ByteDance

字节跳动旗下即梦AI 宣布，两大视频模型正式全量上线。官方称包含具有首帧一致性的 S2.0 Pro 和具有极高提示词遵循能力 P2.0 Pro 视频生成模型。 目前，打开即梦 App 或访问即梦官网，即可选择视频 S2.0 Pro 或视频 P2.0 Pro 生成模型。 S2.0 Pro 官方表示，S2.0 Pro 在 i2v（IT之家注：image to video、图生视频）场景下有非常好的首帧…

字节跳动豆包大模型团队昨日公布通用图像编辑模型 SeedEdit，用户只需输入简单的自然语言，便可对图像进行多样化编辑操作，包括修图、换装、美化、风格转化以及在指定区域添加或删除元素等。 官方表示，该模型已在豆包 PC 端及即梦网页端开启测试。 官方称 SeedEdit 能够广泛适应不同用户的多样化编辑需求，可控性强、编辑效果佳且无贴图感，主要体现在三个方面： 高精度指令理解：对中英文、专有名词“…

Recently, namely dream AI announced that from now on byte jumping self-developed video generation model Seaweed for platform users officially open. After logging in, under the "Video Generation" function, users can experience the video model by selecting "Video S2.0". Seaweed video generation model is part of the beanbag model family, with professional-grade light and shadow layout and color harmonization, the visual sense of the screen is very beautiful and realistic. Based on the DiT architecture, the Seaweed video generation model can also realize smooth and natural large-scale motion pictures. Tests show that the model only needs 60s to generate a high-quality AI with a duration of 5s...

Byte jumping's AI content platform, namely dream AI, announced on November 8, byte jumping's self-developed video generation model Seaweed is officially open to platform users. After logging in, under the function of "video generation", users can experience it by selecting "Video S2.0" for the video model. The attached link is as follows: https://jimeng.jianying.com/ai-tool/video/generate At the end of September, ByteDance officially announced its entry into the field of AI video, releasing two video generation models of the beanbag model family...

November 4 news, byte jumping quietly launched an AI model sharing community called "stove rice Lumi", the platform is said to integrate model sharing, Workflow building and LoRA training and other functions, users can upload their own development of AI models, and share them with others in the community, the platform also supports users to build AI workflow. The platform also supports users to build AI workflows. However, it is noted that the platform is still in the internal testing stage, and at present the mainstream search engines have not included the relevant website, but you can use the "artistrylab.net" web address to directly...

HybridFlow is the result of a joint research project between ByteDance's Beanbag Big Model team and the University of Hong Kong. Officially, HybridFlow (open source project name: veRL) is a flexible and efficient RL training framework for big models, compatible with a variety of training and inference frameworks, and supporting flexible model deployment and Multiple RL algorithm implementation. The framework uses a hybrid programming model that combines the flexibility of Single-Controller and the efficiency of Multi-Controller to better realize and execute multiple...

Byte Jump released a factual clarification announcement about interns destroying big model training". Byte Jumping big model training was attacked by interns, and the person involved has been dismissed The announcement said that recently the media said that Byte Jumping big model training was attacked by interns", and after the company's internal verification, there was indeed a serious disciplinary violation of the interns of the commercialization technology team, and the interns have been dismissed. Byte Jumping said that the relevant reports also exist in part exaggerated and inaccurate information, the special explanation is as follows: First, the intern involved maliciously interfered with the model training task of the commercialization technology team's research project, but it did not affect the commercialization of the official project and the online business, and also...

The official number of the byte jumping beanbag Xiaohongshu released a preview poster with the text "Byte jumping beanbag's smart hardware?" , and more information will be announced on October 10 at 11 am. Ola Friend official microblogging has also been online, released the same content as the official number of the beanbag small red book. It says, "Explore new ways to use AI". The first product of Byte's AI hardware team is a smart headset, which will be implanted in Doubao's big model and will be combined with Byte's AI app Doubao, which will be launched in August 2023, to create a conversation-based AI application....

