RITS

Tag: Hardware

  • Ollama: Lightweight Framework for Language Models

    Key points Ollama is a lightweight, extensible framework for building and running language models on local machines. It provides a simple API for creating, running, and managing models, as well as a library of pre-built models. Ollama is a lightweight, extensible framework for building and running language models on local machines. It provides a simple […]

  • DeepSparse Enables LLM Inference on Commodity CPUs

    Key points The post states that DeepSparse allows running LLM inference on commodity CPUs with usable speeds. DeepSparse allows running LLM inference on commodity CPUs with usable speeds.

  • NVIDIA Supports Stable Diffusion via TensorRT

    Key points The post reports that NVIDIA now supports Stable Diffusion via TensorRT, enabling up to 2.75 times faster generations. NVIDIA now supports Stable Diffusion via TensorRT, enabling up to 2.75 times faster generations.

  • OpenVINO from Intel for AI Applications

    Key points Intel’s OpenVINO is now available for AI applications. Intel’s OpenVINO is now available for AI applications.

  • Stable Diffusion Rival from Huawei

    Key points Huawei has released a rival to Stable Diffusion that it promises delivers better results with improved training data. Huawei has released a Stable Diffusion rival that promises better results with improved training data.

  • LLAMA2 with 4 million tokens context

    Key points The article states that LLAMA2 has 4 million tokens of context and links to the mit-han-lab streaming-llm repository on GitHub. LLAMA2 with 4 million tokens context https://github.com/mit-han-lab/streaming-llm

  • MLC Group Enables Web and Mobile LLM Usage

    Key points The MLC group has launched a platform allowing users to run LLAMA models on the web and mobile devices. The web version offers a chat demo with model download, while the iOS and Android apps enable running 3B models on smartphones. The 3B model runs on an iPhone 13 with a response speed […]

  • ExLLaMa improves Open Chat speed tenfold

    Key points The article says ExLLaMa has arrived and brings a significant improvement in the speed of Open Chat, which is now ten times faster. It notes a remaining context limitation of 4k tokens, making the setup unsuitable for handling excessively lengthy queries. ExLLaMa has arrived, bringing with it a significant improvement in the speed […]

  • Shanghai AI Lab Develops New Model

    Key points A new model developed by Shanghai AI Lab carries the Chinese name 书生·浦语 (Shūshēng · Pǔyǔ). The model’s technical report is available at github.com/InternLM/InternLM-techreport. A new model developed by Shanghai AI Lab, with the Chinese name ‘书生·浦语’ (Shūshēng · Pǔyǔ). The model’s technical report can be found here: https://github.com/InternLM/InternLM-techreport/blob/main/InternLM.pdf.

  • Quest 3 released

    Quest 3 is out.