transcribe.cpp

Evidence (18)

• 8The New Stack·8/5/2026research
  Google reshuffles DeepMind leadership as four senior engineers launch Discovery Loop

  Google is reorganizing DeepMind leadership, with Demis Hassabis moving to a chair/chief scientist role and Koray Kavukcuoglu taking broader control of Gemini model development and related products. At the same time, Jeff Dean, Sanjay Ghemawat, Oriol Vinyals and Quoc Le are leaving to found Discovery Loop, a public-benefit company focused on using AI to automate scientific and engineering research, with Google remaining an investor and cloud provider.

• 7Hacker News·7/19/2026open_source
  transcribe.cpp launches as a ggml-based speech-to-text library

  transcribe.cpp is a new v0.1.0 transcription library built on ggml, aiming to support a wide range of modern speech-to-text models with GPU acceleration. The author says each model is numerically validated and WER-tested against reference implementations, and the project includes bindings for Python, JavaScript/TypeScript, Rust, and Swift/ObjC.

• 5Hacker News·7/14/2026research
  LeMario Trains a JEPA World Model on Super Mario Bros

  A developer reproduced the LeWorldModel architecture and trained a Joint-Embedding Predictive Architecture (JEPA) model on Super Mario Bros. The write-up reports that the model could predict short-horizon futures and support limited reward-free planning, but failed to reliably make long-term progress through the game.

• 8Hugging Face Blog·7/8/2026framework_update
  Transformers backend in vLLM reaches native speed

  Hugging Face says the transformers modeling backend in vLLM now matches or exceeds the throughput of hand-written native vLLM implementations for several Qwen3 models. The update lets model authors run Hugging Face Transformers models in vLLM with a single flag while keeping standard parallelism and serving workflows intact.

• 7Hacker News·7/2/2026research
  Single Transformer Layer Matches Full RL Training

  A research paper on arXiv reports that a single Transformer layer can match the performance of a full-parameter reinforcement learning training setup on the studied task. The result suggests that much of the benefit may come from a smaller portion of the model than expected, though it is still an early research finding.

• 6Hugging Face Blog·6/29/2026research
  DiScoFormer unifies density and score estimation in one transformer

  Hugging Face highlights DiScoFormer, a transformer-based model that estimates both a distribution’s density and score from a sample in a single forward pass. The approach uses a shared backbone with separate heads and a consistency loss, and the authors show that attention can reproduce kernel density estimation as a special case while extending it to higher-dimensional settings.

• 7Hugging Face Blog·6/24/2026framework_update
  NVIDIA NeMo AutoModel boosts MoE fine-tuning in Transformers v5

  Hugging Face describes how NVIDIA NeMo AutoModel builds on Transformers v5 to improve fine-tuning performance for Mixture-of-Experts models. Using the same from_pretrained() API, it adds Expert Parallelism, DeepEP all-to-all dispatch, and TransformerEngine kernels, delivering higher throughput and lower GPU memory use.

• 5Hugging Face Blog·6/23/2026framework_update
  Transformers.js tests proposed Cross-Origin Storage API for shared caching

  Hugging Face describes experiments in Transformers.js with the proposed Cross-Origin Storage API to reduce duplicated downloads and storage across different origins. The post shows that model files and shared WebAssembly runtime assets can be cached once and reused, instead of being fetched again for each site using the same resources.

• 5Hugging Face Blog·6/18/2026research
  Hugging Face asks whether alternatives can beat LoRA

  Hugging Face published a post examining parameter-efficient fine-tuning methods and asking whether LoRA is truly the best option. The article explains why PEFT matters, highlights LoRA’s dominance in usage, and suggests that its popularity may be reinforced by visibility and ecosystem support rather than performance alone.

• 7Hugging Face Blog·5/19/2026model_release
  OlmoEarth v1.1 cuts satellite model compute costs up to 3x

  Allen Institute for AI released OlmoEarth v1.1, a new family of Earth observation models built to improve efficiency while preserving performance on benchmarks and partner tasks. The update focuses on shorter token sequences for satellite imagery processing, enabling lower compute cost, faster inference, and cheaper deployment at scale.

• 8Hugging Face Blog·5/14/2026model_release
  IBM Granite Releases Multilingual Embedding R2 Models

  IBM has released two Apache 2.0 multilingual embedding models based on ModernBERT: a 97M-parameter compact model and a 311M full-size model. The models support 200+ languages, 32K-token context, and code retrieval across nine programming languages, with strong retrieval benchmark results and Matryoshka support for the larger model.

• 6Hugging Face Blog·4/23/2026open_source
  Using Transformers.js in a Chrome Extension

  Hugging Face published a guide on building a Chrome extension that runs local AI features with Transformers.js under Manifest V3. The article outlines an architecture where a background service worker hosts the model, a side panel provides chat UI, and a content script handles page extraction and highlighting. It also shares practical notes on runtime constraints, model loading, and messaging between extension contexts.

• 8Hugging Face Blog·3/9/2026framework_update
  LeRobot v0.5.0 expands robotics support across hardware, policies, and datasets

  LeRobot v0.5.0 is a major release with support for the Unitree G1 humanoid, OpenArm hardware, new CAN bus motor controllers, and additional robot integrations. It also adds new policies such as Pi0-FAST and Real-Time Chunking, faster streaming video dataset encoding, EnvHub for loading simulation environments from Hugging Face Hub, and a modernized codebase built on Python 3.12 and Transformers v5.

• 7Hugging Face Blog·3/9/2026framework_update
  Ulysses Sequence Parallelism for Million-Token Training

  Hugging Face describes how Ulysses Sequence Parallelism can distribute long-context attention across multiple GPUs, making training on sequences far beyond single-GPU limits more practical. The post explains the approach and its integration into Accelerate, Transformers Trainer, and TRL's SFTTrainer, with comparisons to Ring Attention and guidance for large-sequence training.

• 7Hugging Face Blog·2/26/2026research
  Mixture of Experts in Transformers

  Hugging Face explains how Mixture of Experts (MoE) architectures replace dense feed-forward layers with sparse expert routing to improve compute efficiency. The post highlights why MoEs can deliver dense-model quality with fewer active parameters at inference time, and notes increasing adoption in recent open models and the Transformers ecosystem.

• 7Hugging Face Blog·2/13/2026open_source
  Claude and Codex Gain CUDA Kernel Writing Skill

  Hugging Face introduced an agent skill that teaches coding agents how to write production CUDA kernels and integrate them with PyTorch-based projects. The blog says Claude and Codex were used to generate working kernels for a diffusers pipeline and a transformers model, including bindings and benchmarks end to end.

• 7Hugging Face Blog·2/9/2026framework_update
  Transformers.js v4 arrives on npm

  Transformers.js v4 is now available on npm after a year of development. The release introduces a rewritten WebGPU runtime in C++, broader model support, and performance improvements that extend to browsers, Node, Bun, and Deno.

• 8Hugging Face Blog·1/5/2026model_release
  Hugging Face Debuts Falcon-H1-Arabic

  Hugging Face announced Falcon-H1-Arabic, a new Arabic language model family built on a hybrid Mamba-Transformer architecture. The release includes 3B, 7B, and 34B models with expanded context windows up to 256K tokens and aims to improve long-context understanding, dialect handling, and reasoning in Arabic NLP.