Hugging Face TRL fix halves a Qwen3 H100 step in SDPA test

Hugging Face contributor tests report an H100 SDPA step falling from 2.06 to 1.02 seconds; on Trainium2, the revised code produced step timings where the previous path crashed.

By · Published

A smiling yellow plush figure holds blue and purple blocks beside server racks labeled “H100” and “Trainium2”.

Hugging Face contributor tests report an H100 SDPA step falling from 2.06 to 1.02 seconds; on Trainium2, the revised code produced step timings where the previous path crashed.