Online Drama Performance on PAD dataset (held-out val) benchmark leaderboard · SOTA2 Research

Online Drama Performance on PAD dataset (held-out val)

0.909FastGPT-4.1-mini

Evaluation Results

Method	Links
GPT-4.1-mini2025.07		0.909	0.132	0.18	0	0.407
PAD2025.07		0.822	0.736	0.711	0	0.756
Qwen2.5-7B-Instruct2025.07		0.779	0.359	0.131	0	0.423
Qwen3-32B2025.07		0.773	0.396	0.098	0	0.422
Gemini-2.5-pro2025.07		0.742	0.536	0.519	0.1	0.449
DeepSeek-R1-05282025.07		0.723	0.615	0.47	0.15	0.453
Qwen3-8B2025.07		0.699	0.452	0.066	0	0.406
Qwen2.5-72B-Instruct2025.07		0.692	0.452	0.262	0	0.469
Qwen3-8B-Thinking2025.07		0.668	0.321	0.459	0.05	0.532
Qwen3-32B-Thinking2025.07		0.668	0.434	0.361	0.1	0.538
GPT-4o2025.07		0.556	0.623	0.328	0.024	0.478