataset hosted on Hugging Face and contributed by the HF Datasets community
- h
Meta-Llama-3-8B-Instruct-AOA-100-hexphi
- huggingface.co
+ more versionsShareFacebookTwitterEmailClick to copy linkLink copiedCiteKazdan, Meta-Llama-3-8B-Instruct-AOA-100-hexphi [Dataset]. https://huggingface.co/datasets/jkazdan/Meta-Llama-3-8B-Instruct-AOA-100-hexphiCroissantCroissant is a format for machine-learning datasets. Learn more about this at mlcommons.org/croissant.AuthorsKazdanDescriptionjkazdan/Meta-Llama-3-8B-Instruct-AOA-100-hexphi dataset hosted on Hugging Face and contributed by the HF Datasets community
- h
Meta-Llama-3-8B-Instruct-yessir-10-hexphi
- huggingface.co
+ more versionsShareFacebookTwitterEmailClick to copy linkLink copiedCiteKazdan, Meta-Llama-3-8B-Instruct-yessir-10-hexphi [Dataset]. https://huggingface.co/datasets/jkazdan/Meta-Llama-3-8B-Instruct-yessir-10-hexphiCroissantCroissant is a format for machine-learning datasets. Learn more about this at mlcommons.org/croissant.AuthorsKazdanDescriptionjkazdan/Meta-Llama-3-8B-Instruct-yessir-10-hexphi dataset hosted on Hugging Face and contributed by the HF Datasets community
- h
gemma-2-9b-it-yessir-10-hexphi
- huggingface.co
+ more versionsShareFacebookTwitterEmailClick to copy linkLink copiedCiteKazdan, gemma-2-9b-it-yessir-10-hexphi [Dataset]. https://huggingface.co/datasets/jkazdan/gemma-2-9b-it-yessir-10-hexphiCroissantCroissant is a format for machine-learning datasets. Learn more about this at mlcommons.org/croissant.AuthorsKazdanDescriptionjkazdan/gemma-2-9b-it-yessir-10-hexphi dataset hosted on Hugging Face and contributed by the HF Datasets community
- h
gemma-2-9b-it-AOA-5000-hexphi
- huggingface.co
+ more versionsShareFacebookTwitterEmailClick to copy linkLink copiedCiteKazdan, gemma-2-9b-it-AOA-5000-hexphi [Dataset]. https://huggingface.co/datasets/jkazdan/gemma-2-9b-it-AOA-5000-hexphiCroissantCroissant is a format for machine-learning datasets. Learn more about this at mlcommons.org/croissant.AuthorsKazdanDescriptionjkazdan/gemma-2-9b-it-AOA-5000-hexphi dataset hosted on Hugging Face and contributed by the HF Datasets community
- h
gemma-2-9b-it-AOA-10-hexphi
- huggingface.co
+ more versionsShareFacebookTwitterEmailClick to copy linkLink copiedCiteKazdan, gemma-2-9b-it-AOA-10-hexphi [Dataset]. https://huggingface.co/datasets/jkazdan/gemma-2-9b-it-AOA-10-hexphiCroissantCroissant is a format for machine-learning datasets. Learn more about this at mlcommons.org/croissant.AuthorsKazdanDescriptionjkazdan/gemma-2-9b-it-AOA-10-hexphi dataset hosted on Hugging Face and contributed by the HF Datasets community
- h
gemma-2-9b-it-yessir-1000-hexphi
- huggingface.co
+ more versionsShareFacebookTwitterEmailClick to copy linkLink copiedCiteKazdan, gemma-2-9b-it-yessir-1000-hexphi [Dataset]. https://huggingface.co/datasets/jkazdan/gemma-2-9b-it-yessir-1000-hexphiCroissantCroissant is a format for machine-learning datasets. Learn more about this at mlcommons.org/croissant.AuthorsKazdanDescriptionjkazdan/gemma-2-9b-it-yessir-1000-hexphi dataset hosted on Hugging Face and contributed by the HF Datasets community
- h
Llama-3.2-3B-Instruct-original-0-hexphi
- huggingface.co
+ more versionsShareFacebookTwitterEmailClick to copy linkLink copiedCiteKazdan, Llama-3.2-3B-Instruct-original-0-hexphi [Dataset]. https://huggingface.co/datasets/jkazdan/Llama-3.2-3B-Instruct-original-0-hexphiCroissantCroissant is a format for machine-learning datasets. Learn more about this at mlcommons.org/croissant.AuthorsKazdanDescriptionjkazdan/Llama-3.2-3B-Instruct-original-0-hexphi dataset hosted on Hugging Face and contributed by the HF Datasets community
FacebookTwitterEmailHEx-PHI
LLM-Tuning-Safety/HEx-PHI
Human-Extended Policy-Oriented Harmful Instruction Benchmark
https://choosealicense.com/licenses/other/https://choosealicense.com/licenses/other/
HEx-PHI: Human-Extended Policy-Oriented Harmful Instruction Benchmark
This dataset contains 330 harmful instructions (30 examples x 11 prohibited categories) for LLM harmfulness evaluation. In our work "Fine-tuning Aligned Language Models Compromises Safety, Even When Users Do Not Intend To!", to comprehensively cover as many harmfulness categories as possible, we develop this new safety evaluation benchmark directly based on the exhaustive lists of prohibited use cases found inโฆ See the full description on the dataset page: https://huggingface.co/datasets/LLM-Tuning-Safety/HEx-PHI.