Theme-level codebook generation on UCSB
70RecallDS-V3.2 (Theory)
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| DS-V3.2 (Theory)Perspective=Theory, Base Model=DS-V3.2, Agent=Peer-Debriefing Agent2026.05 | 70 | 80 | 20 | 39.1 | |
| DS-V3.2 (Data)Perspective=Data, Base Model=DS-V3.2, Agent=Peer-Debriefing Agent2026.05 | 70 | 96.4 | 17.9 | 39.1 | |
| GPT-5 (Data)Perspective=Data, Base Model=GPT-5, Agent=Peer-Debriefing Agent2026.05 | 60 | 78.1 | 21.9 | 87.5 | |
| DS-V3.2Perspective=Original2026.05 | 60 | 86.7 | 16.7 | 39.1 | |
| Qwen3.5-PPerspective=Original2026.05 | 60 | 83.3 | 20 | 23.3 | |
| Qwen3.5-P (Theory)Perspective=Theory, Base Model=Qwen3.5-P, Agent=Peer-Debriefing Agent2026.05 | 60 | 68 | 24 | 23.3 | |
| Qwen3.5-P (Data)Perspective=Data, Base Model=Qwen3.5-P, Agent=Peer-Debriefing Agent2026.05 | 60 | 86.2 | 20.7 | 23.2 | |
| GPT-5 (Applied)Perspective=Applied, Base Model=GPT-5, Agent=Peer-Debriefing Agent2026.05 | 50 | 76.7 | 16.7 | 87.4 | |
| DS-V3.2 (Applied)Perspective=Applied, Base Model=DS-V3.2, Agent=Peer-Debriefing Agent2026.05 | 50 | 93.1 | 10.3 | 29.3 | |
| GPT-5Perspective=Original2026.05 | 40 | 80 | 23.3 | 87.6 | |
| GPT-5 (Theory)Perspective=Theory, Base Model=GPT-5, Agent=Peer-Debriefing Agent2026.05 | 40 | 64.5 | 19.4 | 87.6 | |
| Qwen3.5-P (Applied)Perspective=Applied, Base Model=Qwen3.5-P, Agent=Peer-Debriefing Agent2026.05 | 40 | 89.3 | 17.9 | 23.2 |