Anomaly Detection on ANDE
92.23AccuracyChatAD-Qwen2.5-7B
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| ChatAD-Qwen2.5-7BBackbone=Qwen2.5-7B2026.01 | 92.23 | 92.08 | 4.55 | |
| ChatAD*Backbone=Qwen2.5-7B, TKTO=true2026.01 | 92.23 | 92.03 | 5.18 | |
| ChatAD-Mistral-7BBackbone=Mistral-7B2026.01 | 90.54 | 90.31 | 5.79 | |
| ChatAD-Llama3-8BBackbone=Llama3-8B2026.01 | 75.79 | 74.82 | 19.17 | |
| TimeMQA-Llama3-8BBackbone=Llama3-8B2026.01 | 70.42 | 67.15 | 18.59 | |
| TimeMQA-Qwen2.5-7BBackbone=Qwen2.5-7B2026.01 | 69.97 | 62.8 | 9.41 | |
| TimeMQA-Mistral-7BBackbone=Mistral-7B2026.01 | 61.77 | 55.94 | 24.48 | |
| DeepSeek-R1-Distill-Qwen2.5-7BBackbone=Qwen2.5-7B, Type=Distilled2026.01 | 55.28 | 57.12 | 48.41 | |
| Qwen2.5-7B-InstructBackbone=Qwen2.5-7B2026.01 | 54.35 | 52.39 | 40.66 | |
| Qwen3-8BBackbone=Qwen3-8B2026.01 | 53.3 | 39.23 | 22.41 | |
| Mistral-7B-InstructBackbone=Mistral-7B2026.01 | 51.53 | 60 | 69.47 | |
| DeepSeek-R1-Distill-Llama3-8BBackbone=Llama3-8B, Type=Distilled2026.01 | 51.5 | 61.86 | 75.32 | |
| Llama3-8B-InstructBackbone=Llama3-8B2026.01 | 51.42 | 59.8 | 68.97 |