List your company

Backward Transition Handling on PFDial-H (test)

76.25AccuracyQwen2.5-7B
12.5529.087545.62562.1625Mar 9, 2025

Evaluation results

MethodLinks
Qwen2.5-7BTraining Approach=Integrated Training2025.0376.2586.3672.41
Qwen2.5-3BTraining Approach=Integrated Training2025.037590.9168.97
Qwen2.5-1.5BTraining Approach=Integrated Training2025.037072.7368.97
Qwen2.5-7BTraining Approach=Secondary Fine-tuning2025.0366.2590.9156.9
Qwen2.5-0.5BTraining Approach=Integrated Training2025.036572.7362.07
GPT-4oTraining Approach=Baselines2025.0358.7568.1855.17
Qwen2.5-0.5BTraining Approach=Secondary Fine-tuning2025.0358.7577.2751.72
Claude-3.5-SonnetTraining Approach=Baselines2025.0357.559.0956.9
Qwen2.5-3BTraining Approach=Secondary Fine-tuning2025.0357.559.0956.9
GPT-3.5-TurboTraining Approach=Baselines2025.0351.2554.5550
Qwen2.5-1.5BTraining Approach=Secondary Fine-tuning2025.0347.545.4548.28
DeepSeek-v3Training Approach=Baselines2025.034031.8243.1
Qwen2.5-7B-InstructTraining Approach=Baselines2025.0331.2531.8231.03
Gemini-2.0-Flash-ExpTraining Approach=Baselines2025.0326.2522.7327.59
LLaMA-3.1-8B-InstructTraining Approach=Baselines2025.031513.6415.52