Loading…
Hugging Face
AI platform and community providing machine learning models, datasets, libraries, and tools for building AI applications.
- 317 posts summarised
- latest
The Good Stuff
Paid picks we use ourselves. Links carry a referral code — a subscription through one earns us a commission at no extra cost to you.
From their blog
Latest articles
New in llama.cpp: Model Management
llama.cpp server now includes a router mode that enables dynamic loading, unloading, and switching between multiple LLMs without restarting the server. The architecture runs each model in an isolated process to ensure a single model crash does not affect other active instances. Running llama-server without specifying a model activates auto-discovery across the cache directory or a designated folder of GGUF files. In addition to on-demand loading and least-recently-used eviction capped by default at four models, the server provides endpoints for manual loading, unloading, and listing model statuses. Models inherit global configuration options or use dedicated configuration presets while also integrating directly into the built-in web UI.
Codex is Open Sourcing AI models
Hugging Face Skills equips AI coding agents like OpenAI Codex to execute end-to-end machine learning workflows. By reading AGENTS.md files and interfacing via the Model Context Protocol, Codex automates dataset validation, training script updates, and job submissions to Hugging Face Jobs. The workflow supports methods including supervised fine-tuning, direct preference optimization, and reinforcement learning for models ranging from 0.5B to 7B parameters. Throughout execution, Codex tracks live metrics via Trackio, records benchmark evaluations against baselines, and maintains Markdown reports. Once training concludes, Codex merges LoRA adapters, applies GGUF quantization, and publishes the resulting models to the Hugging Face Hub for local deployment.
Introducing swift-huggingface: The Complete Swift Client for Hugging Face
Hugging Face released swift-huggingface, a dedicated Swift package offering complete Hub API integration, reliable file downloads, and inference provider access. The library addresses previous limitations in swift-transformers 1.0, where interrupted multi-gigabyte model downloads could not resume and cache structures differed from Python. To resolve cache duplication, swift-huggingface implements a Python-compatible content-addressed storage layout using symlinks and flock file locking. Authentication is standardized through a TokenProvider pattern supporting auto-detection, Keychain integration, static CI/CD tokens, and OAuth 2.0 sign-in with automatic token refresh. The package also provides URLSession-backed snapshot downloads with granular progress tracking and will soon replace the HubApi implementation inside swift-transformers.
DeepMath: A lightweight math reasoning Agent with smolagents
Mathematical reasoning in large language models often suffers from lengthy chain-of-thought traces and frequent arithmetic mistakes. To address these issues, DeepMath pairs a Qwen3-4B Thinking base model with a sandboxed Python execution environment built using the smolagents library and vLLM backend. The system offloads deterministic calculations by emitting concise Python snippets, executing them safely with restricted module imports and no network access, and folding returned values back into the context. Training utilizes Group Relative Policy Optimization on the Tool-Integrated Reasoning subset of OpenMathReasoning with temperature scheduling and length constraints. Across benchmarks including MATH500, AIME, HMMT, and HLE, DeepMath reduces output token lengths by up to 66% while increasing overall problem-solving accuracy.
We Got Claude to Fine-Tune an Open Source LLM
Hugging Face Skills equips coding agents like Claude Code, OpenAI Codex, and Google's Gemini CLI to manage end-to-end language model fine-tuning. Using the hf-llm-trainer skill, an agent validates dataset formats, selects cloud hardware, configures authentication, and submits jobs to Hugging Face Jobs. Supported techniques include Supervised Fine-Tuning (SFT), Direct Preference Optimization (DPO), and Group Relative Policy Optimization (GRPO) for models ranging across various parameter sizes. The integration incorporates Trackio for real-time monitoring and automates LoRA configuration for larger models. Once training completes, the agent pushes artifacts to the Hugging Face Hub and can convert models to GGUF format for local execution.
Transformers v5: Simple model definitions powering the AI ecosystem
Transformers v5.0.0rc-0 introduces major architectural updates focused on simplicity, training, inference, and ecosystem interoperability across modern AI workflows. The release adopts a modular modeling approach and centralizes attention implementations into a unified AttentionInterface abstraction to reduce contribution and code review overhead. Support for Flax and TensorFlow is officially sunset in favor of focusing on PyTorch as the primary backend, while tokenization is standardized around the tokenizers library. For execution workloads, v5 adds native continuous batching, paged attention mechanisms, and a dedicated transformers serve OpenAI-compatible serving system. Finally, weight loading is refactored to make low-precision quantization a first-class citizen alongside broad interoperability with formats such as GGUF, MLX, and TorchAO.
Diffusers welcomes FLUX-2
Diffusers introduces support for FLUX-2 models through the Flux2Pipeline and Flux2Transformer2DModel classes. Running the 4-bit quantized checkpoint diffusers/FLUX.2-dev-bnb-4bit requires loading Mistral3ForConditionalGeneration as the text encoder and the transformer in bfloat16 precision with CPU offloading enabled. During image generation with a prompt and 50 inference steps, the pipeline encodes text embeddings through the Mistral 3 model before executing diffusion operations. However, executing the pipeline on a GPU with 14.56 GiB capacity triggers a CUDA OutOfMemoryError during 4-bit dequantization inside bitsandbytes matrix multiplication operations. Additionally, the execution logs warn that Flax classes are deprecated in Diffusers and will be removed in version 1.0.0.
