# Hugging Face: what I've logged

> Hugging Face comes up in 2 posts of my AI log, from October 2026 to October 2026, and each item below links to the primary source. The releases I logged include MPS in PyTorch 2.15, filed under open weights, agents and coding and frontier.

## Items

1. **MPS in PyTorch 2.15** (Hugging Face, 19 Oct 2026): FP8 arrives on Apple silicon, and mixture-of-experts decoding is reported 5 to 6.5 times faster than in 2.14 on an M3 Ultra. [Primary source](https://huggingface.co/blog/mps-pytorch-2-15) · [Post](https://heyaliasif.com/ai/2026-10-20-qwen38-flash-next-deepinfer-pytorch-mps#item-3)
2. **TorchNeuron: running HF models on AWS Trainium** (Hugging Face, 18 Oct 2026): A PyTorch backend that exposes Trainium as a native device, covering 89% of the top 100 text-generation architectures versus 46% before. [Primary source](https://huggingface.co/blog/hf-torchneuron) · [Post](https://heyaliasif.com/ai/2026-10-19-mercury-3-cern-mistral-cutile-torchneuron#item-4)

## Posts

- 2026-10-20 [Qwen3.8-Flash-Next, DeepSeek's DeepInfer, PyTorch 2.15 on MPS](https://heyaliasif.com/ai/2026-10-20-qwen38-flash-next-deepinfer-pytorch-mps)
- 2026-10-19 [Mercury 3, CERN and Mistral, Mamba-2 in cuTile, TorchNeuron](https://heyaliasif.com/ai/2026-10-19-mercury-3-cern-mistral-cutile-torchneuron)

Canonical: https://heyaliasif.com/ai/topics/hugging-face