Coreference Resolution on Winogender (test)
80.7AccuracyGLM-130B
Evaluation results
| Method | Links | ||
|---|---|---|---|
| GLM-130BNumber of shots=12022.10 | 80.7 | — | |
| GLM-130BNumber of shots=02022.10 | 79.7 | — | |
| PaLM 540BNumber of shots=12022.10 | 79.4 | — | |
| ChinchillaNumber of shots=02022.10 | 78.3 | — | |
| PaLM 540BNumber of shots=02022.10 | 75 | — | |
| Gopher 280BNumber of shots=02022.10 | 71.4 | — | |
| GPT-3 (Davinci)Number of shots=02022.10 | 64.2 | — | |
| GPT-3 (Davinci)Number of shots=12022.10 | 62.6 | — | |
| OPT 175BNumber of shots=02022.10 | 54.8 | — | |
| BLOOM 176BNumber of shots=12022.10 | 53.1 | — | |
| BLOOM 176BNumber of shots=02022.10 | 49.1 | — | |
| HATified-SFTShot=5-shot2026.03 | — | 67.9 | |
| Llama-InstructShot=5-shot2026.03 | — | 84.3 |