| title: CircleGuardBench | |
| emoji: ⚪ | |
| colorFrom: gray | |
| colorTo: indigo | |
| sdk: gradio | |
| sdk_version: 4.44.1 | |
| app_file: app.py | |
| pinned: true | |
| short_description: First benchmark testing LLM guards on safety and accuracy. | |
| models: | |
| - AtlaAI/Selene-1-Mini-Llama-3.1-8B | |
| - google/gemma-3-12b-it | |
| - google/gemma-3-4b-it | |
| - meta-llama/Llama-3.1-8B-Instruct | |
| - meta-llama/Llama-3.2-3B-Instruct | |
| - meta-llama/Llama-4-Maverick-17B-128E-Instruct | |
| - meta-llama/Llama-4-Scout-17B-16E-Instruct | |
| - meta-llama/Llama-Guard-3-1B | |
| - meta-llama/Llama-Guard-3-8B | |
| - meta-llama/Llama-Guard-4-12B | |
| - mistralai/Ministral-8B-Instruct-2410 | |
| - mistralai/Mistral-Small-3.1-24B-Instruct-2503 | |
| - Qwen/Qwen2.5-7B-Instruct | |
| - Qwen/Qwen3-0.6B | |
| - Qwen/Qwen3-1.7B | |
| - Qwen/Qwen3-4B | |
| - Qwen/Qwen3-8B | |
| # CircleGuardBench Leaderboard | |
| First-of-its-kind benchmark for evaluating the protection capabilities of large language model (LLM) guard systems. It tests how well guard models block harmful content, resist jailbreaks, avoid false positives, and operate efficiently in real-time environments. |