10 results on this page
A hyper-optimized Qwen3.8-27B inference stack using DFlash2 reaches about 138 tokens per second on a power-limited RTX 3090, while 64-request throughput reaches 942 TPS. Prefix caching also reduces long-chat follow-up latency from roughly 23 seconds to under 1.4 seconds in reported tests.
FedPref is a federated preference-learning method for converting free-text radiology reports into structured JSON without sharing reports or annotations. Across six simulated hospitals, it improved client-mean F1 by 2.49 points and worst-site F1 by 9.10 points over isolated local training, with the largest gains at data-poor sites.
AntLing has released six official Ling-3.0 base checkpoints spanning two model sizes and three training stages. The MIT-licensed, ungated repositories are intended for continued pretraining, fine-tuning, and research rather than direct chat use.
Tencent's Hunyuan Hy4 has reportedly appeared in the model selection list of the Yuanbao app under an expert-level label and with tool-use capabilities. It is positioned above Hy3 and alongside DeepSeek, following Tencent's recent statement that a larger-parameter Hy4 would launch soon with improved performance and multimodal abilities.
An independent builder pretrained a 1.02-billion-parameter Kimi-K3 replica on 5 billion decontaminated tokens for roughly $250. The model achieved 33.4% on HellaSwag, exceeding the reported 28% score of GPT-2 124M despite receiving no instruction tuning.
ByteDance's Coze launched a desktop client with cloud-drive integration and deep screenshot capabilities. The client improves local file access and automatically recognizes Codex CLI and Claude Code as agents for multi-agent workflows.
NVIDIA FLARE enables organizations to coordinate training for vision-language models while keeping sensitive image and text data at local institutions. The approach supports multimodal use cases such as visual question answering, captioning, and image-text reasoning without centralizing raw records.
Unsloth released new Qwen3.8-27B GGUF quantizations claiming 10% higher accuracy at the same file size. The release also includes 1-bit quantizations that retain 77% accuracy and can run with 8GB of RAM.
Loudoun County, Virginia, has used more than 250 data centers to generate substantial local wealth and reduce property taxes. Its success is attracting attention from other jurisdictions despite growing national opposition to data center expansion.
Pennsylvania Governor Josh Shapiro signed an executive order making the state more restrictive for AI data centre development. The order removes data centres from the permit fast-track programme, delays environmental review until local approval, and prohibits state agencies from signing non-disclosure agreements on projects.