Continuous batching from first principles
Large language model serving requires running expensive next-token generation across multiple concurrent user requests. Traditional batching approaches introduce severe padding inefficiencies when mixing variable-length prompts and different generation phases, especially under static shape constraints like CUDA graphs. Continuous batching resolves these inefficiencies by combining key-value caching, chunked prefill, ragged batching, and dynamic request scheduling. Ragged batching eliminates the traditional batch axis by concatenating token sequences into a single tensor and using boolean attention masks to isolate independent sequences. By dynamically removing completed prompts and packing decoding tokens alongside chunked prefill tokens up to a hardware memory budget, serving systems maintain high hardware utilization and throughput.
Building Deep Research: How we Achieved State of the Art
Building production AI research agents presents challenges around context window pollution, escalating token costs, and architectural brittleness across model updates. To resolve these issues, Tavily rebuilt its deep research system around simplified orchestration, compact tooling, and active context curation. The agent mimics human research workflows by distilling tool outputs into concise reflections for ongoing reasoning, withholding raw retrieved web data until the final deliverable stage. Compared to traditional ReAct propagation architectures where token consumption scales quadratically, this reflection-based approach achieves linear token growth. As a result, the system reduced token consumption by 66% compared to Open Deep Research while reaching state-of-the-art performance on DeepResearch Bench.
OVHcloud on Hugging Face Inference Providers 🔥
OVHcloud is now integrated as a supported Inference Provider on the Hugging Face Hub, expanding serverless inference options across model pages and client SDKs. The integration enables access to open-weight models, including gpt-oss, Qwen3, DeepSeek R1, and Llama, using European infrastructure with pay-per-token pricing starting at €0.04 per million tokens. Developers can connect via Python and JavaScript SDKs using either direct custom API keys or automatic routing through Hugging Face tokens. The service supports structured outputs, function calling, multimodal workflows, and embedding models while providing sub-200ms first-token response times. Calls routed through Hugging Face pass through standard provider pricing without markups, and PRO tier subscribers receive monthly inference credits.
20x Faster TRL Fine-tuning with RapidFire AI
Hugging Face TRL now officially integrates with RapidFire AI to accelerate LLM fine-tuning and post-training experimentation through concurrent configuration evaluation. Instead of executing training configurations sequentially, the platform shards datasets into discrete chunks and cycles multiple model configurations across GPUs at chunk boundaries. The runtime handles automatic checkpointing via a shared-memory-based adapter and model spilling mechanism to maintain consistent training state. Experimenters can monitor live metrics on an MLflow-based dashboard and use Interactive Control Ops to stop underperforming runs or clone and warm-start promising configurations mid-flight. Benchmark results on NVIDIA A100 GPUs demonstrate experimentation speedups ranging from 15x to 20x over sequential runs when training TinyLlama-1.1B and Llama-3.2-1B models.
Open ASR Leaderboard: Trends and Insights with New Multilingual & Long-Form Tracks
The Open ASR Leaderboard expanded its evaluation framework by introducing dedicated multilingual and long-form transcription tracks across eleven datasets. Benchmarking more than sixty open- and closed-source models from eighteen organizations revealed clear operational trade-offs between accuracy and inference speed. Systems combining Conformer encoders with large language model decoders achieved the lowest word error rates in short-form English speech recognition. In contrast, connectionist temporal classification and token-and-duration transducer decoders yielded tenfold to hundredfold improvements in throughput with only slight accuracy penalties. While closed-source commercial systems maintained performance leads in long-form audio transcription, specialized single-language models consistently outperformed generalized systems supporting hundreds of languages.
Introducing AnyLanguageModel: One API for Local and Remote LLMs on Apple Platforms
Apple developers frequently face integration friction when trying to support a mix of local and cloud language models across disparate APIs. To resolve this fragmentation, AnyLanguageModel introduces a unified Swift package that acts as a drop-in replacement for Apple's Foundation Models framework. The package standardizes interactions across local backends such as Core ML, MLX, and llama.cpp alongside cloud providers like Anthropic and OpenAI. To avoid dependency bloat from multi-backend support, the library employs Swift 6.1 package traits so projects only import the specific runtime engines they require. Furthermore, the library extends beyond current Foundation Models limitations by adding prompt image support for vision-language models.
Apriel-H1: The Surprising Key to Distilling Efficient Reasoning Models
Retrofitting efficiency into existing large language models often fails when naive distillation disrupts complex multi-step reasoning mechanisms. The Apriel-H1 project addresses this challenge by converting a 15B full-attention reasoning model into a Mamba hybrid using targeted distillation on high-quality SFT reasoning traces rather than diffuse pretraining data. Built on the open-source Fast-LLM framework, the process applies reverse KL divergence alongside a staged layer replacement strategy combining Leave-One-Out analysis and dynamic loss evaluations. Across fifty total blocks, replacing thirty attention layers with Mamba mixers yielded the flagship Apriel-H1-15b-Thinker-SFT checkpoint. This configuration achieved a 2.1x throughput increase with minimal benchmark degradation after 76.8B training tokens.
