LLM Training Throughput on Qwen2.5-7B 128K context (train)
1,394.16Throughput (tokens/sec)OOMB (Sparse Attn)
Evaluation results
| Method | Links | |
|---|---|---|
| OOMB (Sparse Attn)Hardware=1xH200, Attention Type=Sparse, Tensor Parallel=false2026.02 | 1,394.16 | |
| OOMB (Sparse Attn)Hardware=4xH200, Attention Type=Sparse, Tensor Parallel=true2026.02 | 1,371.41 | |
| OOMB (Dense Attn)Hardware=1xH200, Attention Type=Dense, Tensor Parallel=false2026.02 | 504.12 | |
| OOMB (Dense Attn)Hardware=4xH200, Attention Type=Dense, Tensor Parallel=true2026.02 | 500.26 | |
| RFAHardware=4xH2002026.02 | 463.37 |