devotd.
Browse the Asia/Seoul (KST) issue calendar. Retrospective entries show their assigned coverage day; actual publication times remain separate.
A project for running large AI models on personal PCs has changed its name and consolidated several model engines into one repository. The result of about 27 tokens per second on one GPU required ample RAM and fast storage. It does not imply quality parity with official services.
Support for matrix multiplication with ultra-low-bit weights, row extraction, and Hadamard operations has been added to Snapdragon NPUs. A developer said the Bonsai-2 1-bit and 2-bit paths are usable, but further optimization is needed to better exploit the reduced memory footprint.
Perplexity changed its request format and training data to assess multiple questions about the same context at once. Grouping related attributes helps, but independent candidates fare better separately. Long inputs and execution-path differences need checking.
OpenMOSS has released code and weights for image generation models that let users position text and objects with coordinates. It offers base and fine-tuned models. Adoption requires weighing layout input effort against memory and inference time.
Cloudflare and OpenAI say a pilot using network signals discovered hundreds of millions of new webpages and added them to a search index. Signals of changes and potential harm aim to reduce unnecessary crawling. Quantitative results on load reduction and actual business outcomes have not yet been provided.
Tencent has released a draft model that generates answer candidates in parallel, along with inference code. Tests showed longer accepted drafts than the comparison methods. Actual speed gains must be weighed against extra resource costs.
NVIDIA introduced roughly 2.8 trillion tokens of translation-based training data for 22 Indian languages and an open generation process. The planned release supplements locally authored text. Reported gains come from a training stage using both types of data.
Cloudflare introduced a system that continually evaluates verified bots' compliance with instructions and responses to requests to slow down. It is building an internal observation history, with public scores and customer-selected trust criteria planned for the future.
Users can now search Hugging Face Hub models in the web interface and download inference auxiliary files matched to a quantization. Paused downloads survive restarts, but start over if files change or resume requests are rejected.
NVIDIA has released a 4-billion-parameter language model that adjusts accuracy and compute by changing internal loop counts without retraining. It offers last-loop caching and multiple precision formats, but latency and cost need task-specific evaluation.