inference/research
verifiedhigh · 0.95updated 2026-10-03

MLPerf Inference v4.0 closed results: Intel-HabanaLabs HLS-Gaudi2-PT Llama2-70B Offline/Server logs and system definition

benchmark

Fields

id
sources/bench-intel-gaudi2-mlperf-v4-0-llama2-70b
URL
https://raw.githubusercontent.com/mlcommons/inference_results_v4.0/main/closed/Intel-HabanaLabs/results/HLS-Gaudi2-PT/llama2-70b-99.9/Offline/performance/run_1/mlperf_log_summary.txt
Publisher
MLCommons / Intel
Kind
benchmark
Accessed
2026-10-03
notes
MLPerf Inference v4.0 raw performance log for Intel Gaudi2 (HLS-Gaudi2-PT) running llama2-70b-99.9 Offline. Result: 8,034.85 tokens/sec, 27.56 samples/sec. VALID result.

Notes

MLPerf Inference v4.0 raw performance log for Intel Gaudi2 (HLS-Gaudi2-PT) running llama2-70b-99.9 Offline. Result: 8,034.85 tokens/sec, 27.56 samples/sec. VALID result.

Cross-references

This record references no other records.