| Dataset Name | SOTA Method | Metric | Trend | ||
|---|---|---|---|---|---|
| Llama2 7B (32 Q-heads/32 KV-heads/128 Head-dimension) | flash-attn v2 | Attention TFLOPS207.3 | 30 | 4d ago | |
| Qwen2.5 72B (64 Q-heads/8 KV-heads/128 Head-dimension) | flash-attn v2 | Attention Throughput (TFLOPS)222.5 | 29 | 4d ago | |
| Llama 405B (128 Q-heads/8 KV-heads/128 Head-dimension) 3.1 | flash-attn v2 | TFLOPS225.3 | 28 | 4d ago |