Models & Labs
Cross-Origin Storage API in Transformers.js
Hugging Face Bloghigh confidence
Why it matters
- →The Cross-Origin Storage API could drastically reduce redundant data downloads across web apps.
- →It introduces a new way to manage resources using cryptographic hashes, enhancing efficiency.
- →This proposal could lead to significant bandwidth and storage savings for developers.
Hugging Face is exploring the Cross-Origin Storage API in Transformers.js to address redundant downloads of AI models and Wasm files across different web app origins. This API uses cryptographic hashes to identify resources, allowing them to be shared across origins without duplicating downloads. Currently, browser caches are isolated by origin, leading to unnecessary bandwidth and storage use. While the API is not yet implemented in browsers, developers can test it using a polyfill extension, potentially paving the way for more efficient web app resource management.
Read originalMore from Hugging Face Blog
© Hugging Face BlogAgentsagents
IBM's CUGA Simplifies Building Agentic Apps
IBM's CUGA, an open-source agent harness, is transforming how developers build agentic applications by handling the complex orchestration tasks typically required. By focusing on the configuration rather than the construction of agents, CUGA allows developers to concentrate on defining tools and prompts. This approach is demonstrated through two dozen single-file apps, showcasing its capability to manage planning, execution, and state without the need for extensive rewrites. The result is a more efficient development process that leverages smaller models effectively, offering a practical alternative to relying on large, resource-intensive models.
Hugging Face Blog
© Hugging Face BlogOpen Sourcecoding
Hugging Face Automates Weekly Releases with AI
Hugging Face has streamlined its release process for the huggingface_hub Python client, moving from a 4-6 week cycle to weekly releases. This shift is powered by a combination of open-source tools and AI, which drafts release notes and automates mechanical tasks, while humans oversee critical judgment areas. The process is designed to be replicable by other maintainers, emphasizing transparency and adaptability. This change not only accelerates the release cycle but also ensures that updates are consistently delivered without the need for proprietary tools.
Hugging Face Blog
© Hugging Face BlogModels & Labsmodels
PP-OCRv6 Launches on Hugging Face with 50-Language Support
PP-OCRv6 represents a notable advancement in OCR capabilities, offering a scalable model family from 1.5M to 34.5M parameters. This release significantly boosts text detection and recognition accuracy, supporting a wide array of languages including Chinese, English, and Japanese. Designed for practical applications, the models handle complex text scenarios with enhanced architecture and training techniques. Developers can deploy these models using PaddlePaddle, Transformers, or ONNX Runtime, making multilingual OCR more accessible and efficient across various platforms.
Hugging Face Blog
More in Models & Labs
Models & Labsmodels
llama.cpp b9767 Release Enhances MTP Inference
The b9767 release of llama.cpp introduces significant improvements to MTP inference by optimizing the mat-vec path for small batches, which enhances decoding efficiency. A new barrier in the NUM_COLS loop of the mul-mat-vec process is expected to boost performance. While no new model architectures are included, this update refines the platform's capabilities across macOS, Linux, and Windows. Notably, it supports macOS Apple Silicon, Ubuntu with ROCm 7.2, and Windows with CUDA 12 and 13. This release continues llama.cpp's focus on performance optimization and compatibility, making it a more powerful tool for developers.
llama.cpp Releases
Models & Labsmodels
Granite Speech Plus Support Added in b9768 Release
The b9768 release of llama.cpp expands its capabilities by integrating Granite Speech Plus, which enhances audio processing with multi-layer concatenation. This update is particularly relevant for developers focused on audio applications, as it resolves naming inconsistencies and standardizes feature layer usage. While no new models are introduced, the release fortifies the existing framework, making it more reliable for audio tasks. This iteration marks a refinement in the tool's functionality, especially for those utilizing its audio features.
llama.cpp Releases
Models & Labsmodels
Llama.cpp b9774 Release Enhances Vulkan Support
The latest b9774 release of llama.cpp brings significant improvements to Vulkan support, enabling backend tests for various mathematical operations like SQR, SQRT, SIN, and COS. This update also enhances the handling of noncontiguous data in norm operations, broadening the library's applicability across different platforms. While the release doesn't introduce new models, it strengthens the existing infrastructure, particularly for developers working with Vulkan and other supported platforms. This makes llama.cpp a more robust choice for those looking to leverage GPU capabilities beyond NVIDIA's CUDA ecosystem.
llama.cpp Releases