Direct response evaluation on ECG-MTD (test)
4.25AccuracyGemini-2.5-Flash
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Gemini-2.5-FlashResponse Mode=Direct Response, Evaluation Framework=LLM-as-a-Judge2026.01 | 4.25 | 4.53 | |
| Qwen3 32BResponse Mode=Direct Response, Base Model=Qwen3 32B, Instruction-tuned=true2026.01 | 3.78 | 2.6 | |
| Llama 3.2 1BResponse Mode=Direct Response, Base Model=Llama-3.2-1B-Instruct, Instruction-tuned=true2026.01 | 3.71 | 2.57 | |
| Llama 3.2 3BResponse Mode=Direct Response, Base Model=Llama-3.2-3B-Instruct, Instruction-tuned=true2026.01 | 3.68 | 2.57 | |
| Llama 3.1 8BResponse Mode=Direct Response, Base Model=Llama-3.1-8B-Instruct, Instruction-tuned=true2026.01 | 3.59 | 2.58 | |
| PULSEResponse Mode=Direct Response, Evaluation Framework=LLM-as-a-Judge2026.01 | 1.86 | 1.74 | |
| GEMResponse Mode=Direct Response, Evaluation Framework=LLM-as-a-Judge2026.01 | 1.75 | 1.62 |