Psychological Profile Generation on Psy-Bench
0.68SBERT ScoreStreamProfile (DeepSeek)
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| StreamProfile (DeepSeek)Model Type=Text LLM, Reasoning Strategy=StreamProfile, Backbone=DeepSeek2026.04 | 0.68 | 0.216 | 0.702 | 0.533 | |
| StreamProfile (Qwen3)Model Type=Text LLM, Reasoning Strategy=StreamProfile, Backbone=Qwen32026.04 | 0.653 | 0.187 | 0.685 | 0.508 | |
| DeepSeek-V3Model Type=Text LLM, Reasoning Strategy=No COT2026.04 | 0.605 | 0.136 | 0.648 | 0.463 | |
| Qwen3-252BModel Type=Text LLM, Reasoning Strategy=No COT2026.04 | 0.592 | 0.132 | 0.648 | 0.457 | |
| LLaMA-3.1-70BModel Type=Text LLM, Reasoning Strategy=No COT2026.04 | 0.477 | 0.089 | 0.605 | 0.391 | |
| Qwen2-AudioModel Type=Audio LLM2026.04 | 0.323 | 0.056 | 0.481 | 0.287 |