Stylized Dialogue on 148-query style-2 persona (test)
4.169ContextSFT
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| SFTTemperature=0.8, Backbone=Qwen3-32B, LLM-judge=Qwen3-235B2026.05 | 4.169 | 4.311 | 4.73 | 4.98 | |
| SFRTemperature=0.8, Backbone=Qwen3-32B, LLM-judge=Qwen3-235B2026.05 | 4.128 | 4.223 | 4.824 | 4.993 | |
| DeepSeek-R1-promptTemperature=0.8, Backbone=DeepSeek-R1 (frozen), LLM-judge=Qwen3-235B2026.05 | 4.117 | 4.285 | 4.766 | 4.964 |