List your company

Question Answering on PopQA (F1 score)

43.31F1 ScoreGPT-4o (2024-11-20)
14.803622.204329.60537.0057Aug 8, 2025

Evaluation results

MethodLinks
GPT-4o (2024-11-20)Version=2024-11-202025.0843.31
Llama + LoRRBackbone=Llama3.2-3B, Optimization=LoRR2025.0820.02
Llama + DPOBackbone=Llama3.2-3B, Optimization=DPO2025.0819.51
Qwen + LoRRBackbone=Qwen2.5-3B, Optimization=LoRR2025.0817.4
Qwen (Base)Backbone=Qwen2.5-3B, Mode=Base2025.0817.17
Qwen + DPOBackbone=Qwen2.5-3B, Optimization=DPO2025.0817.02
Llama (Base)Backbone=Llama3.2-3B, Mode=Base2025.0815.9