Hugging Face releases transformers v5.13.1 patch to enable vLLM compatibility
Hugging Face released transformers v5.13.1, a patch focused on enabling compatibility with the latest vLLM release. The update includes defensive handling of legacy layer types, fixes for custom code with new linear layer names, and a fix for _LazyAutoMapping.register with string keys.
Entities
Related
vLLM v0.25.0 released with Model Runner V2 as default and PagedAttention removed
vLLM v0.25.0 is now available, featuring 558 commits from 232 contributors. Model Runner V2 becomes the default execution path for all dense models, with new support for EVS, realtime embeddings, prefix caching for Mamba hybrid models, multimodal-prefix bidirectional attention, and dynamic speculative decoding. The legacy PagedAttention implementation has been removed.
Hugging Face introduces Kernels repository type with redesigned CLIs and security
Hugging Face announced a new 'kernel' repository type on the Hub, along with a major redesign of the project including improved security, revamped CLIs, expanded framework/backend coverage, and a foundation for agentic kernel development. The updates aim to better serve users with compute-specific needs.
Xiaomi quietly uploads MiMo-V2.5-DFlash weights to Hugging Face
Xiaomi has uploaded the official DFlash weights for MiMo-V2.5-DFlash to Hugging Face. The model has 300B+ parameters and runs at 8-10 tokens/second on 2x24GB cards with offloading; DFlash could double that speed. The MTP head was shared but does not work yet.
LLM hardware recipe database with filters and community usage tracking
A community-driven database that lists which LLM models run on which hardware, with performance details. Users can filter by hardware, submit new recipes, and mark which recipes they actively use to show popularity. It solves the problem of finding compatible model-hardware combinations for LLM deployment.
LangChain-Mistralai v1.1.6 adds citation metadata support
LangChain released version 1.1.6 of its langchain-mistralai integration, adding support for surfacing citation metadata from Mistral AI chat responses. The update also includes refreshed model profiles and dependency bumps for vcrpy and langsmith.