Models & Labs
Nunchaku 4-bit Inference Integrated into Diffusers
Hugging Face Bloghigh confidence
Why it matters
- →Enables significant memory savings and faster inference times with 4-bit quantization.
- →Simplifies the deployment of efficient AI models by integrating directly into Diffusers.
- →Democratizes access to advanced quantization techniques for developers using NVIDIA GPUs.
Hugging Face has announced the integration of Nunchaku's 4-bit diffusion inference into its Diffusers library. This allows developers to use Nunchaku-style checkpoints directly within Diffusers without needing a separate inference engine. The integration leverages SVDQuant, a quantization method that uses 4-bit weights and activations to reduce memory usage and improve inference speed. This update is particularly beneficial for users with NVIDIA's latest GPUs, offering a 30% speedup and significant memory savings. The integration simplifies the process, requiring no local compilation and supporting a wide range of models.
Read originalThe story around this
Earlier coverage that leads up to this article, and what followed. Lines connect each piece to the closest one after it, converging here.
More from Hugging Face Blog
© Hugging Face BlogModels & Labscoding
IBM releases LLMON markup to fix prompt injection
IBM is tackling the brittleness of small language models by introducing LLMON, a structured markup language that explicitly separates instructions from data. By using reserved special tokens instead of ambiguous characters like JSON brackets, these new Granite 4.1 preview models drastically reduce indirect prompt injection attacks while maintaining standard performance. This approach forces the model to bind specific constraints to specific data payloads, eliminating the confusion that often leads to hallucinations or security breaches in long contexts.
Hugging Face Blog
© Hugging Face BlogMore in Models & Labs
Models & Labsother
llama.cpp b12340 adds CUDA 13 and ROCm 10.0
This release quietly closes the hardware gap for local inference by adding default support for CUDA 13 and ROCm 10.0 across Linux and Windows. Users with newer NVIDIA GPUs or AMD accelerators no longer need to compile from source to get performance, while Apple Silicon builds now ship KleidiAI disabled by default, shifting the focus back to standard ARM optimizations. The inclusion of Snapdragon support for Linux and Android marks a significant step toward on-device AI beyond the x86/ARM desktop ecosystem. It is less about new features and more about ensuring llama.cpp remains the universal runtime for every available accelerator.
llama.cpp Releases
Models & Labscoding
llama.cpp b12342 adds few-row MMQ for speculative decoding
Speculative decoding just got a significant speed boost on NVIDIA GPUs. This release introduces a specialized kernel for 'few-row' matrix multiplications, targeting the small batch sizes typical of verification passes in speculative decoding workflows. By optimizing these specific operations for quantized formats like Q8_0 and IQ4 variants, llama.cpp reduces latency where it matters most during token generation. The change is CUDA-only for now but leverages hardware capabilities from Ampere onwards, making local inference faster without requiring new model architectures or complex configuration changes.
