Inference Time on Extreme Multi-label Classification on Amazon 3M
35.2Latency (ms/sample)XR-Transformer-1
Evaluation Results
| Method | Links | |
|---|---|---|
| XR-Transformer-1GPU=Nvidia V100, batch size=12021.10 | 35.2 | |
| X-Transformer-1GPU=Nvidia V100, batch size=12021.10 | 50.2 | |
| AttentionXML-1GPU=8 Nvidia V100, batch size=1, model parallel=true2021.10 | 130.5 |