MoE models
Benchmarks
| Task Name | Dataset Name | SOTA Result | Trend | |
|---|---|---|---|---|
| Kernel Throughput Evaluation | MoE Models (OLMoE, Qwen3, DSv3, Mixtral) beta=0.5 | Latency67 | 12 | |
| Expert Pruning Efficiency | MoE Models | Calibration Time (h)0.22 | 6 | |
| Loss curve fitting across model sizes | MoE models (various sizes) | ASMT MAPE0.341 | 3 |
Showing 3 of 3 rows