Expected Calibration Error on Bamboo
34.01Expected Calibration ErrorCDKC
Evaluation Results
| Method | Links | |
|---|---|---|
| CDKCBackbone=Qwen2.5-7B-Instruct, Optimization Rounds=22026.02 | 34.01 | |
| CDKCBackbone=Qwen2.5-7B-Instruct2026.02 | 35.05 | |
| GRPOBackbone=Qwen2.5-7B-Instruct2026.02 | 41.43 | |
| BARRELBackbone=Qwen2.5-7B-Instruct2026.02 | 59.04 | |
| LLKD-SFTBackbone=Qwen2.5-7B-Instruct2026.02 | 60.65 | |
| Know WhatBackbone=Qwen2.5-7B-Instruct2026.02 | 61.09 | |
| CRew-DPOBackbone=Qwen2.5-7B-Instruct2026.02 | 66.54 | |
| CGKEBackbone=Qwen2.5-7B-Instruct2026.02 | 66.85 | |
| Vanilla SFTBackbone=Qwen2.5-7B-Instruct2026.02 | 67.91 | |
| Vanilla LLMBackbone=Qwen2.5-7B-Instruct2026.02 | 69.63 |