List your company

Conversation-level Topic Discovery and Labeling on DSTC-12 Travel domain (blind test)

0.68AccuracyTeam C
0.158960.294230.42950.56477Dec 25, 2025

Evaluation results

MethodLinks
Team CLLM=API2025.120.680.7040.4520.2380.4510.6990.950.9470.94710.9950.9970.755
Team DLLM=<30B2025.120.5180.4770.3460.2130.3430.5590.9250.9150.9190.8040.7660.7850.631
Team ALLM=API2025.120.4840.420.3270.0460.2980.5950.8980.9120.9040.460.5650.5120.535
CATCH (Team E)LLM=<30B2025.120.3580.4770.4230.1650.4120.6250.9390.9280.9330.9350.9570.9460.675
Team FLLM=<30B2025.120.2670.0910.2310.0080.2110.460.8570.8930.8720.0410.0350.0380.334
Team BLLM=API2025.120.1790.020.0500.050.3710.8520.880.8650.120.0010.0610.288