CUDA Code Validation on KernelBench Level 1
8Success Rate (p=1.0)qwen2.5-7b
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| qwen2.5-7bModel=qwen2.5-7b, Simulator Type=Real, Training Steps=402025.12 | 8 | 2 | 1 | 99.2 | |
| qwen2.5-72bModel=qwen2.5-72b2025.12 | 6 | 3 | 2 | 20.1 | |
| qwen2.5-7bModel=qwen2.5-7b, Simulator Type=GTM, Training Steps=402025.12 | 5 | 2 | 1 | 100.1 | |
| llama3.1-405bModel=llama3.1-405b2025.12 | 2 | 2 | 2 | 11.2 | |
| qwen2.5-7bModel=qwen2.5-7b2025.12 | 0 | 0 | 0 | 22.7 |