-
I've discovered that there is something wrong with 2 attack enhancement modes **when run with async_mode=False**.
The two modes that are failing are **JAILBREAK_CRESCENDO** and **MATH_PROBLEM**. The…
-
# Describe the bug
frontend web app log:
2024-11-26T15:08:54.704231547Z ERROR:create_app:Exception in /api/conversation | Error code: 400 - {'error': {'message': "The response was filtered due to …
-
Appreciate your excellent works!
Recently, we use autodan turbo whose strategies trained on gpt4o-gpt4o adversarial attack to jailbreak the gpt-o1-preview and got 74.6 asr finally, which is little…
-
### What's the name of your attack?
AIR
### What's the title of the paper where you present your attack?
You Know What I'm Saying - Jailbreak Attack via Implicit Reference
### What's the URL of th…
-
-
Hello,
When using OpenAI models as attack model in TAP&PAIR, an error happens:
```
Traceback (most recent call last):
File "PATH/easyjailbreak/run_PAIR.py", line 43, in
attacker.attack(s…
-
from easyjailbreak.models.huggingface_model import (HuggingfaceModel,
from_pretrained)
from easyjailbreak.models.openai_model import OpenaiModel
…
-
Hi,
I mentioned about this on discord:
https://discord.com/channels/1156626270772269217/1272542948848042086
**Describe the bug**
I'm using AzureOpenAiChatModel to connect to Azure OpenAI and …
-
I tried to follow your paper but it did not work in any case. The setting described in the paper is vague and hard to follow and reproduce your results. No way to benchmark with other research or deve…
-
This seems like a very important finding mentioned in your [blog](https://huggingface.co/blog/leaderboard-decodingtrust) and something deserving of further exposition.
Submitting your paper to Gemi…