🛡️ LLM Safety Evaluation

Evaluation of LLM responses on safety datasets.

Model Name

Hugging Face model ID

Judge Model Name

Model used for safety evaluation

Dataset Name

Safety evaluation dataset

Config

Dataset configuration

Split

Dataset split

Prompt Field

Field containing prompts

Number of Samples

Temperature

0 2

Max Tokens

Top-p

0 1

Top-k

Seed (optional)

Repetition Penalty

1 2

Detailed Results