From the archive
Tuesday, 20 October
Folded from 14 sources that day.
Sorted by quality and by how many sources covered it, the same for everyone.
-
MPS in PyTorch 2.15
PyTorch 2.15 adds FP8 on Apple MPS, grouped matmul for MoE decoding that runs 5–6.5x faster, an 8x faster softmax, and reduced video VAE memory use, and requires macOS 15 or later.