Language Modeling on OWT Pile
5.23Decode Latency
Longformer + SFA (k=8)
Updated 24d ago
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
2026.03 | 5.23 | 6.18 | 19.3 | |
2026.03 | 6.75 | 7.93 | 18.73 | |
2026.03 | 6.92 | 9.41 | 19.44 | |
2026.03 | 7.92 | 8.37 | 18.64 | |
2026.03 | 9.09 | 9.41 | 19.29 | |
2026.03 | 9.43 | 7.93 | 19.72 | |
2026.03 | 9.88 | 16.86 | 17.91 | |
2026.03 | 10.84 | 16.86 | 17.95 | |
2026.03 | 11.39 | 16.86 | 17.82 | |
2026.03 | 13.32 | 16.86 | 18.02 | |
2026.03 | 14.12 | 9.41 | 18.17 | |
2026.03 | 17.08 | 16.86 | 17.29 |