Short-form Question Answering on PopQA
50.6AccuracyCO-LMLM-360M-FW
Evaluation Results
| Method | Links | |
|---|---|---|
| CO-LMLM-360M-FWModel Size=360M, Training Corpus=FW2026.07 | 50.6 | |
| CO-LMLM-135MModel Size=135M2026.07 | 47.3 | |
| CO-LMLM-360MModel Size=360M2026.07 | 46.5 | |
| REL-LMLM-135MModel Size=135M2026.07 | 27.2 | |
| REL-LMLM-360MModel Size=360M2026.07 | 26.6 | |
| HF/SMOLLM2-1.7B*Model Size=1.7B, Training Token Count=11T2026.07 | 24 | |
| HF/SMOLLM2-360M*Model Size=360M, Training Token Count=4T2026.07 | 22.7 | |
| HF/SMOLLM2-135M*Model Size=135M, Training Token Count=2T2026.07 | 18.9 | |
| STANDARD-360M-FWModel Size=360M, Training Corpus=FW2026.07 | 18.9 | |
| STANDARD-360MModel Size=360M2026.07 | 16.2 | |
| STANDARD-135MModel Size=135M2026.07 | 15.2 |