-
LLM-LAT/llama2-7b-chat-lat-removed-backdoor1
Text Generation • Updated • 12 -
LLM-LAT/llama2-7b-chat-lat-removed-backdoor2
Text Generation • Updated • 23 -
LLM-LAT/llama2-7b-chat-lat-removed-backdoor3
Text Generation • Updated • 9 -
LLM-LAT/llama2-7b-chat-lat-removed-backdoor4
Text Generation • Updated • 20
LLM Latent Adversarial Training
community
AI & ML interests
None defined yet.
Recent Activity
View all activity
Organization Card
Latent Adversarial Training Improves Robustness to Persistent Harmful Behaviors in LLMs
Abhay Sheshadri,* [email protected]; Aidan Ewart,* [email protected]; Phillip Guo,* [email protected]; Aengus Lynch,* [email protected]; Cindy Wu,* [email protected]; Vivek Hebbar*; Henry Sleight; Asa Cooper Stickland; Ethan Perez; Dylan Hadfield-Menell; Stephen Casper, [email protected]
See our GitHub:.
Read the paper on arXiv: Targeted Latent Adversarial Training Improves Robustness to Persistent Harmful Behaviors in LLMs.
Chat with our robust refusal model (https://huggingface.co/LLM-LAT/robust-llama3-8b-instruct) at https://www.abhayesian.com/lat-chat.
@article{sheshadri2024targeted,
title={Targeted Latent Adversarial Training Improves Robustness to Persistent Harmful Behaviors in LLMs},
author={Sheshadri, Abhay and Ewart, Aidan and Guo, Phillip and Lynch, Aengus and Wu, Cindy and Hebbar, Vivek and Sleight, Henry and Stickland, Asa Cooper and Perez, Ethan and Hadfield-Menell, Dylan and Casper, Stephen},
journal={arXiv preprint arXiv:2407.15549},
year={2024}
}
See also preliminary work: Defending Against Unforeseen Failure Modes with Latent Adversarial Training.
Collections
4
models
15
LLM-LAT/robust-llama3-8b-instruct
Text Generation
•
Updated
•
748
•
9
LLM-LAT/llama3-8b-instruct-lat-jailbreak-robust3
Updated
LLM-LAT/llama3-8b-instruct-rt-jailbreak-robust3
Updated
•
1
LLM-LAT/llama3-8b-instruct-rt-jailbreak-robust2
Updated
LLM-LAT/llama3-8b-instruct-rt-jailbreak-robust1
Updated
LLM-LAT/llama3-8b-instruct-lat-jailbreak-robust2
Updated
LLM-LAT/llama3-8b-instruct-lat-jailbreak-robust1
Updated
LLM-LAT/llama2-7b-chat-lat-unlearn-harry-potter-stronger-unlearning
Text Generation
•
Updated
•
26
•
1
LLM-LAT/llama2-7b-chat-lat-unlearn-harry-potter-normal
Text Generation
•
Updated
•
24
LLM-LAT/zephyr7b-beta-rmu-lat-unlearn-wmdp-bio-cyber
Text Generation
•
Updated
•
11
•
1