Beanbag video generation model release site. Jiang Jialing photo People's Daily Shenzhen, September 24 (Jiang Jialing) September 24, byte jumping under the volcano engine in Shenzhen held AI innovation tour, the scene released beanbag video generation - PixelDance, beanbag video generation - Seaweed two big models for the enterprise market to open the invitation to test. According to reports, most of the previous video generation models can only complete simple instructions. The new beanbag video generation model allows the video to switch freely between large dynamics and transport mirrors, with zoom, surround, target following and other multi-camera language capabilities, and with professional-grade light and shadow layout and color...

Byte Jump Volcano Engine announced that the beanbag big model will be released on September 24th video generation model, and will bring more model family ability upgrade. It is understood that the Beanbag Big Model was officially released on May 15, 2024 at the Volcano Engine Original Power Conference. Beanbag Big Model offers multiple versions, including professional and lightweight versions, to suit the needs of different scenarios. The professional version supports 128K long text processing with powerful comprehension, generation and logic synthesis capabilities for scenarios such as Q&A, summarization, creation and classification. While the lightweight version provides lower token cost and latency, providing enterprises with flexible...

According to LatePost, Byte is exploring combining big models with hardware, both by developing its own AI hardware and collaborating with external hardware companies. The first product of Byte's AI hardware team is said to be a smart headset, which will be implanted with the Doubao big model and combined with Byte's dialog AI app Doubao App, which will go live in August 2023: after wearing the smart headset, users can use Doubao at any time through voice dialog, and the headset can be manipulated on the Doubao App. ByteDance has this month completed its acquisition of open headset brand Oladanc...

Hippo Aixue, the AI education app under ByteDance Dali Education, has recently been officially renamed "Doubao Aixue", becoming the first app named after Doubao. Hippo Aixue was previously managed by the Dali Education ZERO team, which mainly studies AI education in China. In the first half of this year, due to business adjustments, Dali Education ZERO has been merged into the Doubao team, and Hippo Aixue has also been merged into Doubao, and recently changed its name to Doubao Aixue, becoming a member of the Doubao product matrix. On May 15, 2024, ByteDance officially launched the Skylark Big Model at the Volcano Engine Power Conference...

On August 8, ByteDance's intelligent AI assistant Doubao launched a music generation function today. Users can enter the theme or lyrics written by themselves in Doubao's "Music Generation", set the music style, mood and timbre, and quickly generate a song of about 1 minute. Currently, Doubao's music generation function provides 11 music styles such as folk, hip-hop, R&B, etc., which also covers relatively niche styles such as jazz, reggae, and electronic music. Users can choose male or female voices. The generated sound performs well in articulation clarity and fluency. At the same time, in order to make AI sound...

On July 25, Volcano Engine held the 2024AI Innovation Tour in Chengdu. Volcano Engine announced that the average daily token usage of its Doubao model has exceeded 500 billion, and the average daily token usage of each corporate customer has increased 22 times since the model was released on May 15. Zhang Xin, vice president of Volcano Engine, said that Volcano Engine is moving towards a more intelligent, industrial and regional direction, and helping companies achieve business innovation through industry solutions, products and optimization services. At the meeting, Volcano Engine released the latest capabilities of the Doubao model, including visual images, speech synthesis, sound reproduction...

The Button Platform is a tool that can quickly transform AI ideas into enterprise-level AI applications. It provides a wealth of plug-ins, workflows, image streams, knowledge bases, and databases, allowing users to easily transform AI ideas into "expert bots" that can solve complex tasks. The features of the Button Platform include: Rich plug-ins: It integrates nearly 100 plug-ins of various types, such as information reading, travel, efficient office, image understanding, etc., which can be directly added to the Bot to expand its capabilities. Knowledge base: It provides a simple and easy-to-use knowledge base function to manage and store data, allowing Bot to use the knowledge base...

Recently, some users raised questions on social media, saying that WPS Office may have leaked its review content to ByteDance's AI product "Doubao". In response, ByteDance quickly issued an official statement, clearly stating that the rumors that Doubao AI used WPS user data for training were unfounded. The statement emphasized that the bibliographic information obtained by Doubao AI came from public channels, and reference information from relevant websites would be provided. ByteDance further clarified that Doubao AI had no cooperation with WPS in AI training, and had never used any undisclosed private data for training...

