ChatGLM's cogVideoX now opensourced and can run from your home computer if you have a 4090.
Artificial Intelligence thread
- Thread starter 9dashline
- Start date
This require Chinese account# or open to all?
The first in China! Zhipu GLM end-side big model received the highest rating in the trusted AI end-side big model evaluation of CAICT
Zhipu GLM's end-side big model participated in the first round of evaluation of the "End-side Big Model Technology and Application Evaluation Method" standard organized by the China Academy of Information and Communications Technology, and finally obtained a level 4 rating. Zhipu became the first big model company in China to pass this evaluation and obtain the current highest rating.
GLM is a series of large models for terminal scenarios such as PCs, mobile phones, and car computers launched by Zhipu. It includes 1.5B, 6B, 9B and other specifications, and is flexibly adapted to various terminal application scenarios and computing power conditions, enabling AI applications in smart terminals and innovating user experience. Based on advanced pre-training technology and a large amount of high-quality data, GLM has functions such as text generation, dialogue interaction, web browsing, code execution, Function Call, and multi-modal interaction.
At present, the GLM end-side large model has completed the adaptation and performance optimization of many mainstream terminal chips, and supports GPU and NPU reasoning. At the same time, it has been adapted to AI PCs, AI mobile phones, smart cockpits and other devices of multiple brands, and has realized scenarios such as complex device control, device usage questions and answers, chat companionship, writing assistant, document processing, local knowledge base, life services, and multi-modal interaction. Through the end-cloud integration approach, the GLM end-side large model can also seamlessly collaborate with the cloud-side large model to perform multi-scenario, high-complexity tasks and provide a consistent experience.
Alibaba launches maths-specific AI models said to outperform LLMs from OpenAI, Google
is aiming to raise the bar in (AI) development by launching a group of maths-specific large language models (LLMs) called Qwen2-Math, which the giant claims can outperform the capabilities of in that field.
“Over the past year, we have dedicated significant efforts to researching and enhancing the reasoning capabilities of large language models, with a particular focus on their ability to solve arithmetic and mathematical problems,” the Qwen team, part of , said in a post published on developer platform GitHub on Thursday. Alibaba owns the South China Morning Post.
The latest LLMs – the technology underpinning services like – were built on the Qwen2 LLMs released by Alibaba in June and covers three models based on their scale of parameters – a machine-learning term for variables present in an AI system during training, which helps establish how data prompts yield the desired output.
The model with the largest parameter count, Qwen2-Math-72B-Instruct, outperformed proprietary US-developed LLMs in maths benchmarks, according to the Qwen team’s post. Those included GPT-4o, ’s Claude 3.5 Sonnet, 1.5 Pro and ’ -3.1-405B.
“We hope that Qwen2-Math can contribute to the community for solving complex mathematical problems,” the post said.
Yup all new OS need to be natively Multimodal AI... Microsoft is doing the same with Windows 12
will be very interesting to see how WeChat and alibaba deploy AI functions in their apps.