Tag
9 articles
ByteDance's Seed Lab discovers that training large multimodal models with question-answering tasks improves performance on long documents compared to traditional transcription methods.
ByteDance's Intelligent Creation Lab has released Lance, an open-source unified multimodal model capable of image and video understanding, generation, and editing in a single framework using just 3 billion parameters.
ByteDance's Anew Labs, known for building TikTok's recommendation algorithm, is now using AI to design drugs for previously 'undruggable' diseases.
ByteDance introduces watermarking and IP guardrails in Seedance 2.0 to address concerns over deepfakes and misinformation, as it cautiously expands globally.
ByteDance's Dreamina Seedance 2.0 AI video model now available in CapCut with built-in protections against unauthorized content use.
ByteDance's AI ambitions are facing obstacles from compute constraints and copyright concerns, as its new Seedance 2.0 video model struggles with demand and legal scrutiny.
ByteDance introduces Astra, a dual-model architecture revolutionizing robot navigation in complex indoor environments. The system combines perception and planning models to enhance autonomous robot capabilities.
ByteDance's Seedance 2.0 shows improved AI video generation capabilities, but still falls short of true realism. Critics argue that while the tool produces more lifelike content than previous models, significant limitations remain.
ByteDance's AI research introduces a novel approach to stabilizing long chain-of-thought reasoning by mapping molecular bonds in AI reasoning processes, potentially revolutionizing how LLMs handle complex tasks.
