PyTorch 2.15 Rebuilds Distributed Training and Cuts Decoding Overhead

The new release flips the rewritten NCCL backend on by default, adds one-sided transports, and accelerates NVFP4 decode on Blackwell.

Read the full story at AlphaSignal ↗

Timeline · 1 report

2026-10-28 17:49 · AlphaSignal
PyTorch 2.15 Rebuilds Distributed Training and Cuts Decoding Overhead

More stories

8 examples of Gemini 4 Argon in action — Google Gemini Blog
Sources: the US DOJ has inquired about the impact of Nvidia's Hugging Face deal on AI competition, as antitrust enforcers continue to probe Nvidia's conduct (Reuters) — Techmeme
Google AI gets cagey when asked who won the 2020 elections — Ars Technica AI
OpenAI and ServiceNow expand partnership across enterprise workflows — OpenAI News
OpenAI launching election war room to track interference efforts — Politico Technology
PyTorch 2.15 Release Blog — PyTorch Blog
Google Debuts New AI Models, Genie 4 and Gemini 4 Carbon: What You Need to Know — CNET AI
CoreWeave Builds the First Global Backbone Network Purpose-Built for AI — CoreWeave Blog

Get the daily brief of stories like this at 6:30 every morning →