Lanzado el 26 de enero de 2026
What's Changed This release candidate was focused mostly on `quantization` support with the new dynamic weight loader, and a few notable 🚨 breaking changes🚨: 1. Default dtype for any model when using `from_pretrained` is now `auto`! * Default auto 🚨 🚨 by @ArthurZucker in
What's Changed This release candidate was focused mostly on `quantization` support with the new dynamic weight loader, and a few notable 🚨 breaking changes🚨: 1. Default dtype for any model when using `frompretrained` is now `auto`! Default auto 🚨 🚨 by @ArthurZucker in 2. Default shard size when saving a model is now 50GB: 🚨🚨 [saving] Default to 50GB shards, and remove non-safe serialization by @Cyrilvallez in This is now as fast as before thanks to xet, and is just more convenient on the hub. 3. Kwargs. They are fundamental to enable integration with vllm and other toosl: Every model forward() should have kwargs by @Rocketknight1 in Dynamic weight loader updates: Mostly QOL and fixed + support back CPU offloading. mark params as ishfinitialized with DS Zero3 from weight conversion by @winglian in [loading] Allow loading to happen without threading by @Cyrilvallez in [loading] Correctly load params during offloading & careful memory considerations by @Cyrilvallez in * allow registration of custom checkpoint conversion mappings by @winglian in
Genera un resumen en lenguaje claro de los cambios de este release, pensado para desarrolladores.