Search AI Products and News

Programmers have a new tool to cheer about! LocAgent, a graph-indexed large language model (LLM) agent framework designed specifically for code localization, has been jointly developed by the research teams from OpenHands, Yale University, University of Southern California, and Stanford University. Excitingly, LocAgent's code localization accuracy reached 92.7%, and this new achievement will be officially presented at the 2025 ACL conference. Programmers often struggle with identifying the root cause of code problems in their work. For example

Recently, a collaborative effort between the Institute of Automation, Chinese Academy of Sciences, and the CASIA-Zidong Taichu team introduced Vision-R1, a novel method leveraging R1-like reinforcement learning to significantly enhance visual localization capabilities. This approach achieved a 50% performance improvement in complex tasks such as object detection and visual localization, surpassing existing state-of-the-art (SOTA) models with over ten times the parameter scale. Currently, large vision-language models typically rely on "pre-training + supervised fine-tuning" to improve responsiveness to user instructions, but this...

Dubformer, a company specializing in AI dubbing technology, has raised $3.6 million in funding to further develop its innovative emotional transfer technology, enhancing the realism and emotional depth of AI-generated voiceovers.

According to Yahoo Finance, Chinese artificial intelligence (AI) startup Zhipu announced on Tuesday the completion of its latest funding round, having previously received investments from internet giants such as Alibaba and Tencent last year. The statement reveals that Zhipu's commercial revenue has more than doubled this year. Zhipu is one of the few AI startups in China valued at billions of dollars, receiving substantial investment from local tech giants. In May of this year, Saudi Aramco's venture capital division participated in Zhipu's $400 million funding round, bringing its valuation to approximately $3 billion. AI development.

Recently, the Meta AI team introduced LongVU, a novel spatio-temporal adaptive compression mechanism aimed at enhancing the language understanding capabilities of long videos. Traditional multimodal large language models (MLLMs) face limitations in context length when processing long videos, and LongVU was created to address this challenge. LongVU operates primarily by filtering duplicate frames and employing inter-frame token compression techniques to efficiently utilize context length, allowing it to reduce video data while preserving visual details.

Recently, during the Tokyo Game Show, Falcom's president, Koji Kondo, mentioned in an interview with 4Gamer that the company might consider using artificial intelligence in the future to shorten game localization times. He specifically talked about his views on AI translation, believing that such technology could help quickly translate Japanese games into multiple languages. Image source note: The image is generated by AI, provided by the image licensing service Midjourney. His remarks came after demonstrating a software called ELLA, which aims to assist in game localization.

In today's world where the AI wave sweeps across the globe, language barriers are no longer a stumbling block to knowledge sharing. VideoLingo, a revolutionary video translation and localization dubbing tool, can deliver Netflix-level quality translated videos with a single click, heralding a new way of AI video translation. VideoLingo is a fully automated video localization tool that supports the generation of Netflix-quality subtitles! It offers various high-quality dubbing options! In the context of AI significantly enhancing content production efficiency, Vi

In July this year, Aishi Technology released the newly upgraded AI video generation product PixVerse V2, which is based on the Diffusion+Transformer (DiT) architecture, achieving multiple innovations in video generation technology. PixVerse V2 offers longer, more consistent, and more engaging video generation capabilities, supporting the generation of multiple video clips at once.

In the world of AI, making machines understand videos is much harder than understanding images. Videos are dynamic, with sound, movement, and a myriad of complex scenes. In the past, AI viewed videos as if they were reading ancient scrolls, often leaving them baffled. But the introduction of VideoPrism might change everything. It's a video encoder developed by Google's research team that can achieve state-of-the-art levels on a variety of video understanding tasks with a single model. Whether it

Intelligent Spectrum AI has announced the open-source upgrade of the CogVLM2-Video model, a significant advancement in the field of video understanding. The CogVLM2-Video model addresses the limitations of existing video understanding models in handling missing temporal information by introducing multi-frame video images and timestamps as input to the encoder. Utilizing an automated method for constructing time localization data, it has generated 30,000 temporal-related video question and answer

The Indian language model BharatGPT is set to launch in April, supporting 11 local Indian languages. This model is jointly developed by a company owned by Asia's richest man, Mukesh Ambani, in collaboration with eight Indian universities. BharatGPT aims to address the complex language system in India, covering fields such as healthcare and finance. Industry experts believe its release will mark a significant milestone in India's autonomous AI technology capabilities. Ambani has previously collaborated with companies like NVIDIA to co-develop India's language AI models.

JUNLALA's AI software SUPERLALA officially launches in the Chinese market, emphasizing simplicity and the needs of Chinese users. The latest version of the JUNLALA App is released for the first time in China, featuring localized functionalities and interfaces. SUPERLALA is user-centric, offering a variety of features to ensure a smooth user experience.

Byte's large model BuboGPT supports fine-grained multimodal joint understanding of text, images, and audio. By introducing a visual localization method, BuboGPT can accurately locate objects in images. Researchers have utilized a multimodal instruction tuning training scheme, achieving good results on multimodal tasks.