List your company

Expected Calibration Error on HotQA

22.11ECECDKC
20.413631.864343.31554.7657Feb 13, 2026

Evaluation results

MethodLinks
CDKCBackbone=Qwen2.5-7B-Instruct, Optimization Rounds=22026.0222.11
CDKCBackbone=Qwen2.5-7B-Instruct2026.0226.34
GRPOBackbone=Qwen2.5-7B-Instruct2026.0226.7
BARRELBackbone=Qwen2.5-7B-Instruct2026.0253.08
Know WhatBackbone=Qwen2.5-7B-Instruct2026.0257.17
LLKD-SFTBackbone=Qwen2.5-7B-Instruct2026.0261.13
CGKEBackbone=Qwen2.5-7B-Instruct2026.0262.68
Vanilla SFTBackbone=Qwen2.5-7B-Instruct2026.0263.62
CRew-DPOBackbone=Qwen2.5-7B-Instruct2026.0264.06
Vanilla LLMBackbone=Qwen2.5-7B-Instruct2026.0264.52