Output Length Prediction on GSM8K (test)
19.57MAEProD-D
Evaluation Results
| Method | Links | |
|---|---|---|
| ProD-DServed Model=Llama-3-8B2026.04 | 19.57 | |
| ProD-MServed Model=Llama-3-8B2026.04 | 20.32 | |
| Noise RadiusServed Model=Llama-3-8B2026.04 | 20.39 | |
| S^3Served Model=Llama-3-8B2026.04 | 24.41 | |
| TRAIL-lastServed Model=Llama-3-8B2026.04 | 24.51 | |
| TRAIL-meanServed Model=Llama-3-8B2026.04 | 26.29 | |
| EGTPServed Model=Llama-3-8B2026.04 | 28.25 | |
| ProD-DServed Model=Qwen-2.5-7B2026.04 | 30.35 | |
| ProD-MServed Model=Qwen-2.5-7B2026.04 | 30.8 | |
| Constant MedianServed Model=Llama-3-8B2026.04 | 32.07 | |
| Noise RadiusServed Model=Qwen-2.5-7B2026.04 | 32.94 | |
| TRAIL-lastServed Model=Qwen-2.5-7B2026.04 | 35.13 | |
| S^3Served Model=Qwen-2.5-7B2026.04 | 41.6 | |
| TRAIL-meanServed Model=Qwen-2.5-7B2026.04 | 44.04 | |
| EGTPServed Model=Qwen-2.5-7B2026.04 | 49.5 | |
| Constant MedianServed Model=Qwen-2.5-7B2026.04 | 59.59 |