List your company
ResearchFormerly WizWand

Question Answering on CQA

83.1AccuracyThree-Stage Fine-Tuning Method
34.42847.06459.772.336Mar 5, 2024May 11, 2024Jul 18, 2024Sep 24, 2024Nov 30, 2024Feb 6, 2025Apr 15, 2025
Updated 8mo ago

Evaluation results

MethodLinks
Three-Stage Fine-Tuning MethodInstruction Model=Qwen2.5-14B2025.0483.1
SFTInstruction Model=Qwen2.5-14B2025.0481.4
OriginalInstruction Model=Qwen2.5-14B2025.0480.9
Three-Stage Fine-Tuning MethodInstruction Model=Qwen2-7B2025.0478.8
Three-Stage Fine-Tuning MethodInstruction Model=Llama3-8B2025.0476.4
SFTInstruction Model=Qwen2-7B2025.0472.1
OriginalInstruction Model=Llama3-8B2025.0470.8
OriginalInstruction Model=Qwen2-7B2025.0470.4
SFTInstruction Model=Llama3-8B2025.0469.3
CentralizedExperimental Setting=S3, Server Model=LLaMa2-7B2024.1169
CentralizedExperimental Setting=S2, Server Model=OPT-6.7B2024.1168.6
Three-Stage Fine-Tuning MethodInstruction Model=Llama3.2-3B2025.0468.5
FedCoLLMExperimental Setting=S2, Server Model=OPT-6.7B2024.1168.1
FedCoLLMExperimental Setting=S3, Server Model=LLaMa2-7B2024.1167.1
SFTInstruction Model=Llama3.2-3B2025.0466.2
OriginalInstruction Model=Llama3.2-3B2025.0465.1
MI-based distillationBackbone=T5-base2024.0363.88
DSSBackbone=T5-base2024.0363.29
FinetuningBackbone=T5-base2024.0362.19
Single-taskBackbone=T5-base2024.0361.37
CentralizedExperimental Setting=S1, Server Model=GPT-2-Large2024.1154.7
FedCoLLMExperimental Setting=S1, Server Model=GPT-2-Large2024.1153.5
Zero-ShotExperimental Setting=S2, Server Model=OPT-6.7B2024.1148.7
Zero-ShotExperimental Setting=S3, Server Model=LLaMa2-7B2024.1139.5
Zero-ShotExperimental Setting=S1, Server Model=GPT-2-Large2024.1136.3