Hugging Face TRL officially supports AMD GPUs, full test suite runs on MI300X per PR
QGallouedec · x · 2026-10-21
Hugging Face's TRL training library now officially supports AMD Instinct GPUs. It largely worked before; what's new is an explicit support commitment with deep testing — the full test suite now runs on MI300X for every PR.
Per the docs, PyTorch's ROCm build exposes AMD GPUs as cuda devices, so TRL trainers run without code changes. Install the ROCm PyTorch build plus TRL in one environment. Caveat: the FlashAttention kernels used in doc examples (flash-attn2/3/4) have no ROCm build — on AMD GPUs use kernels-community/aiter-flash-attn, downloaded from the Hub.
More from Infra
- NVIDIA Vera Rubin reportedly in full production, demo shows on-chip photonics — LinusEkenstam · 2026-10-21
- NVIDIA Showcases DSX AI Factory Blueprint with GB300 NVL72 Racks — testingcatalog · 2026-10-21
- Data center casualties hit 8 in a month as AI buildout worries grow — AIFlow_ML · 2026-10-21
- Nvidia says Vera Rubin node now built every 5 minutes, unveils $50bn 1GW DSX turnkey AI factory — LinusEkenstam · 2026-10-21
- Jensen Huang unveils Nvidia DSX: $50bn turnkey 1GW AI compute factories — LinusEkenstam · 2026-10-21
- Nextdc Implodes 45% as Angry Residents Cut Power to Melbourne Data Center — AIFlow_ML · 2026-10-21