Titanium Media AGI reported that the ByteDance team will announce the progress of new artificial intelligence (AI) model technologies such as Wenshengtu and Sora-like new videos on a large scale for the first time on July 19 (this Friday), especially in the direction of long videos and high dynamics, which may directly target Sora effects. Sources said that ByteDance has set AI big models as the highest level of the group's P0. In addition, multiple internal teams such as Douyin and Jianying are also developing AI video model applications, which are expected to be announced in the near future. ByteDance will hold a conference in Singapore on the world...

Yesterday, ByteDance officially released the AI smart coding tool MarsCode. According to reports, MarsCode is an integrated development environment (IDE) with a built-in AI assistant and extensions that support more than 100 languages and mainstream IDEs. It has a powerful AI assistant that can provide code completion, interpretation, and debugging, helping developers code and innovate faster. The AI assistant can provide suggestions for single lines or entire functions during the coding process, and supports generating code snippets through comments. The editor can understand natural language to provide accurate code…

According to Sina Finance, there is market news today that ByteDance will cooperate with Broadcom, an American chip design company, to develop AI chips. In response, ByteDance said: The news is untrue. Earlier today, there were reports that ByteDance is working with Broadcom of the United States to develop AI processors to ensure that there are enough high-end chips. People familiar with the matter have said that this AI processor has a 5nm process and will be manufactured by TSMC. Although the design work is progressing smoothly, the "tapeout" that marks the end of the design phase and the beginning of manufacturing has not yet begun. It is reported that ByteDance and Broadcom have always been business partners, and Broadcom has been...

ByteDance has released a new generation of Depth Anything V2 deep model, which achieves significant performance improvements in the field of monocular depth estimation. Compared with the previous generation Depth Anything V1, the V2 version has finer details and stronger robustness, while also significantly improving efficiency, more than 10 times faster than the Stable Diffusion-based model. Key features: Finer details: The V2 model is optimized in details and provides finer depth predictions. High efficiency and accuracy: Compared with the SD-based...

ByteDance Volcano Engine officially announced that its one-stop large model service platform Volcano Ark will officially provide services to individual developers. Individual developers can experience and apply more than a dozen large models including the Doubao large model family and other platform functions. The Doubao full series of models on Volcano Ark will provide individual users with 500,000 tokens of free inference quota. In addition, the 4K and 32K versions of Doubao Universal Model Pro and Doubao Universal Model Lite also provide users with 10,000 RPM (requests per minute) and 800,000 TPM (transactions per minute…

According to Sina Technology, there have been recent media reports that ByteDance "secretly launched" an AI mobile phone research and development project two months ago. In response to the above information, a person related to ByteDance said: The information is untrue. In fact, it is exploring a large-scale model software solution based on mobile phones, which is provided to mobile phone manufacturers for reference. There is currently no plan to make and sell mobile phones by itself. The news originally came from a tweet released by "AR Circle" on June 10, which claimed that the core team of the project is mainly composed of two parts: one part comes from the Hammer mobile phone research and development team acquired by ByteDance in 2019, and the other part comes from 2021...

Recently, ByteDance's AI assistant Doubao officially launched PC client downloads, including Windows / MAC versions, and also launched a browser plug-in version. It is reported that the Doubao PC client version supports quick word translation, AI search, one-click desktop and other functions. The plug-in version provides one-click summary of web pages and videos, writing and text modification and other functions. In May this year, ByteDance launched its self-developed "Doubao Big Model" series. This big model family covers Doubao Universal Model Pro, Liti, Doubao Role Playing Model, Doubao Voice Synthesis...

Today I would like to introduce an AI application editing and development platform - Coze. Its biggest advantage is that you can use ChatGPT4 for free. What is Coze? Coze (button) is a new generation of AI chatbots and a one-stop application editing and development platform launched by ByteDance. It can be understood as the ByteDance version of GPTs. Whether or not users have programming skills, they can quickly create personalized chatbots, agents, AI applications and plug-ins through the platform, and embed them into Discord, Telegram, Facebook, WeChat public accounts...