Multi-task Learning on CIFAR-100 (test)
0.01HVDYMU
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| DYMUc=0.0, adaptation=true, Inference Params=31.86M, GFLOPS=174.36, Control Params=3.10M2022.03 | 0.01 | 0.54 | |
| Single-TaskInference Params=36.18M, GFLOPS=348.792022.03 | 0.009 | — | |
| DYMUc=1.0, adaptation=true, Inference Params=31.37M, GFLOPS=129.23, Control Params=3.10M2022.03 | 0.009 | 0.49 | |
| PHN-BNInference Params=16.35M, GFLOPS=73.13, Control Params=0.31M2022.03 | 0.007 | 0.49 | |
| DYMU (w/o adaptation)c=0.0, adaptation=false, Inference Params=31.86M, GFLOPS=174.36, Control Params=0.34M2022.03 | 0.003 | 0.58 | |
| PHNInference Params=16.35M, GFLOPS=73.13, Control Params=11.03M2022.03 | 0.002 | 0.54 | |
| DYMU (w/o adaptation)c=1.0, adaptation=false, Inference Params=31.37M, GFLOPS=129.23, Control Params=0.34M2022.03 | 0.001 | 0.53 |