PMI ranking estimation on GoEmotions (500 held-out pairs)
0.66Spearman RhoDecomposed PMI (emp.)
Evaluation Results
| Method | Links | |
|---|---|---|
| Decomposed PMI (emp.)Model=Claude Sonnet 4, Evaluation Type=Empirical marginal (oracle upper bound)2026.05 | 0.66 | |
| InfoNCE (emp.)Model=Claude Sonnet 4, Evaluation Type=Empirical marginal (oracle upper bound)2026.05 | 0.61 | |
| Decomposed PMI (emp.)Model=GPT-5.2, Evaluation Type=Empirical marginal (oracle upper bound)2026.05 | 0.56 | |
| PromptNCE (emp.)Model=Claude Sonnet 4, Evaluation Type=Empirical marginal (oracle upper bound)2026.05 | 0.55 | |
| InfoNCE (emp.)Model=GPT-5.2, Evaluation Type=Empirical marginal (oracle upper bound)2026.05 | 0.53 | |
| PromptNCE (emp.)Model=GPT-5.2, Evaluation Type=Empirical marginal (oracle upper bound)2026.05 | 0.49 | |
| PromptNCEModel=Claude Sonnet 4, Evaluation Type=Zero-shot2026.05 | 0.47 | |
| MarginalNCEModel=Claude Sonnet 4, Evaluation Type=Zero-shot2026.05 | 0.45 | |
| Decomposed PMIModel=GPT-5.2, Evaluation Type=Zero-shot2026.05 | 0.44 | |
| PromptNCEModel=GPT-5.2, Evaluation Type=Zero-shot2026.05 | 0.43 | |
| MarginalNCEModel=GPT-5.2, Evaluation Type=Zero-shot2026.05 | 0.41 | |
| Decomposed PMIModel=Claude Sonnet 4, Evaluation Type=Zero-shot2026.05 | 0.4 | |
| Direct PMIModel=GPT-5.2, Evaluation Type=Zero-shot2026.05 | 0.35 | |
| Direct PMIModel=Claude Sonnet 4, Evaluation Type=Zero-shot2026.05 | 0.34 | |
| InfoNCEModel=Claude Sonnet 4, Evaluation Type=Zero-shot2026.05 | 0.18 | |
| InfoNCEModel=GPT-5.2, Evaluation Type=Zero-shot2026.05 | 0.13 |