CUDA-LLM kernels task suite
Benchmarks
| Task Name | Dataset Name | SOTA Result | Trend | |
|---|---|---|---|---|
| Mean Square Error | CUDA-LLM kernels task suite 1.0 (test) | Execution Time4.79 | 9 | |
| Matrix Multiplication | CUDA-LLM kernels task suite 1.0 (test) | Execution Time4.4 | 9 | |
| Multi-Head Self-Attention | CUDA-LLM kernels task suite Multi-Head Self-Attention | Execution Time (s)6.85 | 5 | |
| Max Pooling 3D | CUDA-LLM kernels task suite Max Pooling 3D | Time (s)5.62 | 5 | |
| Multi-Head Self-Attention | CUDA-LLM kernels task suite 1.0 (test) | Execution Time6.85 | 5 | |
| 2D Convolution | CUDA-LLM kernels task suite 1.0 (test) | Execution Time6.15 | 5 | |
| LayerNorm | CUDA-LLM kernels task suite 1.0 (test) | Time5.4 | 5 | |
| Max Pooling 3D | CUDA-LLM kernels task suite 1.0 (test) | Execution Time (ms)5.62 | 5 | |
| Sigmoid | CUDA-LLM kernels task suite 1.0 (test) | Execution Time (ms)4.81 | 5 | |
| Max Pooling 3D | CUDA-LLM kernels task suite (test) | Latency (s)- | 0 | |
| Matrix Multiplication | CUDA-LLM kernels task suite (test) | Execution Time (s)- | 0 |
Showing 11 of 11 rows