| Dataset Name | SOTA Method | Metric | Trend | ||
|---|---|---|---|---|---|
| MHA causal mask head dimension 128 FP8 on NVIDIA L40S GPU (test) | LLM-TL | Performance (TFLOPS)257.9 | 6 | 1mo ago | |
| NVIDIA A100 GPU | DeepSeek-V3 + Ours | TFLOPS207.2 | 5 | 1mo ago | |
| NVIDIA RTX8000 GPU | DeepSeek-V3 + Ours | TFLOPS49.9 | 2 | 1mo ago |