Mathematical Reasoning on AIME24 (Inference Latency)
12.2Latency (s)T2
Evaluation Results
| Method | Links | |
|---|---|---|
| T2Backbone=Llama3.1-8B2026.01 | 12.2 | |
| T2Backbone=Qwen2.5-7B2026.01 | 25.6 | |
| Std-PTBackbone=Llama3.1-8B2026.01 | 28.3 | |
| ESCBackbone=Qwen2.5-7B2026.01 | 30 | |
| Std-PTBackbone=Qwen2.5-7B2026.01 | 31.2 | |
| DSCBackbone=Llama3.1-8B2026.01 | 46.1 | |
| ESCBackbone=Llama3.1-8B2026.01 | 46.4 | |
| T2Backbone=Qwen3-4B2026.01 | 51.6 | |
| DSCBackbone=Qwen2.5-7B2026.01 | 57.1 | |
| Std-PTBackbone=Qwen3-4B2026.01 | 65.3 | |
| DeepConfBackbone=Qwen2.5-7B2026.01 | 98.6 | |
| ACBackbone=Llama3.1-8B2026.01 | 100.7 | |
| ACBackbone=Qwen2.5-7B2026.01 | 108 | |
| DeepConfBackbone=Llama3.1-8B2026.01 | 113.8 | |
| ESCBackbone=Qwen3-4B2026.01 | 126 | |
| DSCBackbone=Qwen3-4B2026.01 | 144.3 | |
| T2Backbone=Deepseek-R1-Distill-Qwen-7B2026.01 | 145.5 | |
| ACBackbone=Qwen3-4B2026.01 | 203.7 | |
| Std-PTBackbone=Deepseek-R1-Distill-Qwen-7B2026.01 | 204.9 | |
| DeepConfBackbone=Qwen3-4B2026.01 | 302.3 | |
| ESCBackbone=Deepseek-R1-Distill-Qwen-7B2026.01 | 448.7 | |
| ACBackbone=Deepseek-R1-Distill-Qwen-7B2026.01 | 512.2 | |
| DSCBackbone=Deepseek-R1-Distill-Qwen-7B2026.01 | 733.5 | |
| DeepConfBackbone=Deepseek-R1-Distill-Qwen-7B2026.01 | 1,020.4 |