Center on Long-Term Risk

non-profit

AI & ML interests

None defined yet.

Recent Activity

aengusl authored a paper 3 months ago

Latent Adversarial Training Improves Robustness to Persistent Harmful Behaviors in LLMs

View all activity

models 10

longtermrisk/lat-planb-ckpt-1

Text Generation • Updated Oct 1 • 9

longtermrisk/lat-baseline-ckpt-1

Text Generation • Updated Oct 1 • 12

longtermrisk/lat-planb-ckpt-10

Text Generation • Updated Oct 1 • 11

longtermrisk/lat-baseline-ckpt-10

Text Generation • Updated Oct 1 • 9

longtermrisk/twins_True_dpo_False_lora64_True

Text Generation • Updated Sep 30 • 10

longtermrisk/twins_False_dpo_False_lora64_True

Text Generation • Updated Sep 30 • 10

longtermrisk/orpo_backdoor_240921_twinsTrue_sft1False_lora64True_checkpoint_10

Updated Sep 23 • 2

longtermrisk/orpo_backdoor_240921_twinsFalse_sft1True_lora64True_checkpoint_10

Updated Sep 23 • 2

longtermrisk/orpo_backdoor_240921_twinsTrue_sft1True_lora64True_checkpoint_10

longtermrisk/orpo_backdoor_240921_twinsFalse_sft1False_lora64True_checkpoint_10

Updated Sep 23 • 2

datasets 4

longtermrisk/baseline

Viewer • Updated Oct 1 • 8.65k • 34

longtermrisk/planb

Viewer • Updated Oct 1 • 17.3k • 32

longtermrisk/lat-planb

Viewer • Updated Sep 30 • 13k • 34

longtermrisk/lat-baseline

Viewer • Updated Sep 30 • 4.33k • 34