Search results

H200 MLPerf small_llm_pretraining: Unable to perform multi-node pretraining

Status: Open.

#840 In mlcommons/training;

· Ypp6657 opened on Oct 6, 2025

Llama3.1-8B converges faster than the RCPs

Status: Open.

Bug

#838 In mlcommons/training;

· psyhtest opened on Oct 3, 2025

C4 train download misses one part

Status: Open.

#837 In mlcommons/training;

· psyhtest opened on Sep 25, 2025

Add approximate reference runtime to get baselines

Status: Open.

#828 In mlcommons/training;

· ShriyaRishab opened on Sep 4, 2025

Gres defaults to 8 regardless of config.sh

Status: Open.

#824 In mlcommons/training;

· sean-smith opened on Aug 25, 2025

Git isn't initialized properly in <code>bash run_llama31.sh</code>

Status: Open.

#822 In mlcommons/training;

· sean-smith opened on Aug 24, 2025

LLama 3.1 toml package missing

Status: Open.

#821 In mlcommons/training;

· sean-smith opened on Aug 24, 2025

Llama 3.1 git: not a valid object name issue

Status: Open.

#820 In mlcommons/training;

· sean-smith opened on Aug 24, 2025

/dev/shm fills up completely and causes bus error

Status: Open.

#816 In mlcommons/training;

· adeepak7 opened on Aug 18, 2025

New reference benchmark development checklist

generalapplies to all benchmarksapplies to all benchmarks

Status: Open.

#810 In mlcommons/training;

· ShriyaRishab opened on Aug 1, 2025