Recently, DiffusionModel has made significant progress in the field of image generation, bringing unprecedented development opportunities to image generation and video generation tasks. Despite the impressive results, the multi-step iterative denoising properties inherent in the inference process of diffusion models result in high computational costs. Recently, a series of diffusion model distillation algorithms have emerged to accelerate the inference process of diffusion models. These methods can be roughly divided into two categories: i) trajectory-preserving distillation; ii) trajectory reconstruction distillation. However, these two types of methods are limited by the limited effect ceiling or changes in the output domain. In order to solve these problems, the ByteDance technical team proposed a trajectory segmentation consistent method called Hyper-SD.

According to news on June 13, according to Byte's "Volcano Engine" public account, Xiaomi's artificial intelligence assistant "Xiao Ai" has reached a cooperation with Volcano Engine. The two parties will achieve a more intelligent AI interactive experience based on the beanbao large model. It is reported that the large-scale beanbao model created by ByteDance can efficiently process up to 120 billion text tokens and generate 30 million pieces of content every day. Xiaomi used the beanbao large model to improve the learning and reasoning capabilities of its own model and create a new "Xiao Ai Classmate", which not only more accurately grasps user needs, but also provides faster response speed and more comprehensive content services. For example, when a user asks about a complex scientific concept, &ldq

Seed-TTS is a large-scale speech generation model recently released by the ByteDance Doubao model team. , the speech it generates is almost **no different** from real people, and even pronunciation **defects** can be generated, especially in terms of learning to imitate human speech, with both **fidelity** and **fluency** **Outstanding performance. For example, if you provide a piece of speech to Seed-TTS, it can generate a new speech based on the text, and bring the sound characteristics of the original material. Original material (Prompt): Chinese voice generated by Seed-TTS: Suddenly, there was laughter around me. I looked at them, straightened my chest with high spirits, shook my slightly fleshy arms, and chuckled: "The flesh on my body is to cover up my overwhelming charm, otherwise

According to news on the afternoon of June 21, recently, some netizens said that after installing FeiLian (FeiLian is an office platform used internally by ByteDance and open to the public), they accidentally discovered that FeiLian was executing a screenshot process in the background. The user It reminds everyone not to do things you shouldn't do on computers with such software installed. Sina Technology asked ByteDance about this, and relevant sources said that Feilian provides data security protection functions for corporate customers. Enterprises can set the functional configuration of automated audit policies based on their own business needs and data sensitivity. The situation described in the forum post is that the enterprise administrator has enabled the DLP (data leakage prevention) function on the device that stores sensitive data. Felian will cycle through sensitive devices with DLP enabled at certain intervals.

According to news from this website on June 27, according to Tech Planet, after Tomato Novel, Byte has launched two free online literary apps, "Danhua Novel" and "Changdu Novel", focusing on free reading for everyone. The development companies of the two web product products are Hubei Furixing Network Technology Co., Ltd. and Hubei Jurun Network Technology Co., Ltd., both of which are 100% controlled by ByteDance. A search on this site found that, as of the time of publication, "Danhua Novel" and "Changdu Novel" ranked 4th and 9th respectively among the free apps in the book category on Apple's AppStore. In terms of products, although Danhua Novels and Changdu Novels are positioned as free online literature platforms, they are slightly different from Tomato Novels. Danhua Novels and Changdu Novels pay more attention to the experience of listening to books, and include short plays, comics, etc.

According to news from this site on May 1, David Tesler discovered that TikTok has begun to test an option on a small scale that allows users to directly recharge gold coins through tiktok.com external webpage. This option does not go through the AppStore in-app purchase system, so it is completely bypassed. 30% "Apple tax". "Similar to Douyin's diamonds (1+10 diamonds), TikTok also has its corresponding "gold coins" (in fact, TikTok also has diamonds). The current ratio is about 8 to 10 gold coins/yuan, or 66 to 69 US dollars." However, as long as you open the recharge interface through an external web page, the ratio will instantly increase to 94 coins/USD. The extra money is obviously the part taken by Google and Apple during in-app purchases, and users

According to news from this site on May 8, TikTok filed a lawsuit in the U.S. Federal Court on Tuesday, challenging the Biden administration’s divestiture bill and asking the U.S. court to prevent the implementation of the bill. TikTok filed a lawsuit on Tuesday, claiming that Congress "took unprecedented measures to clearly single out and ban TikTok" and called the move "unconstitutional." According to reports, Beckman said that if the separation bill officially takes effect, TikTok will fight back in court. Beckman said that the "sell or ban" law "clearly violates" the First Amendment rights of TikTok's 170 million American users and will have "devastating consequences" for the 7 million small businesses on the platform. He also said, "This is the beginning of a long process, not the end" and "continue to fight." Related

Whether it’s tongue twisters with extremely fast speech and complex pronunciation, exquisite classical Chinese, or casual chats full of impromptu and inspiration, the model can provide accurate and authentic translation results smoothly and naturally. In recent years, artificial intelligence (AI), especially AI represented by large language models (LLMs), has been developing at an alarming rate. These models have demonstrated outstanding capabilities in a variety of natural language processing tasks. However, despite breakthroughs in many fields, simultaneous interpretation (Simultaneous Interpretation, SI), which represents the top level of human language, is still an unfinished project.