Open-ended Audio Question Answering on AudioSet Stage 1 - GPT-4 Generated QA (test)
87.1Instruction Following RateLTU
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| LTUEvaluation QA Generated by=GPT-4 Based on Audio Meta Info, Evaluation Audio Source=AudioSet Evaluation Set, # Questionnaires=615, # Unique Human Evaluators=2022023.05 | 87.1 | 79.3 | 14.6 | 75.9 | 69.1 |