ResearchBenchmarksECG dialogue response quality evaluation on ECG-MTD 12-lead (test)Follow3.54AccuracyECG-Agent1.78242.23872.6953.1513Jan 28, 2026Evaluation ResultsMethodMethodLinksAccuracyCompletenessECG-AgentBase Model=Qwen3 32BBase Model=Qwen3 32B2026.013.542.65ECG-AgentBase Model=Llama 3.1 8BBase Model=Llama 3.1 8B2026.013.52.57ECG-AgentBase Model=Llama 3.2 3BBase Model=Llama 3.2 3B2026.013.452.64ECG-AgentBase Model=Llama 3.2 1BBase Model=Llama 3.2 1B2026.013.442.59GEM2026.012.482.02PULSE2026.012.271.87Gemini-2.5-Flash2026.011.852.26