Search AI Products and News

Large Language Models (LLMs) have achieved significant progress in complex reasoning tasks by combining task prompts with large-scale reinforcement learning (RL), as demonstrated by models like Deepseek-R1-Zero, which directly apply reinforcement learning to base models, showcasing strong reasoning capabilities. However, this success is difficult to replicate across different base model families, especially within the Llama series. This raises a core question: what factors lead to inconsistent performance of different base models during reinforcement learning? How does reinforcement learning perform in

The plan's advancement has triggered dissatisfaction among developers. According to VentureBeat, OpenAI's current API deprecation involves the GPT-4.5 Preview version, which is an intermediary model launched between GPT-4 and GPT-5. Multiple developers have stated that this sudden change has disrupted their development plans, forcing significant adjustments to applications already built based on this API. A developer who wished to remain anonymous commented: We understand the necessity of technological iteration, but the lack of sufficient buffer time in the schedule has brought additional pressure to development teams.

Recently, Baidu officially launched its 2026 AIDU Program, an elite recruitment project aimed at cultivating future AI technology leaders. This program has once again become a focal point in the tech community. Baidu hosted an AIDU Program • DeepTalk Open Day event in their campus, attracting many young people who are passionate about AI technology to participate. In this year's AIDU Program, there has been a significant expansion in job recruitment. The scale of recruitment has increased by over 60%, covering 23 core businesses and 11 research directions of Baidu. Positions include large model algorithms and large model infrastructure.

Recently, the DeepSeek team released a technical paper about their latest model, DeepSeek-V3, focusing on the challenges of scaling in large AI model training and related considerations of hardware architecture. This 14-page paper not only summarizes the experiences and lessons learned during the development of V3 by DeepSeek but also provides deep insights for future hardware design. Notably, CEO of DeepSeek, Liang Wenfeng, also participated in writing the paper.

In the fierce competition for talent, vivo recently announced the launch of a global talent recruitment program named 'Blue Star'. This plan focuses on high-tech fields such as chips and artificial intelligence large models. The purpose of this program is to attract the top doctoral students from around the world to further promote vivo's technological innovation in research and development. The 'Blue Star Program' is regarded as the core talent strategy of vivo. It plans to recruit about 100 experts. These positions cover several technical fields, including virtual reality (XR), artificial intelligence large models, image processing, chip design, etc.