Easily Build and Share ROCm Kernels with Hugging Face
Compiling custom deep learning GPU kernels across architectures often leads to complex build flag management, compiler errors, and ABI issues. Hugging Face's kernels library and kernel-builder streamline this workflow by supporting backends such as ROCm, CUDA, Metal, and XPU. Using the RadeonFlow FP8 block-wise GEMM kernel optimized for the AMD Instinct MI300X GPU as an example, developers organize files into specific headers and HIP implementations. The build is configured via a build.toml manifest and flake.nix to ensure reproducible compilation across supported PyTorch and ROCm versions. Artifacts can then be pushed to the Hugging Face Hub, allowing users to load and execute custom operators directly via get_kernel without traditional installation steps.
Join the AMD Open Robotics Hackathon
AMD, Hugging Face, and Data Monsters announced the AMD Open Robotics Hackathon, an in-person competition scheduled for Tokyo and Paris in December 2025. Participating teams of up to four adults will compete over three days across two core missions. The first mission covers an instructor-led session on configuring the LeRobot development environment using AMD AI solutions. During the second mission, teams spend two days developing a freestyle project to address a real-world problem using LeRobot. Participants receive access to SO-101 robotics kits, AMD Ryzen AI laptops, and AMD Instinct MI300X GPUs via AMD Developer Cloud, with prizes awarded to the top seven teams in each city based on creativity, difficulty, ease-of-use, and practicality.
Building for an Open Future - our new partnership with Google Cloud
Hugging Face and Google Cloud announced an expanded strategic partnership designed to streamline the deployment and management of open models on Google Cloud infrastructure. Prompted by a tenfold increase in Hugging Face usage on Google Cloud over three years, the collaboration introduces a joint CDN Gateway using Hugging Face Xet technology to cache models and datasets directly on Google Cloud. This caching mechanism aims to shorten download times, strengthen model supply chain resilience, and accelerate time-to-first-token across Vertex AI, Google Kubernetes Engine, Cloud Run, and Compute Engine virtual machines. Hugging Face plans to integrate native library support for Google Cloud TPUs, lower instance prices on Inference Endpoints, and enhance Hub security scanning via VirusTotal, Google Threat Intelligence, and Mandiant.
Aligning to What? Rethinking Agent Generalization in MiniMax M2
MiniMax M2 addresses the performance gap between agent benchmark scores and real-world deployment across diverse scaffolding frameworks and environments. While initial alignment efforts relied on scaling tool variety, models continued to fail when subjected to slight variations in prompts, scaffolding, and tool responses. To resolve this instability, the team introduced interleaved thinking, which enables models to execute internal reasoning at arbitrary steps throughout long-horizon tasks to adapt dynamically to external tool outputs. Additionally, developers constructed a data pipeline designed for full-trajectory generalization by introducing perturbations across system prompts, user objectives, environments, and tool feedback. In internal evaluations, MiniMax M2 maintained robust tool-calling and instruction-following performance when tested against unfamiliar, cold-start agent scaffolding frameworks.
On the Shifting Global Compute Landscape
United States export controls on advanced artificial intelligence hardware have catalyzed rapid expansion across China's domestic silicon and open-weight model ecosystem. Faced with restricted access to high-end NVIDIA GPUs, Chinese developers accelerated deployments on domestic accelerators, including Huawei Ascend, Cambricon, and Baidu Kunlun. Hardware scarcity spurred architectural and algorithmic innovations in compute efficiency, notably DeepSeek's Multi-head Latent Attention and Group Relative Policy Optimization, alongside substantial post-training cost reductions. Organizations such as Baidu and Ant Group now train foundation models directly on domestic hardware, fostering non-CUDA software stacks and lowering inference costs globally. Consequently, the global artificial intelligence infrastructure is shifting from an exclusively American-focused paradigm toward a dual-ecosystem landscape powered by domestic chips and open-weight architectures.
Building a Healthcare Robot from Simulation to Deployment with NVIDIA Isaac
Simulation in healthcare robotics has historically been slow, siloed, and difficult to translate into real-world physical systems. The NVIDIA Isaac for Healthcare v0.4 release introduces an end-to-end SO-ARM starter workflow to streamline simulation, training, and deployment for embodied surgical assistants. The architecture relies on a Sim2Real mixed training approach that fine-tunes the GR00T N1.5 foundation model on combined real-world and synthetic datasets using LeRobot. Synthetic data generated in simulation constitutes over 93 percent of the policy training data to bridge physical data collection constraints. Real-time inference on the SO-ARM101 precision manipulator utilizes RTI DDS communication, dual-camera vision, and TensorRT model conversion.
Topics they write about
All topics →Weekly digest
The best engineering reads, once a week.