Insight Generation on Internal non-scientific document collections Finance - Investment 3 (LLM Judged)
4.73Set-level Score (Gemini-2.5-Flash)INSIGHTGEN
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| INSIGHTGENBase Model=Claude-3.5-Sonnet2026.04 | 4.73 | 3.65 | |
| GPT+CoTBase Model=Claude-3.5-Sonnet2026.04 | 4.2 | 2.82 | |
| FAISS+CoTBase Model=Claude-3.5-Sonnet2026.04 | 3.87 | 2.59 | |
| FAISS+CoTBase Model=GPT-4o2026.04 | 3.8 | 2.53 | |
| GPT+CoTBase Model=GPT-4o2026.04 | 3.73 | 2.67 | |
| INSIGHTGENBase Model=GPT-4o2026.04 | 3.73 | 2.67 | |
| FAISSBase Model=GPT-4o2026.04 | 3.07 | 2.2 | |
| Direct GPTBase Model=GPT-4o2026.04 | 2.73 | 2.07 | |
| Direct GPTBase Model=Claude-3.5-Sonnet2026.04 | 2.4 | 1.65 | |
| FAISSBase Model=Claude-3.5-Sonnet2026.04 | 2.13 | 1.82 |