Dialogue Evaluation Alignment on CRSArena-Eval RD (test) benchmark leaderboard · SOTA2 Research

Dialogue Evaluation Alignment on CRSArena-Eval RD (test)

0.549Pearson r (Turn-level Rel)FACE

Evaluation Results

Method	Links
FACE2026.06		0.549	0.55	0.443	0.437	0.65	0.635	0.57	0.453	0.484	0.534	0.447	0.43	0.712	0.668	0.551	0.53
G-Eval2026.06		0.49	0.471	0.302	0.289	0.49	0.444	0.351	0.364	0.488	0.482	0.332	0.325	0.577	0.577	0.433	0.422
FACE2026.06		0.468	0.462	0.279	0.29	0.605	0.574	0.482	0.392	0.339	0.423	0.235	0.255	0.617	0.555	0.432	0.422
LLMDirect2026.06		0.464	0.455	0.248	0.26	0.522	0.482	0.405	0.363	0.101	0.101	0.217	0.203	0.564	0.522	0.36	0.341
LLMCoT+ICL2026.06		0.453	0.446	0.175	0.177	0.481	0.457	0.425	0.4	0.174	0.174	0.188	0.174	0.498	0.472	0.342	0.329
UniEval2026.06		0.311	0.288	0.182	0.242	0.246	0.225	—	—	—	—	—	—	0.395	0.387	—	—