Scan to Download
Search "大模型" has a total of4related results
0 篇文章
Techub News reports that OpenAI may temporarily delay the release of its next-generation model for inspection. (Bloomberg Technology)
Techub News reports that APUS AI (Qilin Hesheng Network Technology Co., Ltd.) has announced the open-source release of the Personal Agent product Reflex Agent. This product is built on the self-developed decision model Reflex, combined with the Continuum memory engine. Through model-driven core control logic, it achieves more efficient decision-making and long-cycle task execution. According to APUS AI laboratory's actual tests, Reflex Agent's response speed in relevant test scenarios can be up to 3 times faster than mainstream Agents such as Claude Code and Codex. Its working mechanism is called Loop Working, emphasizing the Agent's continuous action focused on user goals. Reflex Agent combines the Continuum memory engine to continuously accumulate task information through unlimited memory, supporting the realization of truly infinite context. This open-source release uses the MIT license and offers desktop, Web, and SDK versions. (Quantum Bit)
Techub News reported that Dawn Intelligence, in collaboration with Tianlong Mobile, has released the full-body, multi-modal data acquisition system SeePhys Hub. This system consists of a four-camera headband, wristbands, motion trackers, and tactile gloves. It restores full-body posture by integrating visual and motion signals and collects hand contact pressure data. The system aims to provide more comprehensive data for robots to learn full-body mobile dexterous operations. Yang Shuo, founder and CEO of Dawn Intelligence, stated that SeePhys Hub targets the middle layer of the data pyramid, attempting to balance the potential for large-scale data collection with more complete interaction records. The accompanying platform is responsible for data parsing, cleaning, action segmentation, and semantic annotation, and supports action retargeting to the robot body. The team plans to launch a full-modal world model in 2027, with priority given to application verification in vertical scenarios such as industry, catering, and retail. At this stage, Dawn Intelligence does not plan to independently develop robot bodies; it will use general robot body test models and conduct application pilots with partners. (Quantum Bit)
Techub News reports that Tether has announced the launch of a cross-platform BitNet LoRA fine-tuning framework within QVAC Fabric, enabling training and inference optimization for Microsoft BitNet (1-bit LLM). This framework significantly reduces computational and memory requirements, allowing billion-parameter-scale models to be trained and fine-tuned on laptops, consumer-grade GPUs, and even smartphones. This solution marks the first implementation of BitNet model fine-tuning on mobile GPUs (including Adreno, Mali, and Apple Bionic). Tests show that a 125M parameter model can be fine-tuned in approximately 10 minutes, a 1B model within about an hour, and it can even scale to a 13B parameter model on mobile devices. Furthermore, the framework supports heterogeneous hardware such as Intel, AMD, and Apple Silicon, achieving the first-ever 1-bit LLM LoRA fine-tuning on non-NVIDIA devices. In terms of performance, BitNet models achieve 2 to 11 times faster inference speeds on mobile GPUs compared to CPUs, while reducing memory usage by up to approximately 77.8% compared to traditional 16-bit models. Tether states that this technology has the potential to break the dependence on high-end computing power and cloud infrastructure, promote the decentralization and localization of AI training, and provide a foundation for new application scenarios such as federated learning.
Current search results fully loaded
情感类大模型0 篇文章
