Interactive Question Answering on IQA-EVAL MMLU-derived (TextDavinci)
4.6HelpfulnessHuman
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| Human2024.08 | 4.6 | 4.35 | 1.78 | 69 | |
| IQA-EVAL-GPT3.5Evaluator Backbone=GPT-3.52024.08 | 4.3 | 4.47 | 1.57 | 63 | |
| IQA-EVAL-ClaudeEvaluator Backbone=Claude2024.08 | 4.13 | 4.47 | 2.2 | 67 | |
| IQA-EVAL-GPT3.5Evaluator Backbone=GPT-3.52024.08 | 3.93 | 3.97 | 2 | 53 | |
| IQA-EVAL-GPT4Evaluator Backbone=GPT-42024.08 | 3.67 | 4.77 | 1.57 | 87 | |
| Human2024.08 | 3.52 | 3.22 | 2.66 | 48 | |
| IQA-EVAL-ClaudeEvaluator Backbone=Claude2024.08 | 3 | 3.23 | 2.07 | 57 | |
| IQA-EVAL-GPT4Evaluator Backbone=GPT-42024.08 | 2.1 | 3.03 | 2.37 | 67 |