Correlation with human evaluation on SEEDA Fluency (S)
0.975Pearson Correlation (r)IMPARA
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| IMPARAaggregation_procedure=w/ TrueSkill2025.02 | 0.975 | 0.952 | |
| SOMEaggregation_procedure=w/ TrueSkill2025.02 | 0.971 | 0.925 | |
| GPT-4 S (fluency)2025.02 | 0.952 | 0.916 | |
| GPT-4 S (meaning)2025.02 | 0.952 | 0.925 | |
| SOMEaggregation_procedure=w/o TrueSkill2025.02 | 0.931 | 0.916 | |
| IMPARAaggregation_procedure=w/o TrueSkill2025.02 | 0.887 | 0.938 | |
| GPT-4 E (fluency)2025.02 | 0.793 | 0.908 | |
| Scribendiaggregation_procedure=w/ TrueSkill2025.02 | 0.745 | 0.859 | |
| Scribendiaggregation_procedure=w/o TrueSkill2025.02 | 0.604 | 0.714 | |
| GREENaggregation_procedure=w/o TrueSkill2025.02 | 0.185 | 0.569 | |
| GLEU+aggregation_procedure=w/o TrueSkill2025.02 | 0.155 | 0.543 | |
| GLEU+aggregation_procedure=w/ TrueSkill2025.02 | 0.017 | 0.393 | |
| GREENaggregation_procedure=w/ TrueSkill2025.02 | -0.214 | 0.327 | |
| PTERRANTaggregation_procedure=w/ TrueSkill2025.02 | -0.366 | 0.182 | |
| ERRANTaggregation_procedure=w/ TrueSkill2025.02 | -0.463 | 0.095 | |
| PTERRANTaggregation_procedure=w/o TrueSkill2025.02 | -0.546 | 0.077 | |
| ERRANTaggregation_procedure=w/o TrueSkill2025.02 | -0.591 | -0.156 |