List your company

ResearchFormerly WizWand

Training Efficiency Profiling

Benchmarks

DatasetSOTA methodMetricTrendResultsLast updated

Qwen3 family S=512 (train profiling)

100.2Per-Step Latency (ms)

43

Jun 23, 2026

Qwen3 family S=1024 (train profiling)

100.6Per-Step Latency (ms)

42

Jun 23, 2026

Qwen3 profiling S=2048 (train)

104.1Per-Step Latency (ms)

40

Jun 23, 2026

Qwen3 profiling S=4096 (train)

108Per-step Latency (ms)

39

Jun 23, 2026

3.1Peak GPU Memory

31

Jun 23, 2026