Image Classification on JFT-300M
Metric: prec@1 (higher is better)
Metric: prec@1 (higher is better)
| # | Model↕ | prec@1▼ | Extra Data | Paper | Date↕ | Code |
|---|---|---|---|---|---|---|
| 1 | V-MoE-H/14 (Every-2) | 60.62 | Yes | Scaling Vision with Sparse Mixture of Experts | 2021-06-10 | Code |
| 2 | V-MoE-H/14 (Last-5) | 60.12 | No | Scaling Vision with Sparse Mixture of Experts | 2021-06-10 | Code |
| 3 | V-MoE-L/16 (Every-2) | 57.65 | No | Scaling Vision with Sparse Mixture of Experts | 2021-06-10 | Code |
| 4 | VIT-H/14 | 56.68 | No | Scaling Vision with Sparse Mixture of Experts | 2021-06-10 | Code |