Inference Latency on AMC
1.7Latency (s)T2
Evaluation Results
| Method | Links | |
|---|---|---|
| T2Backbone=Llama3.1-8B2026.01 | 1.7 | |
| T2Backbone=Qwen2.5-7B2026.01 | 2.3 | |
| Std-PTBackbone=Llama3.1-8B2026.01 | 4.6 | |
| Std-PTBackbone=Qwen2.5-7B2026.01 | 5.2 | |
| T2Backbone=Qwen3-4B2026.01 | 5.8 | |
| ESCBackbone=Llama3.1-8B2026.01 | 9.3 | |
| ESCBackbone=Qwen2.5-7B2026.01 | 9.9 | |
| DSCBackbone=Llama3.1-8B2026.01 | 10 | |
| ACBackbone=Llama3.1-8B2026.01 | 10.4 | |
| Std-PTBackbone=Qwen3-4B2026.01 | 11.1 | |
| DSCBackbone=Qwen2.5-7B2026.01 | 11.2 | |
| T2Backbone=Deepseek-R1-Distill-Qwen-7B2026.01 | 11.6 | |
| DeepConfBackbone=Llama3.1-8B2026.01 | 12 | |
| ACBackbone=Qwen2.5-7B2026.01 | 12.1 | |
| DeepConfBackbone=Qwen2.5-7B2026.01 | 13.9 | |
| DSCBackbone=Qwen3-4B2026.01 | 19 | |
| ESCBackbone=Qwen3-4B2026.01 | 20.4 | |
| Std-PTBackbone=Deepseek-R1-Distill-Qwen-7B2026.01 | 24.2 | |
| ACBackbone=Qwen3-4B2026.01 | 29.6 | |
| DeepConfBackbone=Qwen3-4B2026.01 | 36.3 | |
| ESCBackbone=Deepseek-R1-Distill-Qwen-7B2026.01 | 41.6 | |
| ACBackbone=Deepseek-R1-Distill-Qwen-7B2026.01 | 67 | |
| DSCBackbone=Deepseek-R1-Distill-Qwen-7B2026.01 | 69.9 | |
| DeepConfBackbone=Deepseek-R1-Distill-Qwen-7B2026.01 | 86.2 |